[Chris Crawford] The Art Of Computer Game Design (BookFi org) pdf
The Art of Computer Game Design by Chris Cra wford P r e f a c e t o t h e E le c t r o n ic Ve r s io n
This text was originally composed by computer game designer
Chris Crawford in 1982. When searching for literature on the
nature of gaming and its relationship to narrative in 1997, Prof.
Sue Peabody learned of The Art of Computer Game Design,
which was then long out of print. Prof. Peabody requested Mr.
Crawford's permission to publish an electronic version of the
text on the World Wide Web so that it would be available to her
students and to others interested in game design. Washington
State University Vancouver generously made resources avail-
able to hire graphic artist Donna Loper to produce this elec-
tronic version. WSUV currently houses and maintains the site.
Correspondance regarding that site should be addressed to
Prof. Sue Peabody, Department of History, Washington StateUniversity Vancouver, peabody@ vancouver.wsu.edu.
If you are interested in more recent writings by Chris
Crawford, see the "Reflections" interview at the end of The Art
of Computer Game Design. Also, visit Chris Crawford's web-page, Erasmatazz.
This document was convert by Mario Croteau, from the Web
site of the Department of History of Washington State University at Vancouver.
Chris Crawford (the author) and Sue Peabody (of department of
History of Washington State University at Vancouver) gave me
a great support in my project: making that important document available to everyone. The Art of Computer Game Design by Chris Cra wfordTa b le o f C o n t e n t s
A C K N O W L E D G M E N T
am deeply indebted to Madeleine M. Gro ss fo r her painstaking and tho ro ugh criticisms o f this bo o k. In many cases she invested greater effo rts into her criticisms than I had put into my o riginal tho ughts. She stro ve to restrain my wild hyperbo le and place my arguments o n a
firmer fo undatio n o f rigo ro us lo gic. The lo gical co nsistency and reliability in this bo o k I o we to her; the speculative flights o f fancy must be laid at my do o rstep.
P R E FA C E
The central premise o f this bo o k is that co mputer games co nstitute a new and as yet po o rly devel- o ped art fo rm that ho lds great pro mise fo r bo th designers and players. This premise may seem laughable o r flippant. Ho w co uld anybo dy classify the likes o f SPACE
INVADERS and PAC MAN as art? Ho w can TEMPEST o r MISSILE CO MMAND co mpare with Beetho ven’s Fifth Sympho ny, Michelangelo ’s Pieta, o r Hemingway’s A Farewell To Arms? Co mputer games are to o trivial, to o frivo lo us to be called art. They are idle recreatio n at best. So says the skeptic.
But we canno t relegate co mputer games to the cesspit o f po p culture so lely o n the evidence o f the current cro p o f games. The industry is to o yo ung and the situatio n is to o dynamic fo r us to dis- miss co mputer games so easily. We must co nsider the po tential, no t the actuality. We must address the fundamental aspects o f co mputer games to achieve a co nclusio n that will withstand the ravages o f time and change. There are many definitio ns o f art, few o f which make much sense to the uninitiated. I will pres- ent my o wn pedestrian definitio n: art is so mething designed to evo ke emo tio n thro ugh fantasy. The artist presents his audience with a set o f senso ry experiences that stimulates co mmo nly shared fantasies, and so generates emo tio ns. Art is made po ssible o nly by the richness o f the fan- tasy wo rld we share. Art is nevertheless difficult, because there are so many practical pro blems asso ciated with stimulating fantasies deep inside ano ther perso n’s mind. A majo r pro blem is get- ting the attentio n o r participatio n o f the audience. Mo st art allo ws very little participatio n. Yo u sit quietly and listen to music that o ther peo ple created and perfo rm, o r yo u stro ll thro ugh a muse- um and stare at pictures o r statues o ther peo ple made. Yo u sit passively and read a no vel, o r a po em, o r a sho rt sto ry. With all o f these art fo rms, the ro le o f the audience is passive. The artist do es all the active wo rk, makes the biggest emo tio nal investment. The audience is expected to abso rb quietly the fruits o f the artist’s exertio ns. Active participatio n is severely curtailed. Witho ut participatio n, attentio n dwindles and impact crumbles away. This is in no wise a criticism o f art o r artists. The techno lo gies o f art preclude participatio n. If we had every klutz jump into the o rchestra pit, o r prance o n the o pera stage, o r slo p paint with so much in their wo rk and mo st peo ple will hear so little because they canno t participate in the art. Enter the co mputer. Co nceived lo ng ago , bo rn in war, reared as the servant o f business, this no w ado lescent techno lo gy has explo ded o ut o f the co mputer ro o m and invaded sho pping centers, pizza parlo rs, and ho mes. Po pular characterizatio ns o f the co mputer alternate between the o ld image o f the co mputer as o mniscient, co ld blo o ded, giant calculato r, and the new image o f the co mputer as purveyo r o f video thrills and 25 cent fixes. O riginally develo ped as a number crunch- er, the co mputer assumed a new perso nality when it was given graphics and so und capabilities. These capabilities gave the co mputer a po werful asset: it co uld no w co mmunicate with the human, no t just in the co ld and distant language o f digits, but in the emo tio nally immediate and co mpelling language o f images and so unds. With this capability came a new, previo usly undreamed o f po ssibility: the po ssibility o f using the co mputer as a medium fo r emo tio nal co m- municatio n art. The co mputer game has emerged as the prime vehicle fo r this medium. The co m- puter game is an art fo rm because it presents its audience with fantasy experiences that stimulate emo tio n. Unfo rtunately, the current generatio n o f micro co mputers canno t pro duce a senso ry experience as rich as that pro duced by, say, a sympho ny o rchestra o r a mo vie. This weakness is mo re than o ff- set by a fundamental advantage lacking in mo st o ther art fo rms: a game is intrinsically participa- to ry in nature. The artist has here a to o l that is mo re subtly indirect than traditio nal art. With o ther art fo rms, the artist directly creates the experience that the audience will enco unter. Since this experience is carefully planned and executed, the audience must so meho w be prevented fro m disturbing it; hence, no n participatio n. With a game, the artist creates no t the experience itself but the co nditio ns and rules under which the audience will create its o wn individualized experience. The demand o n the artist is greater, fo r s/ he must plan the experience indirectly, taking into acco unt the pro bable and po ssible actio ns and reactio ns o f the audience. The return is far greater, fo r participatio n increases attentio n and heightens the intensity o f the experience. When we pas- sively o bserve so meo ne else’s artistic presentatio n, we derive so me emo tio nal benefit, but when we actively participate in a game, we invest a po rtio n o f o ur o wn ego into the fantasy wo rld o f the game. This mo re sizable investment o f participatio n yields a co mmensurately greater return o f emo tio nal satisfactio n. Indeed, the ro le o f participatio n is so impo rtant that many peo ple derive greater satisfactio n fro m participating in an amateur artistic effo rt than fro m o bserving a pro fes- sio nal effo rt. Hence, games, being intrinsically participato ry, present the artist with a fantastic o ppo rtunity fo r reaching peo ple. Until no w, games in general and co mputer games in particular have no t been very impressive as art fo rms. The co mputer games especially are do wnright puerile. This is because the techno lo gy o f co mputer games has been in the hands o f techno lo gists, no t artists. These guys ( and they are almo st all male) can write beautiful o perating systems, languages, linking lo aders, and o ther tech- no lo gical wo nders, but artistic flair has hereto fo re been treated as subo rdinate to technical pro wess.ketplace. These machines are new; the public is unfamiliar with them and the manufacturers are hesitant to press the public to o hard to o fast. We therefo re o pt to build inhibited little games pathetically whispering so me trivial emo tio n. Truly intense emo tio ns o r situatio ns such as patho s, ecstasy, majesty, rapture, catharsis, o r tragedy intimidate use. We hide behind the defense that we are in the entertainment business, no t the art business, but that defense o nly betrays a pro fo und misunderstanding o f art. Art can be starchily elitist, but go o d art can also be a fo o t sto mping blast. Elitism arises fro m the intellectual co ntent o f art; impact springs fro m its emo - tio nal ho nesty.
Fo rtunately, times are changing. Already, we see a backlash develo ping against co mputer games. It expresses itself in many ways: in o rdinances against the placement o f arcade games in so me areas, in statements by educato rs deno uncing the games, and in mo re vigilant regulatio n o f chil- dren’s game activities by parents. This backlash is viewed by smaller minded members o f the industry with anxiety. Mo re visio nary thinkers watch the backlash with eager interest rather than defensiveness. The American peo ple are telling us so mething here, so mething very impo rtant. It is impo rant eno ugh to them that they are willing to co mpro mise their traditio nal reluctance to interfere with o ther peo ple’s business. While the arguments presented in public debates no rmal- ly fo cus o n fo rmal issues such as delinquency fro m scho o l, creatio n o f large gro ups o f ro wdy teenagers, and so fo rth, the co ncerns expressed privately reflect a distaste fo r the games, a vague suspicio n that the games are a waste o f time. Yo u can’t fo o l all o f the peo ple all o f the time; they are beginning to realize that the wo rld o f co mputer games is as yet a vast wasteland.
Co mputer games are much like candy, co mic bo o ks, and carto o ns. All fo ur activities pro vide intense o r exaggerated experiences. Whether they use sugar, exclamatio n po ints, o r animated explo sio ns, the go al is the same: to pro vide extreme experiences. Children appreciate these activ- ities because their no velty value is still stro ng. Adults, jaded by years o f experience with such things, prefer diversio ns with greater subtlety and depth. We thus have the pano ply o f culinary achievement, the vast array o f literature, and the universe o f mo vies as the adult co unterparts to candy, co mic bo o ks, and carto o ns. Yet, we have no adult co unterpart to co mputer games. This deficit is pregnant with po ssibilities, fo r it suggests a mo mento us upheaval in co mputer game design. This develo ping revo lutio n has no thing to do with the rapid techno lo gical develo pments o f the last few years. While techno lo gical impro vements will surely co ntinue, we are no lo nger ham- pered primarily by the limitatio ns o f the hardware. O ur primary pro blem is that we have little the- o ry o n which to base o ur effo rts. We do n’t really kno w what a game is, o r why peo ple play games, o r what makes a game great. Real art thro ugh co mputer games is achievable, but it will never be achieved so lo ng as we have no path to understanding. We need to establish o ur principles o f aes- thetics, a framewo rk fo r criticism, and a mo del fo r develo pment. New and better hardware will impro ve o ur games, but it will no t guarantee o ur artistic success any mo re than the develo pment o f o rchestras guaranteed the appearance o f Beetho ven. We are a lo ng way fro m a co mputer game these artists sto o d o n the sho ulders o f earlier artists who plunged into an unexplo red wo rld and mapped o ut its territo ries so that later artists co uld build o n their wo rk and achieve greater things. We co mputer game designers must put o ur sho ulders to gether so that o ur successo rs may stand o n to p o f them. This bo o k is my co ntributio n to that enterprise.C H A P T E R O N E W h a t is a G a m e ?
f we desire to understand games and game design, we must first clearly establish o ur funda- mental o rientatio n. We must define what we mean by the wo rd “game.” We must also deter- mine the fundamental characteristics o f all games. After discussing so me o f the o bstacles
inherent in this effo rt, I will briefly describe the salient classes o f games; then I will pro po se a set o f attributes that characterize all games. Games are a fundamental part o f human existence. The parlance o f games has insinuated itself into o ur language to refer to activities that are no t truly games. We play alo ng with activities we find distasteful. We play ball with tho se who require o ur co o peratio n. We play games when we are insincere. A willing participant is game fo r the enterprise. This bro ad penetratio n o f gaming co ncepts into the entire spectrum o f human experience presents us with two po tential barriers to understanding games. First, o ur liberal use o f gaming terms pro mo tes an exaggerated perceptio n o f o ur o wn under- standing o f games. We fail to render unto the subject the careful and critical analysis that we ten- der to mo re academic to pics, and we blithely igno re the co mplexities o f game design. Co mplete amateurs who se o nly relevant skill is pro gramming undertake to design games with no further preparatio n than their o wn experience as game players. Tho se who o verrate their o wn under- standing undercut their o wn po tential fo r learning. The seco nd o bstacle is ambiguity. We have applied the principles and co ncepts o f gaming so wide- ly that we have watered do wn their o riginal meanings. There is no lo nger a clear fo cus to the co n- cepts we seek to understand. Game designers have no well defined set o f co mmo n terms with which to co mmunicate with each o ther. Discussio ns o f game design frequently disintegrate into arguments o ver semantics. To cut thro ugh the tangled undergro wth that has gro wn up aro und gaming we shall need the bulldo zer and the scalpel. Let us begin this endeavo r by stepping back fo r a mo ment and taking o ur bearings. Let us take a brief to ur o f the universe o f games, glancing briefly at each o f the majo r regio ns. In the co urse o f this to ur I ho pe to refresh the reader’s memo ry o f games and make so me simple po ints befo re digging into the serio us analysis o f fundamental game characteristics. I perceive five majo r regio ns o f games: bo ard games, card games, athletic games, children’s games, and co mputer games.
We begin with the bo ard games. These games co nsist o f a playing surface divided into secto rs po p- ulated by a set o f mo vable pieces. In the mo st co mmo n arrangement the pieces are directly asso - ciated with the players, while the playing surface represents an enviro nment beyo nd the players’ direct co ntro l. Players maneuver their pieces acro ss the playing surface in an effo rt to capture mo dity. The player’s primary co ncern in these games is the analysis o f geo metrical relatio nships between the pieces.
A seco nd class o f games is the card games. These games utilize a set o f 52 symbo ls generated fro m two facto rs: rank ( 13 values) and suit ( 4 values) . The games revo lve aro und co mbinatio ns built fro m these two facto rs. Players may gain o r lo se po ssessio n o f symbo ls either by rando m pro cess- es o r by matching so me co mbinatio n allo wed by the rules o f the game. Each legal co mbinatio n is assigned a victo ry value fo r final assessment o f game results. Players must reco gnize bo th exist- ing and po tential co mbinatio ns and estimate pro babilities o f o btaining the cards necessary fo r co mpleting a co mbinatio n. This pro bability must be weighed against the victo ry value o f the co mbinatio n. Since the number o f co mbinatio ns is very large, precise co mputatio n o f the requi- site pro babilities exceeds the mental po wers o f almo st all players, rendering the game a primari- ly intuitive exercise. Thus, the player’s primary co ncern in these games is the analysis o f co mbi- natio ns.
Ano ther traditio nal game fo rm is the athletic game. These games emphasize physical mo re than mental pro wess. The rules o f the game rigo ro usly specify a precise set o f actio ns that the player is either allo wed to execute o r required to execute. Skillful use o f the bo dy is the player’s primary co ncern in these .games.
We must be careful to distinguish between athletic games and athletic co mpetitio ns. Fo r example, a race is a co mpetitio n, no t a game. The line o f demarcatio n between games and co mpetitio n illu- minates o ne o f the fundamental elements o f all games. I distinguish the two by the degree o f interactio n between players. Theo retically speaking, the runners in a race do no t interact with each o ther. Each is racing o nly against the clo ck; the presence o f o ther runners sho uld be immaterial. In truth, the runners do interact psycho lo gically, fo r the perfo rmance o f o ne runner can affect the perfo rmance o f the o ther runners. Furthermo re, in so me races a runner ( o r driver o r pilo t o r cap- tain) can physically interpo se himself in between the go al and ano ther racer, thereby gaining an advantage. I co nclude that the simplest co mpetitio ns, tho se in which each perso n strives to per- fo rm so me task o ptimally witho ut direct interactio n with the o ther co mpetito rs, do no t co nsti- tute games but co mpetitio ns. A co mpetitio n that do es allo w interactio n is a game.
Ano ther type o f gaming activity is the children’s game. Hide and Seek, Red Ro ver, Tag, and Kick the Can are co mmo n examples. Such games frequently take the fo rm o f gro up activities empha- sizing simple physical play. Altho ugh these games co ntain simple mental and physical co mpo nents,
Instead, the player’s primary co ncern in these games is the use o f so cial skills illuminating the fun- damental ro le o f the gro up in human life. A wide variety o f children’s activities are frequently referred to as games. When a child talks to a strip o f bark, maneuvers it, and pro vides so und effects, we are tempted to refer to such behavio r as game playing. Fo r the purpo ses o f this bo o k, I ,exclude such activities fro m the fo ld o f games. These impro visatio nal games are to o ill defined to pro vide us with any useful info rmatio n abo ut games.
The next area o f gaming we shall glance at is the current fad in gaming and the subject o f this bo o k, the co mputer game. These games are played o n five types o f co mputers: expensive dedicat- ed machines fo r the arcades ( “co in o p” machines) , inexpensive dedicated machines ( “hand helds”) , multi pro gram ho me games, machines such as the ATARI 2600 and the ATARI 5200, per- so nal co mputers, and large mainframe co mputers. The co mputer acts as o ppo nent and referee in mo st o f these games; in many o f them it also pro vides animated graphics. The mo st co mmo n fo rm o f co mputer game is the skill and actio n ( “S&A”) game emphasizing hand eye co o rdinatio n. These S&A games are frequently vio lent in nature. There are many o ther areas o f co mputer gam- ing: adventure games, fantasy ro le playing games, and war games. In o ur curso ry o verview, these o ther co mputer games are eclipsed by the sheer vo lume o f the skill and actio n games. This co ncludes o ur quick survey o f the mo st pro minent gro upings in the universe o f games. We shall return to the subject later, to create a taxo no my o f co mputer games, and later still to draw o n specific examples o f games to make po ints abo ut their nature. We must no w address the ques- tio n which mo tivated o ur initial reco nnaissance: what are the fundamental elements co mmo n to these games? I perceive fo ur co mmo n facto rs: representatio n, interactio n, co nflict, and safety.
First, a game is a clo sed fo rmal system that subjectively represents a subset o f reality. Let us exam- ine each term o f this statement carefully. By 'clo sed' I mean that the game is co mplete and self sufficient as a structure. The mo del wo rld created by the game is internally co mplete; no reference need be made to agents o utside o f the game. So me badly designed games fail to meet this require- ment. Such games pro duce disputes o ver the rules, fo r they allo w situatio ns to develo p that the rules do no t address. The players must then extend the rules to co ver the situatio n in which they find themselves. This situatio n always pro duces arguments. A pro perly designed game precludes this po ssibility; it is clo sed because the rules co ver all co ntingencies enco untered in the game.
By fo rmal I mean o nly that the game has explicit rules. There are info rmal games in which the rules are lo o sely stated o r deliberately vague. Such games are far remo ved fro m the mainstream o f game play.
The term 'system' is o ften misused, but in this case its applicatio n is quite appro priate. A game’s co llectio n o f parts which interact with each o ther, o ften in co mplex ways. It is a system.
Representatio n is a co in with two faces: an o bjective face and a subjective face. The two faces are no t mutually exclusive, fo r the subjective reality springs fro m and feeds o n o bjective reality. In a game, these two faces are intertwined, with emphasis o n the subjective face. Fo r example, when a player blasts hundreds o f alien invaders, no bo dy believes that his recreatio n directly mirro rs the o bjective wo rld. Ho wever, the game may be a very real metapho r fo r the player’s perceptio n o f his wo rld. I do no t wish to sully my arguments with po p psycho lo gical analyses o f players giving vent to deep seated aggressio ns at the arcades. Clearly, tho ugh, so mething mo re than a simple blast- ing o f alien mo nsters is go ing o n in the mind o f the player. We need no t co ncern o urselves with its exact nature; fo r the mo ment it is entirely adequate to realize that the player do es perceive the game to represent so mething fro m his private fantasy wo rld. Thus, a game represents so mething fro m subjective reality, no t o bjective. Games are o bjectively unreal in that they do no t physically re create the situatio ns they represent, yet they are subjectively real to the player. The agent that transfo rms an o bjectively unreal situatio n into a subjectively real o ne is human fantasy. Fantasy thus plays a vital ro le in any game situatio n. A game creates a fantasy representatio n, no t a scien- tific mo del.
Games versus Simulations
The distinctio n between o bjective representatio n and subjective representatio n is made clear by a co nsideratio n o f the differences between simulatio ns and games. A simulatio n is a serio us attempt to accurately represent a real pheno meno n in ano ther, mo re malleable fo rm. A game is an artistically simplified representatio n o f a pheno meno n. The simulatio ns designer simplifies reluctantly and o nly as a co ncessio n to material and intellectual limitatio ns. The game designer simplifies deliberately in o rder to fo cus the player’s attentio n o n tho se facto rs the designer judges to be impo rtant. The fundamental difference between the two lies in their purpo ses. A simula- tio n is created fo r co mputatio nal o r evaluative purpo ses; a game is created fo r educatio nal o r entertainment purpo ses.( There is a middle gro und where training simulatio ns blend into edu- catio nal games.) Accuracy is the sine qua no n o f simulatio ns; clarity the sine qua no n o f games.
A game is no t merely a small simulatio n lacking the degree o f detail that a simulatio n po ssesses; a game deliberately suppresses detail to accentuate the bro ader message that the designer wishes to present. Where a simulatio n is detailed a game is stylized. Co nsider, fo r example, the differences between a flight simulato r pro gram fo r a perso nal co m- puter and the co in o p game RED BARO N”. Bo th pro grams co ncern flying an airplane; bo th o per- ate o n micro co mputer systems. The flight simulato r demo nstrates many o f the technical aspects o f flying: stalls, ro lls, and spins, fo r example RED BARO N has no ne o f these. Indeed, the aircraft that the player files in RED BARO N is quite unrealistic. It canno t be stalled, ro lled, spun, o r dived into the gro und. When the stick is released it auto matically rights itself. It is inco rrect to co nclude fro m these o bservatio ns that RED BARO N is inferio r to the flight simulato r. RED BARO N is no t a game abo ut realistic flying; it is a game abo ut flying and sho o ting and avo iding being sho t. The inclusio n o f technical details o f flying wo uld distract mo st players fro m the o ther aspects o f the game. The designers o f RED BARO N quite co rrectly stripped o ut technical details o f flight to fo cus the player’s attentio n o n the co mbat aspects o f the game. The absence o f these technical details fro m RED BARO N is no t a liability but an asset, fo r it pro vides fo cus to the game. Their absence fro m a flight simulato r wo uld be a liability.
Subset of Reality The last term I use is “subset o f reality.” O ne aspect o f this term ( “subset”) is easily justified.
Clearly, no game co uld include all o f reality witho ut being reality itself; thus, a game must be at mo st a subset o f reality. The cho ice o f matter in the subset is the means o f pro viding fo cus to the game. A game that represents to o large a subset o f reality defies the player’s co mprehensio n and beco mes almo st indistinguishable fro m life itself, ro bbing the game o f o ne o f its mo st appealing facto rs, its fo cus.
Summary of Representation
A game creates a subjective and deliberately simplified representatio n o f emo tio nal reality. A game is no t an o bjectively accurate representatio n o f reality; o bjective accuracy is o nly necessary to the extent required to suppo rt the player’s fantasy. The player’s fantasy is the key agent in mak- ing the game psycho lo gically real.
So me media fo r representing reality are static. A painting o r sculpture depicts a snapsho t o f real- ity fro zen in time. So me media are dynamic; they sho w change with time. Mo vies, music, and dance are dynamic in this way. They are able to represent the changing aspect o f reality mo re rich- ly. But the mo st fascinating thing abo ut reality is no t that it is, o r even that it changes, but ho w it changes, the intricate webwo rk o f cause and effect by which all things are tied to gether. The o nly to let them generate causes and o bserve effects. Thus, the highest and mo st co mplete fo rm o f rep- resentatio n is interactive representatio n. Games pro vide this interactive element, and it is a cru- cial facto r in their appeal.
Games versus Puzzles
O ne way to understand the nature o f the interactive element o f games is to co ntrast games with puzzles and o ther no n interactive challenges. Co mpare playing a cube puzzle with playing a game o f tic tac to e. Co mpare the spo rt o f high jumping with the game o f basketball. In each co mpari- so n the two activities pro vide similar challenges to the player. The key difference that makes o ne activity a game and the o ther activity no t a game is the interactive element. A cube puzzle do es no t actively respo nd to the human’s mo ves; a high jump po le do es no t react to the jumper’s effo rts. In bo th tic tac to e and basketball the o ppo sing players ackno wledge and respo nd to the player’s actio ns.
The difference between games and puzzles has little to do with the mechanics o f the situatio n; we can easily turn many puzzles and athletic challenges into games and vice versa. Fo r example, chess, a game, has spawned a who le class o f puzzles, the end game pro blems. Games can include puzzles as subsets, and many do . Mo st o f the time the puzzles are a mino r co mpo nent o f the o ver- all game, fo r a game that puts mo st o f its challenge value o n included puzzles will rapidly lo se its challenge o nce the puzzles have been so lved.
Games versus Stories
Ano ther way to illustrate the ro le o f interactio n is to co mpare games with sto ries. A sto ry is a co l- lectio n o f facts in time sequenced o rder that suggest a cause and effect relatio nship. Frequently, the facts presented are deliberately fictitio us, because the facts o f a sto ry are intrinsically unim- po rtant. Indeed, the entire co ncept o f fictio n ( “an untruth that is no t a lie”) o nly makes sense when o ne realizes that the facts presented in the fictio n are themselves unimpo rtant. The cause and effect relatio nships suggested by the sequence o f facts are the impo rtant part o f the sto ry. Fo r example, we care no t whether Luke Skywalker and the Death Star really existed. We saw that Luke Skywalker was go o d and pure, and that the Death Star was evil, and that Luke Skywalker destro yed the Death Star. The cause and effect relatio nship suggested by the sto ry was that go o d o verco mes evil. Thus, a sto ry is a vehicle fo r representing reality, no t thro ugh its facts per se, but thro ugh the cause and effect relatio nships suggested by the sequence o f facts.
Games also attempt to represent reality. The difference between the two is that a sto ry presents the facts in an immutable sequence, while a game presents a branching tree o f sequences and allo ws the player to create his o wn sto ry by making cho ices at each branch po int. The audience o f a sto ry to explo re alternatives, co ntrapo sitives, and inversio ns. The game player is free to explo re the causal relatio nship fro m many different angles. Indeed, the player expects to play the game many times, trying different strategies each time. A sto ry is meant to be experienced o nce; its representatio nal value decreases with subsequent retellings because it presents no new info rmatio n. A game’s representatio nal value increases with each playing until the player has explo red a representative subset o f all o f the branches in the game net. This do es no t mean that games are better than sto ries. Altho ugh sto ries trace o nly a single sequence o f causal develo pment, they do so with greater intricacy and detail than games. Detail is crucial to the creative success o f a sto ry, fo r it pro vides the texture, the feel o f reality that makes a sto ry co mpelling. The sto ry writer unleashes a mighty swirling to rrent o f facts that sweeps the audience to its predestined co nclusio n. The game designer creates a co mplex netwo rk o f paths cunningly crafted to sho w the player all po ssible facets o f a single truth. In this respect, a sto ry is like a statuette where a game is like a jewel. The statuette’s value arises fro m the fineness o f detail and intricacy o f co nstructio n. A jewel, by co ntrast, has no detail; its faces must be abso lutely smo o th. The jewel’s value arises fro m its ability to refract light into many different angles. A stat- uette is meant to be statio nary; a jewel is meant to be mo ved. So to o , is a sto ry static where a game is dynamic. Sto ries enjo y a particular advantage o ver the current generatio n o f co mputer games: the element o f surprise. A go o d sto ry bo asts an array o f interesting plo t twists. The sto ryteller leads us into a set o f expectatio ns and then cleverly inserts a new facto r that creates a disjunctio n, a new and dra- matically different situatio n. This pro cess can be repeated many times during the co urse o f the sto ry. Amo ng co mputer games, o nly adventures pro vide this element o f surprise. Unfo rtunately, the surprise can o nly be created by limiting the player’s freedo m o f actio n so as to guarantee that the player will enco unter the surprise under the pro per circumstances. After a while, all adventures begin to smell like primro se paths. The really exciting po ssibility o ffered by co mputer games is the pro spect o f fo rmulating a plo t twist in respo nse to the player’s actio ns, instead o f merely drag- ging him do wn a pre-o rdained primro se path. Ho wever, the ability to fo rmulate surprise requires an ability to analyze the player’s actio ns, deduce his expectatio ns, and generate a believable plo t twist that co nfutes his expectatio ns witho ut frustrating him. Artificial intelligence that advanced has yet to be created.
Games versus Toys
Games lie between sto ries and to ys o n a scale o f manipulability. Sto ries do no t permit the audi- ence any o ppo rtunity to co ntro l the sequence o f facts presented. Games allo w the player to manipulate so me o f the facts o f the fantasy, but the rules go verning the fantasy remain fixed. To ys ryteller has direct creative co ntro l o ver his audience’s experience; the game designer has indirect co ntro l; the to ymaker has almo st no ne.
Significance of Interaction
Interactio n is impo rtant fo r several reaso ns. First, it injects a so cial o r interperso nal element into the event. It transfo rms the challenge o f the game fro m a technical o ne to an interperso nal o ne. So lving a cube puzzle is a strictly technical o peratio n; playing chess is an interperso nal o peratio n. In the fo rmer, o ne plays against the lo gic o f the situatio n; in the latter, o ne uses the lo gic o f the situatio n to play against the o ppo nent.
Seco nd, interactio n transfo rms the nature o f the challenge fro m a passive challenge to an active challenge. A puzzle will always present the player with exactly the same challenge. But a game o ppo nent reacts to player’s actio ns, and presents different challenges in each game. This difference has majo r emo tio nal significance. The perso n so lving the puzzle must so meho w divine, guess, deduce, master, o r disco ver the key trick built into the puzzle by the designer. Emo tio nally, the puzzle player is wo rking against the puzzle o r its designer to unmask its secret. O nce the secret is kno wn, the puzzle is no lo nger interesting. The game-player, by co ntrast, faces different challenges each time she plays the game. Where a puzzle is dead a game is alive; the player must create her so lutio n to the game in a manner best suited to her o wn perso nality and that o f her o ppo nent. The key distinctio n between a game and a puzzle is the difference between creating yo ur o wn so lutio n and disco vering the designer’s so lutio n. A game ackno wledges the player’s existence and reacts to the player’s perso nality; a puzzle lies do wn like a dead fish. Co mputer games seldo m pro vide a human o ppo nent, and so they lack the so cial element that o ther games o ffer. They can, ho wever, present an illuso ry perso nality against which the player must wo rk. This is o ne o f the mo st exciting and least develo ped po tentials o f the co mputer as a game techno lo gy. And regardless o f the co mputer’s success o r failure in synthesizing a so cial ele- ment, the co mputer can readily make the game a highly interactive experience fo r the player. It can react to the player’s mo ves with speed and tho ro ughness.
N ature of Interaction
Interactiveness is no t a binary quantity; it is a co ntinuo us quantity with a range o f values. Puzzles have little o r no interactiveness, while games have mo re interactiveness. This suggests that inter- activeness is an index o f “gaminess”. So me games, such as blackjack, tag, o r PO NG pro vide very little interactio n between the players. Altho ugh the players may wish to interact, the games pro - vide very limited mo des o f interactio n ( binary decisio n to stand o r hit, running, and twisting pad- dle) . The games do no t allo w players to invest much o f themselves into the play, o r to react in a rich way to their o ppo nents. O ther games, such as bridge, fo o tball, and LEGIO NNAIRE ( trade- mark o f Avalo n Hill Game Co .) allo w a far richer interactio n between players. Players can grap- dull, while the seco nd gro up o f games is generally regarded as mo re interesting. What is impo r- tant abo ut the mo des o f interactio n is no t their mechanical quality but their emo tio nal signifi- cance. PO NG is insipid because I can’t express much o f my perso nality thro ugh the medium o f a bo uncing ball. Bridge is better because it includes within its interactio n elements o f teamwo rk, deceptio n, and co o peratio n. I can better imprint my perso nality traits o nto a game o f bridge. Thus, degree o f interactio n pro vides a useful index o f “gaminess”.
A third element appearing in all games is co nflict. Co nflict arises naturally fro m the interactio n in a game. The player is actively pursuing so me go al. O bstacles prevent him fro m easily achieving this go al. If the o bstacles are passive o r static, the challenge is a puzzle o r athletic challenge. If they are active o r dynamic, if they purpo sefully respo nd to the player, the challenge is a game. Ho wever, active, respo nsive, purpo seful o bstacles require an intelligent agent. If that intelligent agent actively blo cks the player’s attempts to reach his go als, co nflict between the player and the agent is inevitable. Thus, co nflict is fundamental to all games.
Games without conflict?
So me peo ple shrink’ fro m this aspect o f games. A number o f attempts have been made to design “nice” games cleansed o f co nflict. Such games emphasize co o perative effo rts rather than co nflict. They have no t been successful co mmercially; this suggests that few peo ple enjo y them. Mo re impo rtantly, these games are failures because they are no t games in the first place. Co nflict can o nly be avo ided by eliminating the active respo nse to the player’s actio ns. Witho ut active respo nse, there can be no interactio n. Thus, expunging co nflict fro m a game inevitably destro ys the game.
While it is impo ssible to eliminate co nflict fro m a game witho ut destro ying the game, it is po ssi- ble to include co o perative elements by shifting the co nflict. Members o f a team can co o perate with each o ther in the team’s co nflict with ano ther agent. This o ther agent co uld be ano ther team, an individual human, o r a co mputer simulated player. In all cases, the o ppo nent must be per- ceivable as endo wed with a perso na. Witho ut at least the illusio n o f purpo seful reactio n to the player’s actio ns, the game co llapses.
This “blo o d and guts” view o f co nflict in games is reinfo rced by the so cial co ntext in which they are o ften played. O ur real wo rld co nflicts are always indirect, diffused o ver time, and tightly reg- ulated. Mo reo ver, they all to o frequently lack reso lutio n, fo r seldo m do es o ne achieve an o utright victo ry in the co nflicts o f daily life. Lo cal successes, yes, but the struggle co ntinues witho ut clear reso lutio n. Because games are subjective representatio ns o f the real wo rld, they fo cus o ur atten- tio n o n a particular aspect o f the wo rld by accentuating that aspect. Co nflict in games thus tends is no t essential o r fundamental to games. It is co mmo n in games because it is the mo st o bvio us and natural expressio n fo r co nflict.
Summary of Conflict
Co nflict is an intrinsic element o f all games. It can be direct o r indirect, vio lent o r no nvio lent, but it is always present in every game.
Co nflict implies danger; danger means risk o f harm; harm is undesirable. Therefo re, a game is an artifice fo r pro viding the psycho lo gical experiences o f co nflict and danger while excluding their physical realizatio ns. In sho rt, a game is a safe way to experience reality. Mo re accurately, the results o f a game are always less harsh than the situatio ns the game mo dels. A player can blast the mo nsters all day lo ng and risk o nly her quarter. She can amass huge financial empires and lo se them in an ho ur witho ut risking her piggy bank. She can lead great armies into desperate battles o n which hang the fate o f natio ns, all witho ut shedding a dro p o f blo o d. In a wo rld o f relentless cause and effect, o f tragic linkages and inevitable co nsequences, the disasso ciatio n o f actio ns fro m co nsequences is a co mpelling feature o f games. This is no t to imply that games are devo id o f co nsequences. The penalties fo r lo sing a game can so metimes be a significant deterrent to game play. Lo sing to ano ther perso n always entails so me lo ss o f dignity. This may be an attractio n o f co mputer games there is less shame in lo sing to a co mputer. The lo ser can keep co ming back fo r mo re defeats witho ut lo sing face. Mo reo ver, true victo ry the to tal destructio n o f the co mputer’s fo rces, is ackno wledged to be impo ssible in mo st such games; this further lessens the shame o f defeat.
A seco nd penalty fo r lo sing is the less o f any reward that might have been gained by winning. In almo st all games the reward penalty structure is po sitive. That is, the lo ser is no t punished fo r lo s- ing, the winner is rewarded fo r winning. The lo ser’s o nly lo ss is any investment that he made to enter the game, such as a bet o r entry fee. This investment is usually very small, and may rightly be regarded as a recreatio nal fee fo r the services asso ciated with the administratio n o f the game rather than a penalty fo r all po tential lo sers.
Gambling presents us with so me difficult pro blems related to the issue o f the safety o f games. Gamblers risk mo ney o r go o ds o n the o utco me o f a rando m o r near rando m pro cess. Lo sers fo r- feit their bets and winners reap a large reward. Hence, gambling presents a real financial risk to the player. Ho wever, two extenuating circumstances intervene: first, the recreatio nal gambler risks very little mo ney; seco nd, so me gamblers deny to themselves the laws o f chance. They indulge in the fantasy o f co ntro l. The pro per into natio n in the shake o f the dice, the co rrect twist o n the han- dle o f the slo t machine these things make the difference, o r so they tell themselves. Thus, recre- atio nal gambling, while so mewhat deviant fro m the mainline o f game playing, pro bably deserves expended mo re fo r anticipated financial gain than fo r recreatio n, lies o n the far side o f the gray zo ne. A special fo rm o f gambling, deserving special co nsideratio n here, is po ker. Po ker is a game o f bluffing; the key to success in the game lies in co nvincing yo ur o ppo nent that yo u have better o r wo rse cards than yo u really have. Because mo ney is at stake, the player experiences stresses that strain his ability to deceive his o ppo nents. Thus, the risk o f gambling, a mere o utco me o f o ther games, is an intrinsic part o f the structure o f po ker. This unique aspect o f po ker merits special co n- sideratio n. I wo uld no t hesitate to classify po ker as a game.
Summary of Safety
Games pro vide safe ways to experience reality. Special cases abo und, but the central principle remains: games are safe. In this chapter I have presented a set o f characteristics that defines what I mean by the wo rd “game”. Fo r the mo st part, I have emphasized the characteristics intrinsic to the games themselves rather than the mo tivatio ns o f the players. Such separatio n o f game fro m player is artificial and misleading, fo r neither exists witho ut the o ther. In the next chapter, I turn to lo o k at the players o f games and their mo tivatio ns. C h a p t e r Tw o W h y D o P e o p le P la y G a m e s ?
ame-playing requires two co mpo nents: a game and a player. The game designer wo rks to pro duce a game, and so her immediate preo ccupatio n is with the game itself. Yet, her final go al is to educate, entertain, o r edify the game-player; hence, the human player is
the pro per primary co ncern o f the game designer. Why do peo ple play games? What mo tivates them? What makes games fun? The answers to these questio ns are crucial to go o d game design. O ne way to address the questio n o f the purpo se o f games is to inquire into their histo ry. Games no w are to o varied, to o intricate, to o invo lved, to indicate a single clear functio n. Perhaps their fundamental nature wo uld be mo re evident in their earliest incarnatio ns. Ho w far back must we go ? To MO NO PO LY, created during the Depressio n? No , card games were played lo ng befo re that. Indeed, the disco verers o f King Tutankhamen’s to mb fo und amo ng the wealth there a wo o den surface with regular divisio ns that appears to be so me so rt o f bo ardgame. But even archaeo lo gy do es no t take us far eno ugh back. If we wish to get back to the beginnings o f games, we must go beyo nd the realm o f the archaeo lo gist and into the realm o f the paleo nto lo gist. We must reach no t tho usands but millio ns o f years into the past to find the earliest games, fo r games predate no t just histo ry but all o f mankind. They are no t a human inventio n. Fo rtunately, direct reco urse to paleo nto lo gy is unnecessary. A trip to the zo o will suffice. There we find two lio n cubs wrestling near their mo ther. They gro wl and claw at each o ther. They bite and kick. O ne cub wanders o ff and no tices a butterfly. It cro uches in the grass, creeps ever so slo wly to ward its insect prey, then raises its haunches, wiggles them, and po unces. We laugh at the co m- edy; we say that the cubs are playing a game, that they are having fun, and that they are such fun- lo ving, carefree creatures. We are right o n the first co unt: these cubs do indeed appear to be playing a kind o f game. We can certainly see in their behavio r all fo ur o f the fundamental game attributes described in Chapter 1: representatio n, interactio n, co nflict, and safety. We may be right o n the seco nd co unt; who kno ws if lio ns can have fun? But we are dead wro ng o n the last co unt. These cubs are no t carefree. They do no t indulge in games to while away the years o f their cubho o d. These games are deadly seri- o us business. They are studying the skills o f hunting, the skills o f survival. They are learning ho w to appro ach their prey witho ut being seen, ho w to po unce, and ho w to grapple with and dispatch prey witho ut being injured. They are learning by do ing, but in a safe way. Better to make mistakes with butterfly and sibling than with the ho rns o f the wildebeest. Games are thus the mo st ancient and time-ho no red vehicle fo r educatio n. They are the o riginal educatio nal techno lo gy, the natural o ne, having received the seal o f appro val o f natural selectio n. We do n’t see mo ther lio ns lecturing cubs at the chalkbo ard; we do n’t see senio r lio ns writing their memo irs fo r po sterity. In light o f this, the questio n, "Can games have educatio nal value?" beco mes absurd. It is no t games but scho o ls that are the newfangled no tio n, the untested fad, the vio lato r o f traditio n. Game-playing is a vital educatio nal functio n fo r any creature capable o f learning. o nly in mammals and birds. The phylo genetically earlier o rders ( fish, insects, amphibians, and reptiles) have no t been sho wn to engage in game-playing. ( See Animal Play Behavio r, by Ro bert Fagen, O xfo rd University Press.) Game play seems to be asso ciated with that quality which we have clumsily attempted to measure with brain size, intelligence, and ability to learn. This co rre- spo ndence canno t be attributed to accident; clearly game play is an impo rtant co mpo nent in the develo pment o f many creatures.
We co mmo nly asso ciate the playing o f games with children. Indeed, "play" as an activity is co n- sidered to be the almo st exclusive preserve o f children, and the term is applied to adults either disparagingly o r jo cularly. Children are expected to play games because we reco gnize ( perhaps unco nscio usly) the fundamental utility o f games as an educatio nal to o l. As children gro w up, cul- tural pressures change and they are enco uraged to devo te less time to the playing o f games so that they can devo te themselves to mo re serio us activities.
I claim that the fundamental mo tivatio n fo r all game-playing is to learn. This is the o riginal mo ti- vatio n fo r game-playing, and surely retains much o f its impo rtance. This claim do es no t co nflict with my o ther primary assertio n that co mputer games co nstitute a new art fo rm. Co nsider, fo r example, humans and fo o d. The fundamental mo tivatio n to eat fo o d is the base desire fo r no ur- ishment, yet this has no t prevented us fro m embellishing this fundamental activity with all man- ner o f elabo rate and no n-no urishing custo ms, rituals, seaso nings, and garnishes. I do no t mean to imply that fo o d is an art fo rm; o nly that we humans can take an activity far beyo nd its prime cause witho ut denying that prime cause.
I must qualify my claim that the fundamental mo tivatio n fo r all game-play is to learn. First, the educatio nal mo tivatio n may no t be co nscio us. Indeed, it may well take the fo rm o f a vague predilectio n to play games. The fact that this mo tivatio n may be unco nscio us do es no t lessen its impo rt; indeed, the fact wo uld lend credence to the assertio n that learning is a truly fundamen- tal mo tivatio n. Seco nd, there are many o ther mo tivatio ns to play games that have little to do with learning, and in so me cases these seco ndary mo tivatio ns may assume greater lo cal impo rtance than the ances- tral mo tivatio n to learn. These o ther mo tivatio ns include: fantasy/ explo ratio n, no se-thumbing, pro ving o neself, so cial lubricatio n, exercise, and need fo r ackno wledgment. I shall examine each in turn.
A very impo rtant mo tivatio n to play games is fantasy fulfillment. Like a mo vie, a bo o k, o r music, a game can transpo rt the player away fro m the tawdry wo rld that o ppresses him and create a fan- tasy wo rld in which he can fo rget his pro blems. Games are po tentially superio r to the traditio nal means o f escape ( movies, bo o ks, music) because they are participato ry. Instead o f merely watching the player drives the game, co ntro ls it in a way that is quite impo ssible with the passive fantasies. This need to escape, to fantasize is certainly an impo rtant mo tivatio n.
Fantasy fulfillment frequently takes the fo rm o f symbo lic explo ratio n. There’s a big wo rld o ut there, full o f exciting things, peo ple, and places, yet mo st o f us are co nfined to a wo rld ,o f asphalt, plastic, and paper. Many art fo rms attempt to transpo rt the audience into a different wo rld, to present experiences o r feelings no t o ften kno wn in the everyday wo rld.
Co nsider, fo r example, the success o f Disneyland. This place is undo ubtedly the mo st successful o f its genre. Such parks are o ften called "amusement parks" o r "theme parks." These terms are misleading, fo r the success o f Disneyland canno t be attributed so lely to its amusements and diversio ns. These elements are technically excellent, but o ther amusement parks spo rt technical- ly excellent rides. The success o f Disneyland can be summed up in o ne wo rd: fantasy. Disneyland creates and suppo rts an aura o f fantasy, a co ntext o f make-believe that permeates all o f the activ- ities within the park. Within mo ments o f entering the park, the visito r feels that s/ he is in a dif- ferent wo rld. Fanatic attentio n to detail in signpo sts, walls, windo ws, even railings has created an enviro nment that enco urages suspensio n o f disbelief. Fantasy is an impo rtant co mpo nent o f human play. It is critical to o ur recreatio n, o ur art and o ur games.
A co mmo n functio n o f games is to pro vide a means o f o verco ming so cial restrictio ns, at least in fantasy. Many games place the player in a ro le that wo uld no t be so cially acceptable in real life, such as a pirate o r a thief. An excellent ( albeit extreme) example o f this is the game CRUSH, CRUMBLE, AND CHO MP by Auto mated Simulatio ns. In this game the player is cast as a 1950’s- vintage mo nster go ing o n a rampage thro ugh his favo rite city. He sto mps o n po lice cars, crushes buildings, swats helico pters, and creates general mayhem. The bo x art sho ws a mo nster abo ut to attack an IRS building as terrified citizens flee. This represents an extreme case o f anti-so cial behavio r made acceptable by the safety o f the game. So metimes the player’s ro le is itself so cially acceptable, but the actio ns taken are disco uraged in real life. MO NO PO LY enco urages players to engage in what the Federal Trade Co mmissio n deli- cately calls "predato ry trade practices." Wargames enco urage players to start and win wars. So me games address sexual matters, allo wing players to indulge in make-believe behavio r that they co uld never exhibit in the real wo rld. The mo st telling example o f this no se-thumbing pheno meno n lies in the arcade games. These games emphasize vio lence, and lo ts o f it. The theme is almo st universal in arcades: destro y so me- bo dy. The co up de grace is no t delivered discreetly o r elegantly. O n the co ntrary, the victim is dis- patched with the m o st co lo rful anim ated explo sio n po ssible. Like a Sam Peckinpah m o vie, tasteful emo tio ns, we delicately mask them in less o ffensive garb. We never, never o bliterate human beings; instead, we vapo rize ugly space mo nsters. The mo nsters have perpetrated so me o dio us interstellar crime, so the player is cast as the defender, the pro tecto r, o r the avenger. The case is o ften presented that the game represents a time o f extreme crisis ( "THE FATE O F HUMAN-
ITY IS AT STAKE! ! ! ") . This heightens the player’s sense o f urgency; it also co nveniently justifies the use o f extreme vio lence, thereby allo wing the player to have vio lence witho ut guilt. The player can thumb his no se at so cial strictures and engage in vio lence and mass murder witho ut risking cen- sure. The game pro vides a safe way to thumb o ne’s no se.
Proving O neself
Ano ther functio n o f games is as a means o f demo nstrating pro wess. All games suppo rt this mo ti- vatio n to a greater o r lesser degree. Many game-playing co mmunities spo nso r to urnaments o r player ratings. Arcade games suppo rt this functio n by reco rding and displaying the initials o f the to p-sco ring players. There are also players who carry this to extremes. Their prime go al is no t merely to win, but to beat so mebo dy, preferably so mebo dy wo rth beating. Chess has an unusu- ally high co ncentratio n o f such sharks; so do wargames. A co mmo n questio n asked during a wargame is "Are yo u playing fo r blo o d o r fo r fun?" Such players no rmally prefer games that allo w their skill to be pro perly bro ught to bear, so they tend to wards games in which chance plays a minimal ro le. Despite this co ncentratio n o f such players in deductive lo gic games, almo st all games have sharks preying o n the playful players. When a shark plays fo r serio us rewards ( e.g., so cial do minance) and -takes serio us risks o f failure, the crucial element o f safety is eliminated fro m the game, and the game ceases to be a game; it beco mes a co nflict.
Inasmuch as all games have the po tential fo r being played in an o verly co mpetitive way, so me peo ple who are especially sensitive to the so cial risks o f game-as-co nflict refuse to play games, fo r they do no t perceive the games to be safe. If they do play, they prefer to play games o f pure chance, no t so much to disable o r disco urage the shark as to create a situatio n in which winning is patent- ly unrelated to pro wess. If winning is arbitrary, so cial risk is eliminated and safety is resto red. It is impo ssible to design a game that is unalterably safe ( i.e., invulnerable to sharks) witho ut reso rting to pure chance as the so le determinant o f victo ry. If the game in any way allo ws indi- vidual pro wess to affect the o utco me, then the o utco me is perceivable as a reflectio n o f individ- ual pro wess. In mo st games, safety fro m so cial risk is co nferred o nto the game by the attitudes o f the players, the willingness to say, "It’s o nly a game."
Games are frequently used ( especially by adults) as so cial lubricants. The game itself is o f mino r impo rtance to the players; its real significance is its functio n as a fo cus aro und which an evening o f so cializing will be built. Card games and so me light bo ard games serve this functio n. An excel- lent example o f such a so cial lubricant game is a game utilizing a large plastic gamebo ard abo ut fo ur feet square that is marked with co lo red spo ts. O n each player’s turn, a rando m pro cess is used to determine which o f fo ur appendages ( arms o r legs) is to be placed o n which spo t o n the bo ard. As the players co nto rt to fulfill the game requirements, they inevitably make physical co ntact with each o ther in inno cent and fo o lishly humo ro us ways. So cial interactio n is thereby fo stered.
Exercise is ano ther co mmo n mo tivatio n to play games. The exercise can be mental o r physical o r so me co mbinatio n o f bo th; in either event, the game is an entertaining way to stay in shape. So me players like to exercise their co gnitive skills, while o thers prefer the use o f intuitio n. So me players prefer to exercise their athletic skills. Furthermo re, players need to exercise their skills at an appro priate level. A chess player will get very lit- tle exercise o ut o f a game o f tic-tac-to e. Similarly, a perso n who finds tic-tac-to e chal- lenging will get little useful exercise o ut o f chess. These preferences so rt players o ut and ro ute them to the different games available.
N eed for Acknowledgment
We all need to be ackno wledged, to be reco gnized by o ther peo ple. The ackno wledgment we crave is no t merely an ackno wledgment o f o ur existence, but o f o ur perso nalities. Fo r example, when we meet a casual acquaintance, we usually get a perfuncto ry ackno wledgment ( "Hello there, Jo nes.") We are mo re gratified when the greeting in so me way ackno wledges us as individuals with special perso nalities and pro blems ( "Hello there, Jo nes; is that knee still bo thering yo u?") The po pularity o f pets pro vide ano ther example o f the need fo r ackno wledgment. Why o n earth do we keep in o ur ho mes animals that require fo o d, veterinary attentio n, and sanitary mainte- nance? Because they ackno wledge us. We can interact with pets; we talk to them, play with them, and emo te with them. A do g is an especially respo nsive creature; it can read o ur facial expressio ns and interpret o ur to ne o f vo ice. A smile will trigger tall-wagging; a kind wo rd will precipitate jumping, licking, barking, o r so me o ther expressio n o f affectio n. Go ldfish, by co ntrast, neither appreciate no r express emo tio n. Thus, even tho ugh go ldfish are much easier to care fo r, mo st peo - ple prefer do gs as pets. Peo ple value ackno wledgment eno ugh to expend the effo rt to o btain it.
This is o ne reaso n why interactio n is so impo rtant to a game; it allo ws the two players to ackno wl- edge each o ther. A truly excellent game allo ws us to imprint a greater po rtio n o f o ur perso nalities into o ur game-playing. Such a game allo ws me to play in a way that o nly I co uld have played it. my devio usness, my entire perso nality. When such a game ends, my o ppo nent and I kno w each o ther better than we did befo re we sat do wn to play.
Many facto rs play a ro le in mo tivating a perso n to play a game. The o riginal ( and almo st instinc- tive) mo tivatio n is to learn, but o ther mo tivatio ns co me to bear as well.
MO TIVATIO N VERSUS SELECTIO N
We must be careful to distinguish between facto rs that mo tivate peo ple to play games in the first place and facto rs that allo w peo ple to cho o se between games. In o ther wo rds, the answer to the questio n, "Why do peo ple play games?" can be quite different fro m the answer to the questio n, "What makes o ne game mo re fun than ano ther?" So me facto rs mo tivate a perso n to play games; o ther facto rs help that perso n select a particular game. Fo r example, senso ry gratificatio n is such a selectio n facto r. A player who has decided to play a particular type o f game will prefer a game with excellent graphics o ver games with po o r graphics; yet the graphics alo ne will no t mo tivate many peo ple to play games. Mo tivating facto rs get peo ple to appro ach games in general; enjo y- ment facto rs help them make their cho ice o f particular games.
Distinguishing mo tivatio n fro m enjo yment is no t tantamo unt to denying co rrelatio n’s between mo tivating facto rs and enjo yment facto rs. Clearly, any game that do es no t deliver the experiences implied by the mo tivating facto r will no t be enjo yed. Thus, so me ( but no t all) mo tivating facto rs will also be used as enjo yment facto rs. If a player is mo tivated to play a game fo r mental exercise, that player will pro bably prefer tho se games that o ffer better mental exercise than do o ther games. A game canno t be fun if its facto rs do no t satisfy the mo tivatio ns o f the player. Two enjo yment facto rs that are no t in themselves mo tivatio nal are game play and senso ry gratificatio n.
Game play is a crucial element in any skill-and-actio n game. This term has been used fo r so me years, but no clear co nsensus has arisen as to its meaning. Everyo ne agrees that go o d game play is essential to the success o f a game, and that game play has so mething to do with the quality o f the player’s interactio n with the game. Beyo nd that, nuances o f meaning are as numero us as users o f the phrase. The term is lo sing descriptive value because o f its ambiguity. I therefo re present here a mo re precise, mo re limited, and ( I ho pe) mo re useful meaning fo r the term "game play". I sug- gest that this elusive trait is derived fro m the co mbinatio n o f pace and co gnitive effo rt required by the game. Games like TEMPEST have a demo nic pace, while games like BATTLEZ0NE have a far mo re deliberate pace. Despite this difference, bo th games have go o d game play, fo r the pace is appro priate to the co gnitive demands o f the game. TEMPEST requires far less planning and co n- ceptualizatio n than BATTLEZO NE; the demands o n the player are simple and direct, albeit at a slo wer pace. Thus, bo th games have ro ughly equivalent game play even tho ugh they have very dif- ferent paces. Pace and co gnitive effo rt co mbine to yield game play.
Senso ry gratificatio n is ano ther impo rtant enjo yment facto r. Go o d graphics, co lo r, animatio n, and so und are all valued by game players. They suppo rt the fantasy o f the game by pro viding sen- so ry "pro o f" o f the game’s reality. We see a related pheno meno n in the mo vies: special effects. So me o f the newer mo vies have excited great interest because o f the excellent special effects they utilize. These mo vies have placed us in the thick o f space battles, let us meet strange and wo n- derful creatures, and taken us to faraway places. The things we see lo o k so real that we believe the fantasy; we kno w ( subjectively) that the fantasy is real. Similar pro cesses can be applied to games. Special effects, graphics, so und, animatio n-these facto rs all help distinguish a go o d game fro m a bad game. We must no t co nfuse their ro le, ho wever; senso ry gratificatio n is a crucial suppo rt func- tio n, no t a central feature. Senso ry texture enhances the impact o f the fantasy created by the game o r mo vie, but wo nderful graphics o r so und do no t by themselves make the pro duct. A mo vie witho ut a believable o r enjo yable fantasy is just a co llectio n o f pretty pictures; a game witho ut an entertaining fantasy is just a co llectio n o f interactive pretty pictures.
So far I have discussed mo tivatio nal and enjo yment facto rs as if they were abso lute quan- tities who se significance is independent o f the individual player. Such is no t the case; the respo nse to a given game depends heavily o n the perso nality o f the pro spective player. Ho w are we to deal with the perso nality differences that do minate the individual's respo nse to games? O ne academic so lutio n to this pro blem is to po stulate the existence o f a very large num- ber o f perso nality traits that determine the individual respo nse to a game. We next po s- tulate a like number o f game traits that, taken to gether, co mpletely define the psycho - lo gical pro file o f the game. Next, we measure and catalo g all o f the perso nality traits o f any given individual, presumably with an o mniscient "perso nalito meter". Then we measure all the game traits o f the game in questio n with an equally po werful "gamo me- ter". We then perfo rm a matrix multiplicatio n o f perso nality traits against game traits. So metime befo re the sun enters its red giant phase, o ur mo nster co mputer returns a number telling us ho w much that perso n will enjo y that game.
This appro ach will fo r the mo ment remain a gedanken-experiment. We must devise sim- pler, admittedly less reliable means o f co ping with individual differences. O ne alterna- tive ro ute is to o bserve and catalo g gro ups o f game-players, and identify the game traits valued by these gro ups. This metho d is made difficult by the yo uth o f the co mputer game players: skill-and-actio n enthusiasts, D&D enthusiasts, and strategy gamers. There remain several o ther game types, but they have no t attracted so large a fo llo wing as to present us with a definable gro up o f players. The passage o f time and further research will certainly give us mo re info rmatio n with which to wo rk.
Individual tastes in games are no t static; as a perso n changes so do the tastes. The fo l- lo wing analo gy with music illustrates this po int. As children, we are all expo sed to music in a variety o f fo rms, but it has little impact o n us because o ur tastes are po o rly develo ped. We sing and dance to simple so ngs, but a full appreciatio n o f the emo tio nal range o f music eludes us. The po wer o f music arises fro m o ur ability to asso ciate musical expressio ns with emo tio ns. It takes years to develo p these asso ciatio ns, and they are made in the co ntext o f o ur experiences. Fo r many in my gen- eratio n, the first deep co ntact with music came with ro ck 'n ro ll in the 60’s. The po und- ing beat, simple themes, and sho rt duratio ns were easily grasped by o ur ado lescent and unso phisticated minds. We co uld understand this music. Mo reo ver, the act o f listening to and enjo ying this music was itself an educatio nal experience. As the range o f o ur musi- cal experience expanded, we learned mo re co mplex co mpo nents o f the musical lexico n and develo ped a wider range o f asso ciatio ns. So o n we were able to understand and appreciate o ther musical co mpo sitio ns previo usly inaccessible to o ur untrained ears. Ro ck music changed to reflect this maturatio n; so me o f us stayed with ro ck. O thers mo ved to jazz, co untry, o r fo lk. Like so me o thers, I mo ved fro m ro ck to classical in a series o f stages. As I mo ved alo ng this evo lutio nary path, the lesso ns o f o ne stage enabled me to understand the material o f the next stage. O ther peo ple fo llo wed their o wn paths, explo ring and learning the areas o f musical expressio n that mo st appealed to them. The co mmo n experience was that o ur musical tastes evo lved, no matter what directio n we cho se. Ro ck music was the bro ad base we all shared, the entry po int o r ,junk o ut o f which sprang many branches. Just as ro ck 'n ro ll was the entry po int into the wo rld o f music fo r an entire generatio n, so will skill-and-actio n games be the entry po int into the wo rld o f games fo r the who le po pulatio n. Like early ro ck 'n ro ll, skill-and-actio n games have bro ad appeal, and are easy to understand. As peo ple beco me mo re so phisticated with games, their tastes will evo lve do wn different branches. Like ro ck 'n ro ll, skill-and-actio n games will no t go away; they will change to reflect the evo lving taste o f the public. We can see this hap- pening already. The early arcade games are tame pussycats co mpared to the rip-sno rting, fire-breathing games o f 1982. Had TEMPEST been released in 1977, it wo uld have intim- idated and repelled players. Times change; peo ple change. Skill-and-actio n is here to stay and will always pro vide an entry po int fo r new players, but the public will no t stand still. Many peo ple will mo ve o n to explo re o ther areas o f game-playing. mo tivatio ns. I readily admit that my treatment o f the subject matter is thin, speculative, and unco mpelling. Peo ple are co mplex creatures; we will never fully understand human mo tivatio ns to play games. Yet me must appreciate the impo rtance o f these mo tivatio ns and at least try to understand them if we are to master the art o f co mputer game design.C H A P T E R T H R E E A Ta x o n o m y o f C o m p u t e r G a m e s
ho usands o f co mputer games are co mmercially available o n a variety o f hardware co nfig- uratio ns. These games present a bewildering array o f pro perties. Many sho w clo se similar- ities. Mo st po ssess so me unique design feature. Given this large sample o f games, we can
learn a great deal abo ut game design by establishing a taxo no my o f co mputer games. A taxo no - my wo uld illuminate the co mmo n facto rs that link families o f games, while revealing critical dif- ferences between families and between members o f families. A well-co nstructed taxo no my will o ften suggest previo usly unexplo red areas o f game design. Mo st impo rtant, a taxo no my reveals underlying principles o f game design. In ano ther field o f study, Charles Darwin’s meticulo us tax- o no metric wo rk while o n the Beagle led almo st inevitably to his develo pment o f the theo ry o f evo lutio n. While we canno t ho pe that taxo no metric wo rk in co mputer game studies will be so spectacularly pro ductive, it certainly seems wo rth the effo rt.
I will insist o n an impo rtant qualificatio n: I do no t claim that the taxo no my I pro po se is the co r- rect o ne, no r will I accept the claim that any co rrect taxo no my can be fo rmulated. A taxo no my is o nly a way o f o rganizing a large number o f related o bjects. If there were so me o rganizing agent, so me underlying pro cess that created the gro up o f o bjects, then we co uld reaso nably expect to be able to find a single co rrect taxo no my embo dying the central o rganizing principle in its structure. Fo r example, the wide array o f living creatures o n this earth did no t arise by chance; this array is the pro duct o f natural selectio n. Natural selectio n is a reaso nable, understandable, no narbitrary pro cess. Therefo re, there is o nly o ne reaso nable taxo no my fo r life o n earth, the taxo no my that embo dies the principles o f natural selectio n. In the shape o f an airplane we can see the principles o f aero dynamics; so to o in a taxo no my o f living creatures can we see the hand o f natural selec- tio n.
Such is no t the case with co mputer games. The field is to o yo ung, the sample to o small, fo r what- ever o rganizing principles there may be to have asserted themselves. The games we no w have are mo re the pro duct o f happenstance than the inevitable result o f well-established fo rces. Witho ut a wide array o f games there is little o ppo rtunity to cho o se between games; witho ut cho ice there can be no natural selectio n. It is therefo re impo ssible fo r us to devise a single, abso lute taxo no - my. Many taxo no mies are admissible. Indeed, attempting to co nstruct several alternative tax- o no mies is a useful way to examine the co mmo n traits o f co mputer games. I am no t so ambi- tio us; I shall be happy to pro po se just o ne taxo no my. I divide co mputer games into two bro ad catego ries: skill-and-actio n ( "S&A") games ( emphasizing perceptual and mo to r skills) and strate- gy games ( emphasizing co gnitive effo rt) . Each majo r catego ry has several subcatego ries.
This is easily the largest and mo st po pular class o f co mputer games. Indeed, mo st peo ple asso ci- ate all co mputer games with skill-and-actio n games. All arcade games are S&A games and almo st heavy emphasis o n graphics and so und, and use o f jo ysticks o r paddles rather than a keybo ard. The primary skills demanded o f the player are hand-eye co o rdinatio n and fast reactio n time.
I gro up skill-and-actio n games into six catego ries: co mbat games, maze games, spo rts games, pad- dle games, race games, and miscellaneo us games.
Co mbat games all present a direct, vio lent co nfro ntatio n. The human player must sho o t and destro y the bad guys co ntro lled by the co mputer. The challenge is to po sitio n o neself pro perly to avo id being hit by the enemy while sho o ting him. These games are immensely po pular; they are Atari’s fo rte. There are many variatio ns o n this theme, mo st arising fro m variatio ns o n the geo m- etry o f the situatio n o r the weapo nry o f the o ppo nents.
STAR RAIDERS and SPACEWAR can be co mpared o n these bases o f geo metry and weapo nry. In bo th games the player files thro ugh space in a ro cket ship and engages enemy spaceships in real- time co smic do gfights. STAR RAIDERS presents the co nflict in first-perso n geo metry ( that is, the televisio n screen sho ws the same scene that the pilo t wo uld see.) SPACEWAR uses much the same weapo nry and mechanisms with o ne crucial difference: the geo metry o f the game is third-perso n rather than first-perso n ( that is, the player sees his o wn and his o ppo nent’s spaceships fro m a dis- tance.) The difference in result is o bvio us to anyo ne who has played bo th games. The first-perso n game is mo re exciting and co mpelling than the third-perso n game. Unfo rtunately, the first-per- so n geo metry is so technically difficult to execute that it has been implemented o n o nly a few games. Mo st games use third-perso n geo metry.
ASTERO IDS is a sho o t-em-up game that uses the same space enviro n that STAR RAIDERS uses. The primary difference between the two games is in the nature o f the o ppo sitio n. The enemy in ASTERO IDS is no t a small number o f intelligent o ppo nents armed with weapo ns identical to the player’s; instead, the enemy is a large number o f stupid ro cks armed o nly with their ability to destructively co llide with the player.
MISSILE CO MMAND is ano ther co mbat game with several interesting twists. First, the player must defend no t o nly himself but also his cities fro m descending nuclear bo mbs. Seco nd, the game is a purely defensive game in that the player never has the o ppo rtunity to attack his enemy. Third, while sho ts in o ther games are very rapid events, the sho o ting pro cess in this game is slo w- er and takes time to develo p because the missiles must fly to their targets befo re deto nating. Because the time between firing and impact is so lo ng, the player must plan his sho ts with greater fo resight and make use o f multiple explo sio ns. Thus, altho ugh this is a skill-and-actio n game, there are mo re strategic elements invo lved than in many games o f this catego ry. SPACE INVADERS ( trademark o f Taito America Co rp.) is o ne o f the mo st successful co mbat games o f all time. It was o ne o f the first smash hit games and co ntributed to the upsurge o f po p- player great mo bility and MISSILE CO MMAND gives him no ne, SPACE INVADERS gives the play- er limited mo bility in o ne dimensio n o nly. As in ASTERO IDS, the player must face a multitude o f rather stupid o ppo nents who can win by to uching the player ( landing) ; in additio n, as in STAR RAIDERS, the mo nsters also sho o t back. The mo nsters march back and fo rth acro ss the screen, slo wly descending o nto the player. As the player kills mo re and mo re mo nsters, they march faster and faster. This gives the game a hypno tic accelerating tempo . SPACE INVADERS is definitely a classic.
The success o f SPACE INVADERS has spawned a who le series o f co pies and derivatives. There are a large number o f co pies who se o nly go al was to cash in o n the success o f the o riginal game. There are also several genuine derivative games. Fo r example, GALAXIAN ( trademark o f Midway) is a simple variatio n o n SPACE INVADERS. Individual invaders peel o ff and attack the player with mo re fero city than the do cile mo nsters o f the o riginal game. CENTIPEDE; is also a derivative o f SPACE INVADERS; it is different eno ugh to be a new design, but the internal game structure is very similar to the o riginal. The invaders have been gro uped into a segmented centipede; their side-to -side mo tio n is bo unded no t by the edges o f the screen but by mushro o ms rando mly scat- tered acro ss the screen. Numero us embellishments ( spiders, fleas, and sco rpio ns) extend the game co nsiderably. TEMPEST is a three-dimensio nal first-perso n derivative o f SPACE INVADERS using vecto r graphics. The amo unt o f design attentio n that SPACE INVADERS has attracted is a tribute to the game’s o riginality, appeal, and durability There are many, many o ther co mbat games. BATTLEZO NE and RED BARO N are two first-perso n co mbat games utilizing vecto r displays. O ther co mbat games include CAVERNS O F MARS, YAR’S REVENGE, CRO SSFIRE ( trademark o f O n-Line Systems) and DEFENDER ( trademark o f Williams) .
Yo u may wo nder why so many co mbat games are set in o uter space. There are three reaso ns. First, space is easy to depict and animate with a co mputer---all the designer need do is draw a blank screen with a few white do ts fo r stars. Seco nd, space is no t encumbered by the expectatio ns o f the players. A designer enco untering pro blems can always co nco ct so me super-duper zapper to so lve any design pro blems with the game and no bo dy can o bject that it is unrealistic. Earthbo und games co nstrain the designer to lo o k reality squarely in the eye---such a tireso me burden fo r a "creative" mind. Third, space is an intrinsically fantasy-laden enviro nment that enco urages sus- pensio n o f disbelief because it is unfamiliar to its audience.
Co mbat games have always been at the heart o f co mputer gaming. Players never seem to tire o f them; it appears that they will be aro und fo r a lo ng time to co me.
The seco nd subgro uping o f S&A games is the set o f maze games. PAC-MAN ( trademark o f Namco ) is the mo st successful o f these, altho ugh maze games predate PAC-MAN. The defining
So metimes o ne o r mo re bad guys pursue the player thro ugh the maze. So me maze games ( MAZE CRAZE fo r the ATARI 2600 is a go o d example) specifically require that the player make his way to an exit. O ther maze games require that the player mo ve thro ugh each part o f the maze. DO DGE 'EM is an early example o f such a game. In either case, the number, speed, and intelligence o f the pursuers then determines the pace and difficulty o f the game. PAC-MAN has a very carefully bal- anced co mbinatio n o f these facto rs. The pursuers are just slightly slo wer than the human player; their intelligence and number make up fo r this. The o verall pace o f the game makes it difficult fo r the player to fully analyze the po sitio ns o f the five pieces in real time.
Any successful game is certain to attract co pies, variatio ns, and derivatives, and PAC-MAN is no exceptio n. O ne o f the first such games fo r the ATARI Ho me Co mputer System was the first edi- tio n o f JAWBREAKERS ( trademark o f O n-Line Systems) . This game, no w remo ved fro m the mar- ket, clearly demo nstrates the difference between structural changes and co smetic changes. Structurally, it is indistinguishable fro m PAC-MAN. The play o f the game is almo st identical to that o f PAC-MAN. Co smetically, there are a number o f differences: the pursuers are faces rather than gho sts; the player is a set o f teeth rather than a head with mo uth; the maze is laid o ut dif- ferently; the so unds are different. This game pro vides a go o d example o f the metho ds that can be used to co py games while attempting to minimize legal pro blems. Ano ther PAC-MAN derivative is MO USKATTACK ( trademark o f O n-Line Systems) . This game sho ws so me structural changes relative to PAC-MAN. The player is again pursued thro ugh a maze by fo ur co mputer-co ntro lled creatures, but the basic scenario co ntains a number o f embellish- ments. First, merely passing thro ugh every po int in the maze is no t eno ugh; so me po ints, ran- do mly cho sen by the co mputer, must be passed thro ugh twice. Seco nd, the player is allo wed to fight back against the pursuers in a very different way ( setting mo usetraps) . The strategic and tac- tical effects o f this co unterfo rce capability yield a game that plays rather differently. Finally, there is a very interesting two -player game that allo ws bo th co o perative and co mpetitive strategies. In MO USKATTACK we see the basic structure o f PAC-MAN with a number o f embellishments and extensio ns that pro duce a distinct game.
The appeal o f maze games can be attributed to the cleanliness with which they encapsulate the branching structure that is a fundamental aspect o f all games. The reader will remember fro m Chapter O ne that a game has a tree structure with each branch po int representing a decisio n made by the player. In a maze game, each branch po int is neatly depicted by an intersectio n in the maze, and the o ptio ns available to the player are visually presented as the paths available at the inter- sectio n. Thus, a maze game presents a clear visual representatio n o f the branching structure o f the game. Even mo re fascinating is the lo o ping structure po ssible with maze games. A player can return to an intersectio n in the maze many times. Yet, each time he do es so , the o ptio ns he has take dif- ferent meanings because the o ther maze-inhabitants have mo ved in the interim to a different pat- tern o f po sitio ns. In this way, a small number o f displayed intersectio ns can represent a huge small number o f pro gram instructio ns, thro ugh lo o ping and branching, can address a large num- ber o f specific cases, is striking.
These games mo del po pular spo rts games. They are anachro nisms derived fro m the early days o f co mputer game design when co mputer games had no identity o f their o wn. Peo ple witho ut o rig- inal ideas fo r games fell back o n the spo rts games as mo dels aro und which to design. This also served a useful marketing purpo se: why wo uld a co nservative co nsumer buy a game with a title and subject co mpletely alien to his experience? Better to o ffer him a game he is already familiar with. Thus we have games based o n basketball, fo o tball, baseball, so ccer, tennis, bo xing, and o th- ers. All o f these games take liberties with their subject matter to achieve playability. The mo st enjo yable aspects o f the co mputer game have very little to do with the real game. This is fo rtu- nate, fo r a slavish attempt at replicatio n wo uld have pro duced a po o r co mputer game. O nly by substantially altering the o riginal games were the autho rs able to pro duce a decent design. Even so , spo rts games remain the wallflo wers o f co mputer gaming. I suspect that spo rts games will no t attract a great deal o f design attentio n in the future. No w that co mputer games have an accepted identity o f their o wn, the need fo r reco gnizable game titles has diminished.
Pad d le Gam es
I use the title "Paddle Games" to co ver the PO NG-based games. PO NG is certainly o ne o f the mo st successful and fertile o f game designs, fo r it has many grandchildren and great-grandchil- dren. The central element o f the game, that o f intercepting a pro jectile with a paddle-co ntro lled piece, has been used in endless variatio ns. The o riginal PO NG pitted two players in an electro nic versio n o f ping-po ng, hence the name. BREAKO UT was a so litaire versio n that required the play- er to chip away at a wall with the ball. The player received po ints fo r each brick destro yed. SUPER- BREAKO UT intro duced variatio ns o n this theme with mo ving walls, extra balls, and o ther tricks. CIRCUS ATARI intro duced parabo lic trajecto ries fo r the pro jectiles and a co mplex mo ving wall o f ballo o ns. WARLO RDS; to o k the genre even further; up to fo ur players ( o ne in each co rner) defend brick castles against a pro jectile bo unced aro und the field by their shield-paddles. In the abo ve games, the player uses the ball as a weapo n to batter; in o ther paddle games the play- er must o nly catch the ball, o r many balls, rather than deflect it. AVALANCHE is o ne such game. In this game, the player is at the bo tto m o f the screen and large numbers o f ro cks are failing; each o ne must be caught with the player’s piece. The game beco mes quite frantic as mo re and mo re ro cks fall at a faster and faster pace. Ano ther game, CHICKEN, ( trademark o f Synapse So ftware) expands o n this theme by replacing the ro cks with eggs and making each o ne hatch o n striking the gro und, fo rcing the player-hen to jump o ver it as she mo ves abo ut. po tential remaining, I am hesitant to pro no unce such a durable o ld system dead.
So me co mputer games invo lve a straightfo rward race. Mo st o f these games allo w the player to mo ve at co nstant speed, but extract time penalties fo r failure to skillfully nego tiate an asso rtment o f hazards. Thus, a player in the APX skiing game DOWNHILL must avo id the trees and ro cks; the player’s sco re is based o n his time to co mplete the co urse. MATCH RACER by Gebelli So ftware is a car-racing game with o il slicks and o bstacles. NIGHT DRIVER is a car-racing game featuring a first-perso n view o f the ro ad. O ne pro blem with all o f these games is that they are no t true games but puzzles, fo r there is no real interactio n in a race between a player and his o ppo nent. Indeed, it is difficult to identify the o ppo nent in these games.
A mo re invo lved variatio n o n the race game is DO G DAZE by Grey Chang. This is a true game, no t a puzzle. It presents a two -player co mpetitive race game with variable go als and asymmetric o bstacles. Each player has a do g; hydrants po p o nto the screen at rando m lo catio ns; the players must race to be the first to to uch the hydrant, thereby claiming it as their o wn. Players may no t to uch hydrants o wned by their o ppo nents o n pain o f being tempo rarily paralyzed. The game has many interesting twists and turns witho ut being o verly co mplex; it demo nstrates that the race game can be a flexible vehicle o f game design.
My taxo no my is flawed; there exist a number o f games that do no t fit into this taxo no my very well. The first I will mentio n is DO NKEY KO NG, ( trademark o f Nintendo ) a game that lo o ks vaguely like a race game with intelligent o bstacles. FRO GGER ( trademark o f ________) is ano th- er game that defies classificatio n in this taxo no my. It co uld perhaps be called a maze game with mo ving walls o r o bstacles, but the fit is po o r. APPLE PANIC by Bro derbund So ftware also defies my taxo no my. In so me ways it is like a maze game and in so me ways it is a co mbat game. The pace o f the game is o ddly slo w. I do n’t kno w what to call this game. The fact that these games do no t fit my taxo no my do es no t bo ther me o verly much; I certainly do n’t want to create ad ho c cat- ego ries fo r individual games. I am co ntent to wait and see o ther develo pments befo re I create new catego ries o r revise o ld o nes.
Strategy games co mprise the seco nd bro ad class o f co mputer games. These games emphasize co g- itatio n rather than manipulatio n. I do no t mean to imply that S&A games are devo id o f strategic co ntent; so me S&A games do indeed have a strategic element. The majo r distinguishing facto r between strategy games and S&A games is the emphasis o n mo to r skills. All skill-and-actio n games require so me mo to r skills; strategy games do no t. Indeed, real-time play is rare in strategy
Strategy games typically require mo re time to play than S&A games. Strategy games are no nexistent in the arcades; they are rare o n the ATARI 2600; they are almo st exclusively restricted to perso nal co mputers. I divide strategy games into six catego ries: Adventures, D&D games, wargames, games o f chance, educatio nal games, and interperso nal games
Ad ven t u res
These games derive fro m o ne o f the o ldest co mputer games, called "Adventure". In these games the adventurer must mo ve thro ugh a co mplex wo rld, accumulating to o ls and bo o ty adequate fo r o verco ming each o bstacle, until finally the adventurer reaches the treasure o r go al. Sco tt Adams created the first set o f Adventures widely available fo r perso nal co mputers; his so ftware ho use ( Adventure Internatio nal) is built o n tho se games. The Sco tt Adams games are pure text adven- tures that run in a small amo unt o f memo ry, so they do no t need disk drives; they are also read- ily transpo rtable to different machines. A sho rt time later Ken and Ro berta Williams built O n-Line Systems with THE WIZARD AND THE PRINCESS ( trademark o f O n-Line Systems) , an adventure that presented pictures o f the scenes in which the adventurer fo und himself. The game itself was no t particularly new; the inno vatio n was primarily the use o f graphics. Bo th firms have expand- ed their lines with mo re games using the systems they pio neered. Mo st o f these derivative games are structurally similar to the o riginals, differing in detail, po lish, and size. The next variatio n o n the adventure theme was the giant adventure, o f which there are several. TIME ZO NE by O n-Line Systems is o ne o f these. These giant adventures use multiple diskettes to link to gether a gigantic adventure. As the player so lves the puzzle in o ne enviro nment he mo ves o n to ano ther enviro nment o n ano ther disk. The games are structurally identical to earlier games; the o nly difference is o ne o f magnitude. They take many weeks o f play to so lve.
A new variatio n o n the adventure game genre is DEADLINE ( trademark o f Info co m) , a detective adventure with a number o f interesting twists. Its heritage as an adventure is evident in its lack o f graphics and its use o f an excellent sentence parser. This adventure puts the player in the ro le o f a detective attempting to so lve a murder. The game is played in a real-time mo de that adds to the interest and challenge o f the game. The player searches no t fo r treasure but fo r info rmatio n with which to so lve the murder. This game sho ws the po tential o f the adventure system in that the same system can be used, with the sto ryline and go als altered, to appeal to a different audience. O ne o f the mo st clever adventures ever do ne is Warren Ro binett’s ADVENTURE o n the ATARI 2600. This adventure fo llo ws the same basic fo rmat as all adventures, except that it uses abso lute- ly no text. Instead, the user mo ves thro ugh a series o f ro o ms presented in rather simple graphics. Altho ugh the graphics and input schemes are radically different, the basic feel o f the adventure system has been successfully retained. SUPERMAN, HAUNTED HO USE, and GALAHAD AND THE HO LY GRAIL by Do ug Cro ckfo rd are all derivatives o f this game. guished fro m games by the static nature o f the o bstacles they present to the player. Adventures present intricate o bstacles that, o nce cracked, no lo nger pro vide challenge to the player. It is true that so me adventures push clo ser to being games by inco rpo rating o bstacles such as hungry drag- o ns that in so me way react to the player. Nevertheless, they remain primarily puzzles.
A co mpletely independent thread o f develo pment co mes fro m the D&D style games. Fantasy ro le- playing was created by Gary Gygax with Dungeo ns and Drago ns ( trademark o f TSR Ho bbles) , a co mplex no nco mputer game o f explo ratio n, co o peratio n, and co nflict set in a fairytale wo rld o f castles, drago ns, so rcerers, and dwarves. in D&D, a gro up o f players under the guidance o f a "dun- geo nmaster" sets o ut to gather treasure. The game is played with a minimum o f hardware; play- ers gather aro und a table and use little mo re than a pad o f paper. The dungeo nmaster applies the rules o f the game structure and referees the game. The dungeo nmaster has autho rity to adjudicate all events; this allo ws very co mplex systems to be created witho ut the frustratio ns o f co mplex rules. The atmo sphere is quite lo o se and info rmal. Fo r these reaso ns, D&D has beco me a po pu- lar game, with endless variatio ns and derivatives.
D&D first appeared in the mid-70’s; it didn’t take lo ng fo r peo ple to realize that it had two seri- o us limitatio ns. First, the game needed a gro up o f players and a dungeo nmaster, so it was impo s- sible to play the game so litaire. Seco nd, the game co uld so metimes beco me tedio us when it required lengthy co mputatio ns and thro wing o f dice. Many peo ple reco gnized that these pro b- lems co uld be so lved with a micro co mputer. The first co mpany to make a D&D style co mputer game available was Auto mated Simulatio ns. Their TEMPLE O F APSHAI pro gram has been very successful. They also market a number o f o ther D&D-style games. So far, ho wever, few games have been marketed that truly capture the spirit o f D&D. There are sev- eral reaso ns fo r this. First, mo st D&D-players are yo ung and do n’t have the mo ney fo r such pack- ages. Seco nd, the adventure games have slo wly abso rbed many o f the ideas o f the D&D games. There was a time when we co uld easily distinguish an adventure fro m a D&D game with several facto rs.Adventures were pure text games, while D&D games used so me graphics. Adventures were puzzles; D&D games were true games. Adventures were by and large no nvio lent, while D&D games tended to be quite vio lent. Lately, we have seen adventures taking o n many o f the traits o f D&D games, so that it is no w harder to tell the difference between them. An ideal example o f this pheno meno n is ALI BABA AND THE FO RTY THIEVES ( trademark o f Q uality So ftware) , a game with the basic elements o f bo th adventures and D&D games. The play- er must search thro ugh a large maze to find and rescue a princess, but o n the way he must fight mo nsters and thieves. The player, as Ali Baba, po ssesses perso nal characteristics ( dexterity, speed, etc.) that are reminiscent o f a D&D game, but he must explo re the maze as in an adventure. Fo r these reaso ns, I feel that this game canno t be classified as either an adventure o r a D&D game, but ro le-playing ( "FRP") games. This suggests that we will see mo re such games co mbining the "search and disco ver" aspects o f adventure games with the "defeat o ppo nents" aspects o f D&D games.
A third class o f strategy games is pro vided by the wargames. No nco mputer wargames as a gaming fo rm have a lo ng heritage. Co mmercial wargaming go es all the way back to the 1880’s with an American wargame design using wo o den blo cks. The British have lo ng had a dedicated gro up o f wargamers using miniature mo dels o f so ldiers and very co mplex rules. Their games, called minia- tures games, have gro wn in po pularity and are no w played in the USA. But the largest segment o f wargamers in recent years has been the bo ardgamers. This ho bby was fo unded in the late 1950’s by Charles Ro berts, who fo unded the Avalo n-Hill Game Co mpany and created such classic games o f the 60’s as BLITZKRIEG, WATERLO O, and AFRIKA KO RPS ( all trademarks o f the Avalo n-Hill Game Co mpany) . During the 1970’s a new co mpany, Simulatio ns Publicatio ns, Inc., turned bo ard wargaming into the largest segment o f wargaming.
Wargames are easily the mo st co mplex and demanding o f all games available to the public. Their rules bo o ks read like co ntracts fo r co rpo rate mergers and their playing times o ften exceed three ho urs. Wargames have therefo re pro ven to be very difficult to implement o n the co mputer; we have, nevertheless, seen entries.
The co mputer wargames available no w fall into two distinct gro ups. The first gro up is co mpo sed o f direct co nversio ns o f co nventio nal bo ardgam es. CO MPUTER BISMARK, CO MPUTER AMBUSH, and CO MPUTER NAPO LEO NICS ( trademarks o f Strategic Simulatio ns, Inc.) are examples o f this gro up o f games. These games illustrate the fo lly o f direct co nversio n o f games o f o ne fo rm to ano ther. They parro t successful and respected bo ardgames, but are themselves no t as successful. Because they attempt to replicate bo ardgames, they are, like bo ardgames, slo w and clumsy to play. The seco nd gro up o f co mputer wargames are less slavish in their co pying o f bo ard wargames. My o wn EASTERN FRO NT 1941 is generally co nsidered to be the best o f this lo t, primarily because o f its graphics and human engineering features. Many o f the games in this catego ry are experi- mental; hence the successes are o utnumbered by the failures. Avalo n-Hill’s first entries into the co mputer wargaming arena were such experiments. My o wn TANKTICS game is an early experi- ment that o nce was the mo st advanced co mmercially available wargame ( it was the O NLY co m- mercially available wargame when I first released it in 1978) . It is no w generally regarded as a medio cre game. It can safely be said that co mputer wargaming is no t a well-develo ped area o f co mputer gaming. Fo r the mo ment, co mputer wargaming is to o clo sely asso ciated with bo ard wargaming in the minds o f the public and mo st designers; until it can shake free fro m the co n- straints o f bo ardgames and, establish its o wn identity, co mputer wargaming will evo lve slo wly.
Games o f chance have been played fo r tho usands o f years; their implementatio n o nto co mputers is therefo re quite expectable. They are quite easy to pro gram, so we have seen many versio ns o f craps, blackjack, and o ther such games. Despite their wide availability, these games have no t pro ven very po pular, mo st likely because they do no t take advantage o f the co mputer’s stro ng po ints. Furthermo re, they lo se the advantages o f their o riginal techno lo gies. These games demo n- strate the fo lly o f mindlessly transpo rting games fro m o ne medium to ano ther.
Educational and Children’s Games
The fifth catego ry o f strategy games is that o f the educatio nal games. Altho ugh all games are in so me way educatio nal, the games in this set are designed with explicit educatio nal go als in mind. This gro up is no t heavily po pulated as yet, perhaps because the peo ple interested in educatio nal uses o f co mputers have no t yet co ncentrated much attentio n o n game design. The Tho rne-EMI puzzles are go o d entries in this field, and APX sells a co llectio n o f very simple children’s games that have so me educatio nal value. Several o f the classic co mputer games are educatio nal: HANG- MAN, HAMMURABI, and LUNAR LANDER are the three mo st no tewo rthy o f these early educa- tio nal games. SCRAM ( a nuclear po wer plant simulatio n) and ENERGY CZAR ( an energy eco - no mics simulatio n) are two o f the mo re co mplex pro grams in the educatio nal games field. My favo rite entry to date is RO CKY’S BO O TS ( trademark o f The Learning Co mpany) , a children’s game abo ut Bo o lean lo gic and digital circuits. The child assembles lo gic gates to create simulated lo gical machines. This game demo nstrates the vast educatio nal po tential o f co mputer games. Educato rs are beco ming mo re aware o f the mo tivatio nal po wer o f co mputer games; with time we can expect to see mo re entries o f the caliber o f RO CKY’S BO OTS.
I have been explo ring a class o f games that fo cus o n the relatio nships between individuals o r gro ups. O ne such game explo res go ssip gro ups. The player exchanges go ssip with up to seven o ther co mputer-co ntro lled players. The to pic o f co nversatio n is always feelings, po sitive o r nega- tive, expressed by o ne perso n fo r ano ther. Adro it po sturing increases po pularity. Similar games co uld address co rpo rate po litics, so ap-o pera situatio ns, go thic ro mances, internatio nal diplo ma- cy, and espio nage. Altho ugh the catego ry is undevelo ped, I believe it is impo rtant because it addresses fantasies that are very impo rtant to peo ple. Many o ther art fo rms devo te a great deal o f attentio n to interperso nal relatio nships. It is o nly a matter o f time befo re co mputer games fo llo w a similar co urse.
This co ncludes the descriptio n o f my pro po sed taxo no my. O bvio usly, this taxo no my has many flaws. This is prim arily because the basis o f divisio n is no t any grand principle but is instead differently than D&D games. Yet, bo th game systems evo lved separately and are histo rically quite distinct. Similarly, the creatio n o f an educatio nal games catego ry is my respo nse to the effo rts o f educato rs to create educatio nal games. With the passage o f time, market fo rces will assert them- selves, and a mo re o rganized and co nsistent taxo no my will beco me po ssible. Peo ple have tried to create educatio nal games, so we no w have them. My taxo no my is a patchwo rk because the set o f available co mputer games is a patchwo rk.
This taxo no my suggests a number o f o bservatio ns abo ut the state o f game design with co mput- ers. Fo r example, it sho uld be o bvio us that there are very few basic scenario s fo r skill-and-actio n games, each scenario taking o ne catego ry. The archetypical game in each catego ry spawned a who le family o f imitato rs, variatio ns, and impro vements. Mo reo ver, the archetypical game in each catego ry was seldo m the big mo neymaker; instead, the archetypical game was fo llo wed by sever- al successo r games that impro ved o n it until o ne game hit the nail o n the head. Thus we have CO MBAT leading to SPACE INVADERS in the co mbat catego ry, DO DGE 'EM leading to PAC-MAN in the maze catego ry, and PO NG leading to SUPERBREAKO UT in the paddle catego ry.
Ano ther lesso n that arises fro m this taxo no my is that the Analo gy games are still in a very po o r- ly-develo ped state in co mpariso n to the S&A games. While S&A games have fairly clear-cut cate- go ries that make sense, the catego ries in strategy games are less satisfying and the distinctio ns between catego ries are muddier. This ambiguity suggests that much creative o ppo rtunity remains in the strategy games field. A taxo no my reflects the bo dy o f material it attempts to o rganize. The state o f co mputer game design is changing quickly. We wo uld therefo re expect the taxo no my presented here to beco me o bso lete o r inadequate in a sho rt time. New taxo no mies must be created to reflect the changes in the marketplace in the next few years. Fo r the present, ho wever, the pro po sed taxo no my can pro - vide us with an o rganized way to view the menagerie o f games while suggesting new areas to explo re. C H A P T E R F O U R T h e C o m p u t e r a s G a m e Te c h n o lo g y GAME TECHN O LO GIES
very art fo rm is expressed thro ugh a physical medium. The co ntro l and manipulatio n o f this physical medium is a technical pro blem that the artist must master befo re she can express herself thro ugh it. Thus, the sculpto r must tho ro ughly understand the limitatio ns
o f marble, brass, o r whatever medium she uses. The painter must fully understand the techno lo - gy o f paint and the behavio r o f light. The musician must be deeply skilled in the techno lo gy o f so und creatio n. So to o must the co mputer game designer tho ro ughly understand the medium with which she wo rks. The co mputer o ffers special po ssibilities and impo ses special co nstraints o n the designer. In this chapter I will discuss the nature o f these po ssibilities and co nstraints. A few examples o f a game techno lo gy o perating at a simpler level may help establish basic princi- ples. Cards are o ne such simpler game techno lo gy. We have here a very simple set o f physical equip- ment---52 pieces o f cardbo ard, imprinted o n o ne side with a unifo rm pattern, and o n the o ther side with distinct symbo ls. The key traits o f this equipment can be summarized as fo llo ws: 1) There are many cards.
2) Each card is unique. 3) Each card po ssesses a numeric value. 4) Each card po ssesses a suit, a two -bit value. 5) The identity o f a card can be selectively revealed. 6) Each card is easily assignable to an o wner.
These six characteristics are the fundamental pro perties o f the card, techno lo gy that co nstrain the design o f all card games. Each characteristic carries implicatio ns fo r game design with cards. So me things are easy to do with this techno lo gy and so me things are hard to do with it. Fo r example, games o f pro bability are easily implemented with this techno lo gy, fo r the two characteristics ( numeric value and suit) can be co mbined into many, many sets acco rding to laws o f pro babili- ty. The limitatio ns o n info rmatio n created by the cards can be used to create games o f guesswo rk and intuitio n. Indeed, o ne o f the mo st intriguing o f card games, po ker, is based no t so much o n co ld pro bability assessments as o n the deceptio ns made po ssible by the limited info rmatio n emplo yed in the game. Like every o ther techno lo gy, cards also have their weaknesses. Fo r example, it wo uld be very tricky to design a card game fo r mo re than 52 players, because there are o nly 52 cards in o ne deck. It wo uld also be very difficult to design a go o d skill-and-actio n game using cards as a techno lo gy. Ano ther to ugh design challenge wo uld be a go o d athletic game using cards. Games meeting these co nditio ns co uld be implemented with cards, but they pro bably wo uld no t be very go o d games. and o ther things can’t. Ano ther game techno lo gy, that o f the bo ardgame, is so mewhat mo re flex- ible than cards. This techno lo gy is so much mo re flexible than cards that I canno t devise a list o f defining characteristics as I co uld with cards. Bo ardgames can be described but no t rigo ro usly defined. They use a large surface o f paper o r cardbo ard o n which are printed vario us images, no r- mally taking the fo rm o f a stylized map. Frequently the area represented o n the map is divided into discrete regio ns by either a regular geo metric pattern ( rectgrid o r hexgrid) , a segmented path to be traversed, an irregular divisio n o f regio ns, o r a netwo rk o f po ints co nnected by paths. The map itself remains the same thro ugho ut the game; players designate changes in the situatio n with a set o f markers that can be mo ved abo ut o n the map. So metimes a rando mizing machine is used to determine o utco mes o f rando m pro cesses; a spinner o r dice are mo st frequently used fo r this purpo se. So metimes cards fro m a special set are drawn to pro vide this rando mizing functio n. This techno lo gy has pro ven to be very successful fo r game designers. It easily acco mmo dates gro ups o f players, and with appro priate game design can address a very wide range o f gaming sit- uatio ns. Chess is certainly the all-time classic bo ardgame. MO NO PO LY ( trademark o f Parker Bro thers) , a successful early bo ardgame, co ncerns real estate transactio ns. O ther bo ardgames have addressed such to pics as life go als, so lving a murder, and race relatio ns. The mo st ambitio us mo d- ern bo ardgames are the wargames. Amo ng these are games with bo ards o f so me 25 square feet, several tho usand mo vable pieces, and a rules manual 50 pages lo ng. A small industry has sprung up aro und these designs, co mplete with histo rical research, star designers, and its o wn jargo n. Bo ardgames pro vide a flexible and po werful techno lo gy fo r game designers. In recent years, ho w- ever, we have seen a stagnatio n in designs with the bo ard techno lo gy. Many new bo ardgames lo o k like cheap co pies o f MO NO PO LY. Wargames, after sho wing a burst o f creative energy in the 60’s and 70’s, have started to stagnate. Few fundamentally new ideas are being intro duced in this arena. It may be that we have mined this vein to the limits o f its pro ductive capacity. What are the limitatio ns o f this techno lo gy? First and fo remo st, it is very difficult to maintain privileged info rmatio n in a bo ardgame. All players can see the bo ard and the po sitio n o f all the markers. Seco nd, the mechanics o f handling all the pieces must be managed by the players. In so me cases this can beco me a sizable cho re, as in the afo rementio ned mo nster wargame. Fo r this reaso n mo st bo ardgames are lo ng affairs, frequently filling an evening. Sho rt bo ardgames playable in twenty minutes o r less are quite rare. Finally, sho uld the pieces be disturbed, a bo ardgame is easily ruined. The central po int o f the preceding discussio n is that every game utilizes so me techno lo gy, and that each techno lo gy has strengths and weaknesses, things that it can do well and things that it can do po o rly. The astute game designer must fully grasp the strengths and weaknesses o f the techno lo gy s/ he uses. Let us no w examine the co mputer as a game techno lo gy.COMPUTERS
The mo st striking feature o f the co mputer in a game co ntext is its respo nsiveness. Respo nsiveness is vital to the interactiveness that is so impo rtant to any game. The co mputer can respo nd to the human player’s wishes in a huge variety o f ways. If the actio n in a card game o r bo ard game starts to drag, the players have no cho ice but to plo d thro ugh it o r take desperate measures. There is no reaso n why a co mputer game in similar straits co uld no t speed up the game o n demand. It co uld change the length o f the game, o r the degree o f difficulty, o r the rules themselves. SPACE
INVADERS ( trademark o f Taito America) fo r the ATARI 2600 pro vides an example o f such per- fo rmance. The player can select o ne o r two -player versio ns, visible o r invisible invaders, statio n- ary o r mo ving shields, fast o r slo w bo mbs, and a variety o f o ther o ptio ns. In effect, the player cho o ses the rules under which he plays. The game is respo nsive to his wishes.
This respo nsiveness arises fro m the co mputer’s plasticity. The co mputer is dynamic; it impo ses lit- tle co nstancy o n any element o f the game. Bo ardgames, cardgames, and athletic games all have invariables that co nstrain the designer. O nce yo u have printed up 100,000 game bo ards it beco mes very difficult to mo dify the map. Try as we may, we can’t have 53-card stud; the card decks aren’t made that way. And sho uld so me miracle o f science pro duce mo re elastic fo o tballs that kick further, we will no t be able to simply extend fo o tball stadiums witho ut spending many millio ns o f do llars. The co mputer is far less restrictive. All o f the game parameters are readily changed, even during the co urse o f the game. There is no thing sto pping us fro m creating a fo o t- ball game in which the go al po st recedes fro m the visiting team. Territo ries in wargames can be switched aro und the map o f the glo be mo re easily than we mo ve a chair in the living ro o m. This flexibility is o f paramo unt impo rtance to the game designer. As yet, it has been put to little use. A seco nd feature o f great value is the co mputer’s ability to Mo tio n as game referee. All o ther game techno lo gies demand that so mebo dy take the time to handle the administrative respo nsibilities o f the game. Athletic games are mo st demanding; they require several impartial referees o r umpires to administer the rules o f the game and adjudicate disputes. Card games and bo ardgames require that the players also functio n as referees. This is seldo m a pro blem with card games, but it can be a big lo ad with bo ardgames, especially the mo re co mplex o nes such as the wargames. Rules disputes and administrative fo ul-ups are part o f the unavo idable dangers o f bo ardgames. The co mputer can eliminate all o f these pro blems. It can administer the game, freeing the player to co ncentrate o n playing it. This allo ws o ne o ther big advantage: the co mputer can implement co mplex arithmetic and lo gical rules. With o ther techno lo gies, game rules must be o verly simple because the humans implementing them canno t be trusted to perfo rm simple numerical co mpu- tatio ns. The co mputer eliminates this restrictio n. Fo r example, in the o riginal versio n o f EASTERN FRO NT 1941, I was able to use exceptio nally co mplex victo ry calculatio ns. Mo st bo ard-level wargames abo ut the eastern fro nt in Wo rld War II assign victo ry po ints fo r captured cities, and perhaps fo r casualties inflicted and sustained. A mo re co mplex calculatio n reco gnizing the realities o f the campaign wo uld be to o tedio us fo r human co mputatio n. O riginal EASTERN FRO NT 1941 was able to calculate no t o nly cities captured and the westward resistance o f every Russian unit. The game is thereby able to pro vide a mo re realis- tic and meaningful measure o f the player’s perfo rmance. The third advantage o f the co mputer is in real-time play. O ther game techno lo gies must have pauses and pro cedural delays while administrative matters are dealt with. The co mputer is so fast that it can handle the administrative matters faster than the humans can play the game. This makes real-time games po ssible. Skill-and-actio n games are the direct result. The speed o f the co mputer also eliminates the need fo r turn-sequencing so co mmo n in card games and bo ardgames. The fo urth strength o f co mputers fo r game design purpo ses is their ability to pro vide an intelli- gent o ppo nent. All o ther games require a human o ppo nent ( exceptio n: so litaire card games, but they are actually puzzles rather than games) . The greatest success so far has been with chess-play- ing games. Pro grams written fo r micro co mputers can no w play a chess game well eno ugh to chal- lenge mo st no n-rated players. These games represent the best we have achieved to date in game artificial intelligence. Mo st games are far less intelligent. Instead, they rely o n o verwhelming numerical advantage to make up fo r the superio r intelligence o f the human player. With the pas- sage o f time, we can expect to see mo re intelligent algo rithms that pro vide mo re challenging play fro m the co mputer. The fifth strength o f the co mputer is its ability to limit the info rmatio n given to the players in a purpo seful way. This capability can be o f great value. Limited info rmatio n fo rces the player to use guesswo rk. The nature o f this guesswo rk can be very intriguing. Fo r example, guessing a rando m number between o ne and ten is no t a very interesting challenge, but guessing yo ur o ppo nent’s reso urces based o n yo ur assessment o f his actio ns and perso nality is a far mo re interesting exer- cise. When the guesswo rk is included in the framewo rk o f a co mplex and o nly partially kno wn system, the challenge facing the human player takes o n a decidedly real-life texture. Limited info rmatio n pro vides ano ther impo rtant bo nus. Games are an unreal representatio n o f a real-wo rld pro blem. The player must use his imaginatio n to make the unreal situatio n seem real. Limited info rmatio n enco urages the use o f imaginatio n. If we kno w all the pertinent facts, we can treat the pro blem as a simple pro blem o f deductio n. But if we kno w o nly a po rtio n o f the truth, o ur minds gro pe fo r an appro priate mo del o n which to hang o ur pro jectio ns. What mo del co uld be mo re appro priate than the reality that the game attempts to re-create? We are therefo re fo rced by lack o f info rmatio n to imagine o urselves in the real-wo rld predicament po stulated by the game so that tie may deal with the pro blems impo sed by the game. In the pro cess, the illusio n o f real- ity is heightened. The game draws us into its fantasy wo rld mo re effectively. The sixth feature o ffered by co mputers is their ability to utilize data transfer o ver telepho ne lines fo r game play. The use o f teleco mmunicatio ns fo r game play makes po ssible game structures that are o ut o f the reach o f o ther techno lo gies. It allo ws us to create games with huge numbers o f play- ers. Until no w, administrative pro blems have made it necessary to limit the number o f players inseveral referees and twenty players o r mo re will require many referees. O bvio usly, games with hundreds o f players will face many administrative pro blems. Indeed, the lo gistic pro blems o f assembling all th players are themselves pro hibitive. All these pro blems are so lved by co mputers linked thro ugh a teleco mmunicatio ns netwo rk. With this techno lo gy it sho uld be po ssible to design games knitting to gether tho usands o f players scattered all o ver the co ntinent. Players co uld drift into and o ut o f the game at their whim; with large numbers o f players the co ming and go ing o f individuals will no t be detrimental to the game. Like any techno lo gy, co mputers have weaknesses as well as strengths. The first and mo st painful weakness is the limited I/ O capability o f mo st co mputers. The co mputer itself may be supremely respo nsive, but if the human player can’t tell it what he wants, o r fails to understand the co m- puter’s respo nse, the co mputer’s effective respo nsiveness is nil. In o ther wo rds, the co mputer must co mmunicate its respo nsiveness to the human; it do es so thro ugh I/ O. Mo st o utput is thro ugh graphics and so und; mo st input is thro ugh keybo ard, jo ystick, and paddle.
Graphics are the first co mpo nent o f o utput. Go o d graphics are hard to co me by. Even the Atari Ho me Co mputer System, bo asting the best graphics in the micro co mputer wo rld, has graphics limitatio ns that severely co nstrain the game designer. Yo u simply canno t sho w all the graphic details that yo u wo uld like to sho w. Fo r example, I suspect that few bo ardgame bo ards co uld be duplicated o n a single screen by this machine. The number o f co lo rs, the mixing o f text with high- reso lutio n graphics, and the size o f the bo ard all co mbine to make the task ho peless. It is po ssi- ble to use a variety o f tricks to pro duce so mething that is functio nally similar to any given game bo ard. We co uld reduce the number o f co lo rs displayed, we co uld dispense with text, and we co uld design an o versize display thro ugh which the user must scro ll. EASTERN FRO NT 1941 uses all o f these tricks, and the result is quite usable, but the game wends a to rtuo us path past the graphics co nstraints o f the co mputer.
O f co urse, the co mputer also bo asts so me graphics advantages. I have yet to see the bo ardgame that co uld sho w animatio n o r change itself aro und the way a co mputer game co uld. These sen- so ry features can dramatically increase the impact o f any game. So the graphics picture is no t all bad.
Ano ther I/ O restrictio n co mes fro m the input requirements. Input to the co mputer must co me in thro ugh the keybo ard o r the co ntro llers. This can make things very difficult fo r the game design- er. In the first place, yo u can’t say much with a jo ystick o r keybo ard. A jo ystick can say o nly five fundamental wo rds: "up", "do wn", "right", "left", and "butto n". A keybo ard can say mo re, but o nly thro ugh a lengthy and erro r-pro ne sequence o f key presses. The human who wishes to express a meaningful co mmunicatio n to the co mputer must successfully enter a lo ng and clum- sy string o f simple co mmands. Input is made even mo re difficult by the indirectness o f keybo ards and jo ysticks. There is very little abo ut such devices that directly co rrespo nds to real-wo rld activ- ities. Actio ns that are simple and o bvio us with o ther techno lo gies beco me arcane with the co m- puter. If I give yo u a bat and tell yo u that yo ur go al in baseball is to hit the ball, yo u will have few easy to figure o ut. Do yo u press H fo r "hit" o r S fo r "swing" o r B fo r "bat"? Do yo u press the START key o r press the jo ystick trigger? Perhaps yo u sho uld swing the jo ystick by its cable at the ball dis- played o n the televisio n screen. After I/ O, the seco nd weakness o f the perso nal co mputer is its single-user o rientatio n. These machines were designed fo r o ne perso n to use while a seated at a desk. If two peo ple are to use it, they may be fo rced to exchange seats, a clumsy and distracting pro cedure. With jo ysticks o r paddle co ntro llers the pro blem is diminished but no t eliminated. This is o ne reaso n why so many co mputer games are so litaire and has led to the accusatio n that co mputer games are anti-so cial. A bo ardgame invites a gro up o f peo ple to sit aro und the table. A co mputer game enco urages o ne player, accepts two , and disco urages mo re.
The final weakness o f the co mputer to be co nsidered here is the requirement that it’s pro - grammed. No o ther game techno lo gy impo ses so harsh a demand o n the game designer. The bo ardgame designer can sketch an adequate bo ard and co nstruct so me simple playing pieces that will serve quite effectively. When the time co mes to pro duce the game, the designer’s amateur effo rts can be handed to a pro fessio nal who can pro duce a quality versio n o f the pro to types made by the designer. Fo r this reaso n the designer need no t co ncern himself with the technical aspects o f game pro ductio n. The co mputer game designer do es no t have life so easy. The design must be implemented o n the co mputer by pro gramming it. Pro gramming itself is a tedio us and difficult pro cess, and it is no t easily delegated, fo r the pro gramming effo rt exerts a majo r influence o ver the design pro cess. Implementing a design well is a majo r hurdle fo r any co mputer game designer.
DESIGN PRECEPTS FOR COMPUTER GAMES
Ho w do we translate an understanding o f these strengths and weaknesses o f the co mputer into a set o f guidelines fo r game designers? The characteristics described abo ve imply a variety o f pre- cepts.
PRECEPT # 1: GO WITH THE GRAIN
( Intro ducing o ur idio t carto o n hero . A ro cket lies o n its side. A wheel-less baby carriage lies near- by. O ur hero is walking fro m the baby carriage to ward the ro cket, carrying so me baby carriage wheels and a hammer.) The first-precept can be summarized with the apho rism: "Wo rk with the grain o f the machine, no t against it." To o many game designers set o ut with unrealistic go als. They attempt to fo rce the machine to perfo rm tasks fo r which it is no t well-suited. In saying this, I do no t excuse lazy pro - gramming. We must remember that the co mputer is the servant o f the human; the co nvenience fro m the co mputer, to make it wo rk its best. We can o nly do this by making it perfo rm functio ns which it perfo rms well. Case In Po int: Hexgrids An example o f this principle might be illuminating. Bo ard wargames are traditio nally executed o n maps that use a hexgrid system. This regularizes mo vement and defines po sitio ns. Hexgrids are preferred o ver rectgrids fo r several reaso ns. First, rectgrids have diago nals; two units can be diago nally adjacent. This situatio n can be very messy; rules to co pe with it are always bur- denso me and co nfusing. Hexgrids have no diago nals, so they eliminate the pro blem. Seco nd, hexgrids allo w a player a cho ice o f six directio ns in which to mo ve, while rectgrids o ffer o nly fo ur directio ns. The greater range o f cho ice allo ws the player to co ntro l mo re finely the mo vements and po sitio ning o f his pieces. It therefo re seems natural that designers o f co mputer wargames wo uld also use hexgrids fo r their maps. Indeed, mo st co mputer wargames do so ---but it is a terrible mistake. The hex do es have advantages, but it impo ses a penalty o n co mputer wargames that do es no t apply to bo ardgames. Yo u can print anything yo u desire o n a piece o f paper, but the graphic display o f the co mputer is no t so acco mmo dating. The display system o f the televisio n set is fundamentally rectangular in its architecture. Ho rizo ntal lines are stacked in a vertical sequence. Such a display can very easily handle rectangular shapes; hexago nal shapes just do n’t wo rk very well. To draw a hex the pro gram must draw fo ur diago nal lines, each o ne co mpo sed o f a set o f staggered do ts. To make the hex- grid reco gnizable the lines must be surro unded by an exclusio n zo ne at least o ne pixel wide; this co nsumes a large po rtio n o f the screen area if the hexes are small and dense. If they are larger, less screen area is co nsumed by the gridwo rk but fewer hexes can be sho wn o n a single screen. Mo reo ver, jo ysticks canno t be easily used with hexgrids because jo ysticks are set up with rectan- gular geo metry. I do no t wish to imply that hexgrids canno t be implemented o n perso nal co m- puter displays; o n the co ntrary, they have already been implemented o n many perso nal co mput- ers. The pro blem is that they are clumsy to display, lacking in graphic detail, and difficult to use. They just do n’t wo rk smo o thly. A to po lo gically identical so lutio n has been used in a few games: ho rizo ntally staggered ro ws o f squares ( "bricks") are used in place o f hexes. This system retains the flexibility o f hexes while impo sing fewer display pro blems; it remains very difficult to use with a jo ystick. Fo r these reaso ns I went back to rectgrid fo r EASTERN FRO NT 1941. My decisio n was no t based o n laziness o r unwillingness to tackle the pro blem o f hexgrids; indeed, I had already so lved the pro blem with ano ther game ( TACTICS) and co uld easily have transpo rted the co de. The experi- ence I gained in wo rking with the earlier co de co nvinced me that hexgrids weren’t so impo rtant. The success o f EASTERN FRO NT 1941 seems to indicate that the lack o f hexgrids need no t impo se a handicap.
( No w o ur hero is plummeting earthward fro m the to p o f a cliff, furio usly flapping makeshift wings attached to his arms.) O ne o f the mo st disgusting denizens o f co mputer gamedo m is the transplanted game. This is a game design o riginally develo ped o n ano ther medium that so me misguided so ul has seen fit to reincarnate o n a co mputer. The high incidence o f this practice do es no t excuse its fundamental fo lly. The mo st genero us reactio n I can muster is the o bservatio n that we are in the early stages o f co mputer game design; we have no sure guidelines and must rely o n existing techno lo gies to guide us. So me day we will lo o k back o n these early transplanted games with the same derisio n with which we lo o k o n early aircraft designs based o n flapping wings. Why do I so vehemently deno unce transplanted games? Because they are design bastards, the ille- gitimate children o f two techno lo gies that have no thing in co mmo n. Co nsider the wo rst example I have disco vered so far, a co mputer craps game. The co mputer displays and ro lls two dice fo r the player in a standard game o f craps. The co mputer plays the game perfectly well, but that is no t the po int. The po int is, why bo ther implementing o n the co mputer a game that wo rks perfectly well o n ano ther techno lo gy? A pair o f dice can be had fo r less than a do llar. Indeed, a stro ng case can be made that the co mputer versio n is less successful than the o riginal. Apparently o ne o f the appeals o f the game o f craps is the right o f the player to shake the dice himself. Many players share the belief that pro per grip o n the dice, o r speaking to them, o r perhaps kissing them will impro ve their luck. Thus, the player can maintain the illusio n o f co ntro l, o f participatio n rather than o bser- vatio n. The co mputer pro vides no ne o f this; the mathematics may be the same, but the fantasy and illusio n aren’t there. In o ne way o r ano ther, every transplanted game lo ses so mething in the translatio n. It may also gain so mething, but it always lo ses so mething. This is because any game that succeeds in o ne techno lo gy do es so because it is o ptimized to that techno lo gy; it takes maximum advantage o f the strengths and avo ids the weaknesses. The transplanted versio n uses the same design o n a differ- ent set o f strengths and weaknesses; it will almo st certainly be a lesser pro duct. Any memo rable artistic expressio n is as much a creature o f its vehicle o f expressio n as it is an image o f a tho ught. Shakespeare reads best in Elizabethan English; translatio n to mo dern English lo ses so me o f the verve and linguistic panache that we find so entertaining. The rheto ric o f Iso crates, dull and drab in English, acquires a co mpelling cadence in Greek that thrills the listener. Great bo o ks that to uched o ur so uls when we read them almo st always disappo int us when we see their mo vie adaptatio ns. Why sho uld co mputer games be immune to this law o f lo ss o n translatio n?
PRECEPT # 3: DESIGN ARO UN D THE I/O
( No w o ur man is putting the final to uches o nto a gigantic and co mplex machine with pipes, valves, smo kestacks, and many wires. O n the fro nt face o f the machine is a sign that reads, right o f this are a pair o f illuminable signs, o ne reading, "YO U WIN! ! ! ", the o ther reading "YO U LO SE! ! ! " ) As I mentio ned earlier, the co mputer’s ability to calculate is a strength, but it’s I/ O is a weakness.
Thus, the primary limitatio n facing the co mputer game designer is no t in the machine’s ability to perfo rm co mplex co mputatio ns, but in the I/ O : mo ving the info rmatio n between the co mputer and the human player. The game must be designed in such a way that the info rmatio n given to the player flo ws naturally and directly fro m the screen layo ut and so und o utput. I have seen far to o many games with go o d game structures that were ruined by po o r I/ O structures. The user was never able to appreciate the architectural beauties o f the game because they were buried in a co n- fusing display structure. Even wo rse are the games that spo rt po o r input arrangements, especially po o r use o f the keybo ard. Mo st game players find keybo ards difficult to use smo o thly. Difficulty can in so me cases create challenge, but difficulties with keybo ards generate o nly frustratio n. The implementatio n o f the game will be do minated by the limitatio ns o f I/ O. What can and canno t be displayed, what can and canno t be inputted, these things must decide the shape o f the same. A co mpariso n o f two o f my o wn games pro vides an excellent example o f the impo rtance o f I/ O structures. EASTERN FRO NT 1941 and TANKTICS ( trademark o f Avalo n-Hill) are bo th wargames dealing with Wo rld War II. Bo th pro vide reaso nably intelligent o ppo nents, co mplex detailed sim- ulatio n, a rich variety o f o ptio ns, and tho ught-pro vo king strategic challenges. In all these respects, they are ro ughly equivalent. They differ primarily in their I/ O. EASTERN FRO NT 1941 was designed aro und its I/ O ; it pro vides clean, info rmative graphics and an intuitively o bvio us jo y- stick input system. By co ntrast, TANKTICS was designed aro und its game structure; its keybo ard input system is clumsy and co nfusing and its alphanumeric; screen display is cryptic. EASTERN FRO NT 1941 has been acclaimed by the critics and has received awards; TANKTICS has been panned. The quality o f a game’s I/ O structure is crucial to its success.
PRECEPT # 4: KEEP IT CLEAN
( O ur hero at the co ntro ls o f his custo m mo to rcycle, 20 feet lo ng, equipped with numero us rear- view mirro rs, po wer steering, brakes, and thro ttle, adjustable seats, adjustable handlebars, wind- shield wipers o n several windshields and o n each mirro r, televisio n, hamburger dispenser, etc. The artist can use imaginatio n here.) Many game designers fail to keep the o verall structure o f their game clo se to heart as they devel- o p the details o f the game structure. As they enco unter design pro blems, they reso rt to quick patches that are grafted o nto the main game structure witho ut due regard to the impact such grafts have o n the o verall cleanliness o f the design. A game must have artistic unity if it is to have emo - tio nal impact o n its audience. Artistic unity can o nly be achieved by sticking clo se to the theme and eschewing distracting details. itating nature o f dirt is seldo m reco gnized, because dirt also endo ws a game with "co lo r", name- ly the texture o r feel that makes the game seem real. It is true that pro per use o f this kind o f co lo r will indeed enhance a game. Ho wever, the game designer must realize that co lo r is o btained at the price o f a certain amo unt o f dirt. The critical quantity then beco mes the ratio o f co lo r to dirt. The designer always desires the highest po ssible ratio , but so metimes, to increase the abso lute amo unt o f co lo r, s/ he must accept so me mo re dirt. In all cases, the inclusio n o f dirt into a game must be a co nscio us trade-o ff o n the part o f the game designer, no t an accident springing fro m the desire to quickly reso lve so me irritating pro blem.
Dirt mo st o ften arises fro m special-case rules that are applied rarely. Fo r example, EASTERN FRO NT 1941 has a number o f special-case rules that add dirt to the game. The wo rst is the rule fo rbidding Finnish units to attack. Inasmuch as there are o nly two Finnish units, this rule has very little significance to the game as a who le, yet the player must still be aware o f it. It clutters up the game and the player’s mind witho ut adding much. ( I had to put it in to so lve a design pro blem: what’s to sto p the Finns fro m taking Leningrad all by themselves?) A less dirty rule pro vides that Axis allies ( Rumanian, Hungarian, and Italian units) fight with less determinatio n than the Germans. There are six o f these units in EASTERN FRO NT 1941; thus, the rule is no t quite so special a case and hence no t quite so dirty.
There is a rule in EASTERN FRO NT 1941 that armo red units mo ve faster than infantry units. EAST- ERN FRO NT 1941 has many armo red units; thus, this rule is no t a particularly special case, because it applies to a go o dly po rtio n o f all units. It is therefo re no t dirty.
I can generalize these o bservatio ns by saying that the narro wer the range o f applicatio n o f a rule, the dirtier it is. My precept against dirt thus requires the designer to fo rmulate a set o f rules that co ver the entire game situatio n witho ut reco urse to special case rules. In the perfect game design, each rule is applied universally. We can never achieve the perfect design, but we can and sho uld strive to give each rule the widest po ssible applicatio n. The player must co nsider the implicatio ns o f each rule while making every decisio n in the game.
There is a scho o l o f game design that I derisively label the "humo ngo us heap" scho o l o f game design. Perpetrato rs o f this philo so phy design a game by selecting a simple structure and piling o nto it the largest po ssible jumble o f special o dds and ends ( they call them "features") . These peo ple design with a sho vel instead o f a chisel. They co nfuse magnitude with magnificence, intri- cacy with insight.
PRECEPT # 5: STO RE LESS AN D PRO CESS MO RE
( O ur idio t is juggling. Beside him ano ther man is juggling five o r six numbers co mfo rtably and happily. The idio t is staring upward in stark terro r, arms o utstretched in a futile attempt to catch an avalanche o f numbers that will simply crush him.) rily an info rmatio n sto rage device; it is instead an info rmatio n pro cessing device. Info rmatio n sto rage is a necessary preco nditio n fo r info rmatio n pro cessing, but it is no t an end in itself. Greater amo unts o f sto red info rmatio n permit greater amo unts o f info rmatio n pro cessing, but if the pro cessing capability is insufficient to realize the full po tential o f the sto rage, then that sto r- age is wasted. The ideal pro gram strikes the o ptimum balance between sto rage and pro cessing. Mo st game pro grams I have seen are lo ng o n sto rage and sho rt o n pro cessing. This is because data fo r sto rage facts are easier to co me by than pro cess-intensive material-pro gram co de. In taking the path o f least resistance, mo st game designers end up go ing do wnhill. Thus, a game that spo rts huge quantities o f static data is no t making best use o f the strengths o f the machine. A game that emphasizes info rmatio n pro cessing and treats info rmatio n dynamical- ly is mo re in tune with the machine. Relegate all static info rmatio n to a rules bo o k; paper and ink are still a better techno lo gy than perso nal co mputers fo r sto ring static info rmatio n. Info rmatio n that lies aro und and do es little, that must be dusted o ff befo re using, has no place inside the micro co mputer. As yo u lo o k o ver yo ur pro gram listing, yo u sho uld inspect each byte and ask yo urself, "Am I getting my mo ney’s wo rth fro m this byte? Is it wo rking hard fo r me, do ing useful things frequently? O r is this a lazy byte that sits idle fo r ho urs and is used o nly rarely?" Fill yo ur pro gram with active bytes that do things, no t lazy bytes. Lazy bytes are o ften asso ciated with dirty rules ( they like to hang o ut to gether in sleazy po o l halls) . Dirty rules are special cases that o ccur rarely. If they o ccur rarely, the bytes asso ciated with them are no t used o ften, hence they are lazy bytes. Ano ther argument in favo r o f this precept arises fro m mo re fundamental co nsideratio ns o n the nature o f game play. Interactiveness is a central element o f game enjo yment. As mentio ned earli- er, the co mputer’s plasticity makes it an intrinsically interactive device. Yet, the po tential inherent in the co mputer can easily go unrealized if it is pro grammed po o rly. A pro gram emphasizing stat- ic data is no t very dynamic. It is no t plastic, hence no t respo nsive, hence no t interactive. A pro cess- intensive pro gram, by co ntrast, is dynamic, plastic, respo nsive, and interactive. Therefo re, sto re less and pro cess mo re. O ne last argument has mo re to do with games than co mputers. ( Yo u will remember fro m Chapter O ne that a game is distinguished fro m a sto ry by the netwo rk o f o ptio ns that a game has, as o ppo sed to the single richly-develo ped thread o f a sto ry. Much o f the quality o f a sto ry is derived fro m the richness o f the info rmatio n it co ntains. A sto ry is thus all info rmatio n and no pro cess- ing. A game derives its quality fro m the richness o f the netwo rk o f o ptio ns it presents. These o ptio ns are o nly accessible thro ugh the pro cess-intensive aspects o f the game. Games that are info rmatio n-rich and pro cess-po o r are clo ser to sto ries than to the ideal game.
( O ur hero is no w a po le vaulter handcuffed to a high jumper. They are attempting to leap; their attempt is o bvio usly go ing to co llapse in a tangle o f limbs. Their facial expressio ns indicate that they are aware o f the likely o utco me.) Games must be designed, but co mputers must be pro grammed. Bo th skills are rare and difficult to acquire, and their co mbinatio n in o ne perso n is even mo re rare. Fo r this reaso n many peo ple have attempted to fo rm design teams co nsisting o f a no ntechnical game designer and a no nartis- tic pro grammer. This system wo uld wo rk if either pro gramming o r game design were a straight- fo rward pro cess requiring little in the way o f judicio us trade-o ffs. The fact o f the matter is that bo th pro gramming and game design are desperately difficult activities demanding many painful cho ices. Teaming the two experts to gether is rather like handcuffing a po le vaulter to a high jumper; their resultant disastro us perfo rmance is the inevitable result o f their co nflicting styles.
Mo re specifically, the designer/ pro grammer team is bo und to fail because the designer will igno - rantly make unrealistic demands o n the pro grammer while failing to reco gnize go lden o ppo rtu- nities arising during the pro gramming. Fo r example, when I designed the game ENERGY CZAR ( an energy-eco no mics simulatio n game) , I did no t include an o bvio usly desirable pro visio n fo r reco rding the histo ry o f the player’s actio ns. During the final stages o f the game’s develo pment, virtually everyo ne asso ciated with the pro ject suggested such a feature. Fro m technical experience, I knew that this feature wo uld require an excessive amo unt o f memo ry. A no ntechnical designer wo uld have insisted upo n the feature, o nly to face the disaster o f a pro gram to o big to fit into its allo wed memo ry size. Ano ther example co mes fro m EASTERN FRO NT 1941. While writing the co de fo r the calendar co mputatio ns, I realized that a simple insertio n wo uld allo w me to change co lo r register values every mo nth. I to o k advantage o f this o ppo rtunity to change the co lo r o f the trees every mo nth. The impro vement in the game is small, but it co st me o nly 24 bytes to install, so it pro ved to be a very co st-effective impro vement. A no ntechnical game designer wo uld never have no ticed the o ppo rtunity; neither wo uld a no nartistic pro grammer. There is no easy way to pro duce go o d co mputer games. Yo u must start with a go o d game design- er, an individual with artistic flair and a feel fo r peo ple. That perso n must then learn to pro gram. The o ppo site directio n o f develo pment ( fro m pro grammer to designer) will no t wo rk, fo r pro - grammers are made but artists are bo rn. When eventually yo u get that rare individual who is bo th designer and pro grammer, then yo u can subo rdinate designers and pro grammers underneath her, so as to multiply her creative po wer. In the pro cess, the subo rdinates will receive valuable train- ing. In all cases, the creative pro cess must be unified in a single mind. Co mmittees are go o d fo r generating red tape, deferring decisio ns, and shirking respo nsibility, but they are useless when it co mes to creative effo rts.
In this chapter I have discussed the co mputer as a techno lo gy fo r game design. Discussio ns o f co mputers and their impact o n so ciety tend to beco me po larized between the "gee whiz scho o l and the cynical scho o l. The fo rmer gro up sees a ro sy future o f co untless triumphs wro ught by the co mputer -- "Every day in every way, better and better." The latter gro up sees co mputers as a dehu- manizing threat, a waste o f time, o r yet ano ther vehicle fo r the expressio n o f human perfidy. In this chapter, I have tried to present co mputers as just ano ther techno lo gy, like hammer and nails, clay and sto ne, paper and ink. Like any techno lo gy, they can do so me things well. Like any, tech- no lo gy, they do so me things po o rly. The artist’s ro le is to devio usly evade their weaknesses while capitalizing their strengths to greatest advantage. C H A P T E R F I V E T h e G a m e D e s ig n S e q u e n c e
ame design is primarily an artistic pro cess, but it is also a technical pro cess. The game designer pursues grand artistic go als even as she grinds thro ugh mo untains o f co de. During the pro cess o f develo ping the game, she inhabits two very different wo rlds, the
artistic wo rld and the technical wo rld. Ho w do es o ne manage the integratio n o f such dissimilar wo rlds? In sho rt, ho w do es o ne go abo ut the pro cess o f designing a co mputer game? In previo us chapters I have to uched o n so me o f the questio ns related to this pro cess; I have also laid do wn a few precepts. In this chapter I will suggest a pro cedure by which a co mputer game co uld be designed and pro grammed. The pro cedure I will describe is based o n my o wn experiences with game design, and reflects many o f the practices that I use in designing a game. Ho wever, I have never used this pro cedure in a step-by-step fashio n, no r do I reco mmend that any perso n fo llo w this pro cedure exactly. In the first place, game design is far to o co mplex an activity to be reducible to a fo rmal pro cedure. Furthermo re, the game designer’s perso nality sho uld dictate the wo rking habits she uses. Even mo re impo rtant, the who le co ncept o f fo rmal reliance o n pro cedures is inimical to the creative imperative o f game design. Finally, my experience in game design is primarily with perso nal co m- puters, so my suggestio ns are no t co mpletely applicable to arcade game designers o r ho me video game designers. I therefo re present this pro cedure no t as a no rmative fo rmula but as a set o f sug- gested habits that the pro spective game designer might wish to assimilate into her existing wo rk pattern. With these impo rtant qualificatio ns in mind, let us pro ceed.
CHOOSE A GOAL AND A TOPIC
This vitally impo rtant step seems o bvio us, yet is igno red time and time again by game designers who set o ut with no clear intent. In my co nversatio ns with game designers, I have many times dis- cerned an indifference to the need fo r clear design go als. Game designers will admit under clo se examinatio n that they so ught to pro duce a "fun" game, o r an "exciting" game, but that is mo re o ften than no t the extent o f their thinking o n go als. A game must have a clearly defined go al. This go al must be expressed in terms o f the effect that it will have o n the player. It is no t eno ugh to declare that a game will be enjo yable, fun, exciting, o r go o d; the go al must establish the fantasies that the game will suppo rt and the types o f emo - tio ns it will engender in its audience. Since many games are in so me way educatio nal, the go al sho uld in such cases establish what the player will learn. It is entirely appro priate fo r the game designer to ask ho w the game will edify its audience.
The impo rtance o f a go al do es no t beco me o bvio us until later in the game design cycle. The cru- cial pro blems in game develo pment with micro co mputers are always pro blems o f trade-o ffs. Everything that the game designer wants to do with her game co sts memo ry, and memo ry is always in sho rt supply with micro co mputers. Thus, the designer must make trade-o ffs. when yo u must face the awful decisio n o f rejecting o ne o f two highly desirable features, the o nly criterio n yo u will have fo r making this painful cho ice will be the go al yo u have established fo r the game. If yo ur go als are clear, yo ur decisio n will be painful but o bvio us; if yo ur go als are murky, yo u may well make the wro ng cho ice, and whatever yo u cho o se, yo u will never kno w if yo ur deci- sio n was co rrect. Ho w do yo u select a pro per go al? There is no o bjective answer to this questio n; the selectio n o f a go al is the mo st undeniably subjective pro cess in the art o f co mputer game design. This is yo ur o ppo rtunity to express yo urself; cho o se a go al in which yo u believe, a go al that expresses yo ur sense o f aesthetic, yo ur wo rld view. Ho nesty is an essential in this enterprise; if yo u select a go al to satisfy yo ur audience but no t yo ur o wn taste, yo u will surely pro duce an anemic game. It mat- ters no t what yo ur go al is, so lo ng as it is co ngruent with yo ur o wn interests, beliefs, and passio ns. If yo u are true to yo urself in selecting yo ur go al, yo ur game can be executed with an intensity that o thers will find co mpelling, whatever the nature o f the game. If yo u are false to yo urself, yo ur game will necessarily be seco nd-hand, me-to o . There are situatio ns in which it is no t quite po ssible to attain the purity o f this artistic ideal. Fo r example, I wo uld no t claim that o nly immature, childish peo ple sho uld design games fo r chil- dren. No r wo uld I suggest that go o d sho o t-’em-up games can o nly be do ne by sho o t-’em-up per- so nalities. The realities o f the marketplace demand that such games be written, and it is better that they be written by mature pro fessio nals than by simpering fo o ls. Such emo tio nally indirect games, ho wever, will never have the psycho lo gical impact, the artistic po wer, o f games co ming straight fro m the heart. O nce yo u have settled o n yo ur go al, yo u must select a to pic. The to pic is the means o f expressing the go al, the enviro nment in which the game will be played. It is the co ncrete co llectio n o f co n- ditio ns and events thro ugh which the abstract go al will be co mmunicated. Fo r example, the go al o f STAR RAIDERS apparently co ncerns the vio lent reso lutio n o f anger thro ugh skillful planning and dexterity. The to pic is co mbat in space. The go al o f EASTERN FRO NT 1941 co ncerns the nature o f mo dern war, and especially the difference between firepo wer and effectiveness. The to pic is the war between Russia and Germany. Mo st game designers start o ff by selecting their to pic, with their go als subo rdinated to their to pic. Indeed, they co mmo nly describe a game under develo pment by its to pic rather than its go al. When I tell o ther designers that I am wo rking o n a game abo ut leadership, I am met with quizzi- cal expressio ns. Is it a space game, o r a wargame, o r a dungeo n game, they wo nder; they seem sat- isfied when I tell them it’s a game abo ut King Arthur. It is a serio us mistake to subo rdinate the go al to the to pic. Altho ugh yo ur initial flash o f inspiratio n may fo cus mo re o n the to pic than the go al, yo u must have the determinatio n to take co ntro l o f the design and impo se yo ur o wn go als o nto the to pic rather than allo wing yo urself to be swept away by the mo mentum o f the to pic.ined fo r its ability to successfully realize the go als o f the game. Many to pics carry with them so me excess emo tio nal baggage that may interfere with the go als o f the game. Fo r example, my mo st recent game design effo rt uses the Arthurian legends as its to pic. My go al in the game is to exam- ine the nature o f leadership. I fo und the Arthurian legends to be a co mpelling vehicle fo r this go al. Unfo rtunately these legends co ntain a stro ng co mpo nent o f male braggado cio , the van- quishing o f o ppo nents by brute fo rce. This theme directly co ntradicts so me o f the po ints I want to make with the game, thus weakening the utility o f this to pic fo r my ends. I find the legends so po werful and so malleable that I am willing to accept and wo rk aro und this po tential pitfall.
RESEARCH AND PREPARATION
With a go al and to pic firmly in mind, the next step is to immerse yo urself in the to pic. Read every- thing yo u can o n the to pic. Study all previo us effo rts related to either yo ur go al o r yo ur to pic. What aspects o f these earlier effo rts appeal to yo u? What aspects disappo int o r anger yo u? Make sure that yo u understand the mechanics o f the enviro nment yo ur game will attempt to represent. Yo ur game must give the authentic feel, the texture o f the real wo rld, and this can o nly be achieved if yo u firmly understand the enviro nment o f the game. While researching EXCALIBUR, I studied the histo ry o f Britain during the perio d AD 400-700. I fo und little in the histo ry bo o ks that was harmo nio us with my go al o f depicting the nature o f leadership. But in the Arthurian legends I fo und recurring themes mo re clo sely related to my go al. Yo u may well find yo urself adjusting yo ur go als as yo u perfo rm this research functio n; such erratic decisio n-making is an embarrassing admissio n o f po o rly-defined go als, but reflects an ho nest willingness to adapt to the exigencies o f the to pic-enviro nment. It is a departure fro m the ideal in which I have sinfully indulged myself many times. During this phase it is critical that yo u co mmit little to paper and abo ve all, WRITE NO CO DE! Take lo ng walks as yo u co ntemplate yo ur game. Co gitate. Meditate. Let the go al, the to pic, and the facts gleaned fro m yo ur research simmer to gether in the innards o f yo ur mind. Weave them to gether into a who le. Take yo ur time with this phase; impatience no w will lead to mistakes that will kill the game. I give myself at least three weeks to develo p a game idea in this stage befo re pro ceeding to the next step. With EXCALIBUR I expended several mo nths o n this stage. During this time I kept my fidgeting hands busy by writing an o pening graphic display that had little rel- evance to the final game. Yo u will generate during this phase a great variety o f specific implementatio n ideas fo r yo ur game. They will no t all fit to gether neatly---like any ho dgepo dge, they will require much so rting and rearranging befo re they can be used. Yo u sho uld no t wed yo urself to any o f them. A large co llec- tio n o f candidates fo r implementatio n is a useful reso urce during the design phase. A laundry list o f implementatio n ideas that must be included is a liability. Indulge yo urself in creating imple- mentatio n ideas, but be prepared to winno w them ruthlessly during design.
During the research and preparatio n phase, we came up with a lo ng list o f clever ideas that we wanted to into the game. We had agreed that the game wo uld have a feminist po int o f view with- o ut being preachy. We wanted to have a demanding bo ss, to ugh pro jects, deadlines, bro wnie po ints, o ne male chauvinist pig, neutral males, neutral females, family and ho me o bligatio ns, mento rs, and the co mpetitio n fo r the big pro mo tio n. We managed to include almo st all o f these ideas in the final design. We were no t able to integrate the family elements into the game. Every design we created failed to do justice to o ur desires. In the end, we had to discard this desirable element.
Yo u no w have a clear idea o f the game’s ideals but yo u kno w no thing o f its fo rm. Yo u are no w ready to begin the co ncrete design phase. Yo ur primary go al in the design phase is to create the o utlines o f three interdependent structures: the I/ O structure, the game structure, and the pro gram structure. The I/ O structure is the system that co mmunicates info rmatio n between the co mputer and the player. The game structure is the internal architecture o f causal relatio nships that define the o bstacles the player must o verco me in the co urse o f the game. The pro gram structure is the o rganizatio n o f mainline co de, subro utines, interrupts, and data that make up the entire pro gram. All three structures must be created simultaneo usly, fo r they must wo rk in co ncert. Decisio ns pri- marily relating to o ne structure must be checked fo r their impacts o n the o ther structures..
I prefer to start with the I/ O structure, fo r it is the mo st co nstraining o f the three. I/ O is the lan- guage o f co mmunicatio n between the co mputer and the player; like any human language, it is the funnel thro ugh which we must squeeze the avalanche o f tho ughts, ideas, and feelings that we seek to share with o ur fello w human beings. I/ O will dictate what can and canno t be do ne with the gains. I/ O is co mpo sed o f input and o utput. Unlike human languages, the two are no t symmetric. The co mputer has two means o f o utput to the human: graphics o n the screen and so und. In the future, we may see mo re exo tic devices fo r o utput fo r games, but fo r the mo ment these are the two mo st co mmo n. Graphics are the mo st impo rtant o f the two , perhaps because we humans are mo re o ri- ented to wards visio n than hearing. Fo r this reaso n, many game designers devo te a large po rtio n o f their energy to wards the design o f quality displays. Indeed, so me designers go so far as to design the display first and let the game develo p fro m the display, as extreme an example o f go al- less design as ever there co uld be.
Do n’t make the co mmo n mistake o f creating cute graphics so lely to sho w o ff yo ur ability to cre- ate cute graphics. Graphics are there fo r a reaso n: to co mmunicate. Use graphics to co mmunicate to the user fo rcefully and with feeling, and fo r no o ther reaso n. Plan functio nal, meaningful
Do n’t use graphics tricks as a crutch fo r a bad game design. If the game is dull and bo ring, no amo unt o f graphics gift-wrapping is go ing to fix it. The wo rst examples o f this mistake are the games that alternate bo ring game segments with cute but meaningless graphics displays. Use o f so und sho uld fo llo w the same rules: use it to tell the player what’s go ing o n in the game. The o nly place where striking but uninfo rmative graphics and so und can be useful is at the beginning o f the game, and then o nly if they help to establish the mo o d o r to ne o f the game.
Sto rybo ards are a graphics design to o l that tempt many game designers, fo r they are a well-devel- o ped techno lo gy fro m the film industry. They are no t appro priate to games, because sto rybo ards are an intrinsically sequential techno lo gy. Games are no t sequential, they are branching tree struc- tures. The game designer who uses an intrinsically sequential to o l risks having her designs made subtly sequential. The to o l shapes the mind o f its user; the saw suggests that we cut wo o d, and the freeway suggests that we drive wherever it takes us, no t where we cho o se to go . In like man- ner do es a sto rybo ard impress its sequentiality upo n o ur games. Devo te special care to the input structure o f the game. The input structure is the player’s tactile co ntact with the game; peo ple attach deep significance to to uch, so to uch must be a rewarding experience fo r them. Have yo u ever no ticed the tremendo us impo rtance pro grammers attach to the feel o f a keybo ard? Remember that players will do the same thing with yo ur game. A case in po int is pro vided by the games JAWBREAKER and MO USKATTACK ( trademarks o f O n-Line Systems) . In bo th games the jo ystick entry ro utine admits an unfo rtunate ambiguity when a diag- o nal mo ve is entered. This gives the player the impressio n that the jo ystick is unrespo nsive. I have seen players slam do wn the jo ystick in frustratio n and swear that they wo uld never play the damn thing again. Remember this well as yo u plan yo ur input structure: will yo ur input structure frus- trate and anger yo ur players? The input structure lies at the heart o f a fundamental dilemma all game designers must face. An excellent game allo ws the player to interact heavily with his o ppo nent, to invest a great deal o f his perso nality into the game. This requires that the game o ffer the player a large number o f mean- ingful o ptio ns, eno ugh o ptio ns that the player can express the nuances o f his perso nality thro ugh the cho ices he makes. Yet, decisio ns must be inputted, and a large number o f o ptio ns seem to require an extensive and co mplicated input structure, which co uld well be intimidating to the player. O ur dilemma, then, is that an excellent game seems to require a hulking input structure. The dilemma is reso lved thro ugh the designer’s creativity in designing a clean input structure that allo ws many o ptio ns. This do es no t co me easily. Many schemes must be co nsidered and rejected befo re a satisfacto ry so lutio n is fo und. Yet, such a so lutio n is o ften po ssible. In designing SCRAM, a nuclear po wer plant game, I faced the fo llo wing pro blem: ho w can a player co ntro l an entire nuclear po wer plant with o nly a jo ystick? At first glance, the task seems ho peless. Nevertheless, the so lutio n I eventually disco vered wo rks very well. The player mo ves a curso r thro ugh the plant stick butto n and pushes the stick up to turn o n o r increase po wer, and do wn to turn o ff o r decrease po wer. The system is simple and easily understo o d o nce the player has seen it. There is a general so lutio n, at the theo retical level, to the dilemma o f o ptio n richness versus input cleanliness; I call this so lutio n "the webwo rk". To design a webwo rk game, we start with a small number o f pieces. We then define a relatio nship that applies to all pairs o f pieces. The set o f rela- tio nships between pieces co nstitutes a webwo rk. The webwo rk can easily beco me quite co mplex, yet few pieces are required to create the webwo rk. In general, the number o f pairwise relatio nships is equal to N* ( N-1) , where N is the number o f pieces. Thus, fo ur pieces can generate 12 pairings, 8 pieces can generate 56 pairings, and 16 pieces can generate 240 pairings. With fewer pieces to manipulate the player faces fewer I/ O pro blems witho ut sacrificing a rich set o f relatio nships in the game. Backgammo n illustrates the simplicity and po wer o f webwo rk games. Backgammo n has o nly 30 pieces and 26 po sitio ns fo r them to o ccupy. The relatio nships between pieces are fairly simple and are expressed thro ugh the ability to mo ve and bump. Yet, o n any given mo ve, each piece has an o ffensive, defensive, blo cking, o r blo cked relatio nship with mo st o f the o ther pieces o n the bo ard. This is partly because almo st every o ther bo ard po sitio n in fro nt o f the piece can be reached, given the right die ro ll. It is no accident that the length o f the playing area ( 24 steps) is exactly equal to the maximum die ro ll. It had to be that way to squeeze all o f the pieces into range o f each o ther, thereby maximizing the number o f significant pairwise relatio nships. Mo st webwo rk games rely o n spatially expressed webwo rks; these are easy to depict and easy fo r the player to visualize. Few games have no n-spatial webwo rks; my o wn GO SSIP is o ne such game. Curio usly, GO SSIP uses a spatial webwo rk fo r its internal co mputatio ns even tho ugh the game webwo rk is no n-spatial. This may imply that game webwo rks are intrinsically spatial; it may equally well imply that I canno t shake my mind-set free fro m spatial webwo rks. The cho ice o f input device is an impo rtant design decisio n. I maintain that a go o d game design- er sho uld eschew the use o f the keybo ard fo r input and restrict herself to a single simple device, such as a jo ystick, paddle, o r mo use. The value o f these devices do es no t arise fro m any direct superio rity o ver the keybo ard, but rather in the discipline they impo se o n the designer. Simple input devices go hand-in-hand with simple input structures. Co mplex input devices enco urage co mplex input structures. The I/ O structure is the mo st impo rtant o f the three structures in a co mputer game, fo r it is the face o f the game that the player sees. It is the vehicle o f interactio n fo r the game. It is also the mo st difficult o f the three structures to design, demanding bo th human sensitivity and co mplete tech- nical mastery o f the co mputer. Give it the care it deserves.
The central pro blem in designing the game structure is figuring o ut ho w to distill the fantasy o f the go al and to pic into a wo rkable system. The game designer must identify so me key element fro m the to pic enviro nment and build the game aro und that key element. This key element must be central to the to pic, representative o r symbo lic o f the issues addressed in the game, manipula- ble, and understandable. Fo r example, in EASTERN FRO NT 1941, I started with the eno rmo us co mplexity o f mo dern warfare and extracted a key element: mo vement. Mo vement dictates the dispo sitio ns o f the military units. Mo ving into an enemy’s po sitio n initiates co mbat with him. Mo ving behind him disrupts his supplies and blo cks his retreat ro uts. Mo ving into a city captures it. Mo vement is no t equitable with all aspects o f war; it is, instead, the key element thro ugh which many o ther aspects o f war are expressible. It is easily manipulable and immediately understand- able.
A mo re difficult design challenge came fro m the game GO SSIP. This game addresses so cial rela- tio nships. The eno rmo us co mplexity o f the subject matter and the intricate twists and turns o f human interactio n to gether suggest that the subject is beyo nd treatment in a game. After much tho ught I was able to iso late a key element: the "statement o f affinity". O ne way o r ano ther, many o f o ur so cial interactio ns bo il do wn to o ne o f two declaratio ns: a first-perso n statement o f feel- ing ( "I rather like Sandra") , and a third-perso n statement ( "Well, To m to ld me that he do esn’t like Sandra o ne bit") . The key element encapsulates the grander array o f human interactio ns rather well. It is easily manipulable; indeed, it is quantifiable. And it is quite understandable. The iso la- tio n o f the statement o f affinity as the key element o f human interactio n made po ssible the game GO SSIP.
The nature o f manipulability assumes tremendo us impo rtance to the success o f the game. The key element must be manipulable, but in a very specific set o f ways. It must be expressively manipu- lable; that is, it must allo w the player to express himself, to do the things that he wants o r needs to do to experience the fantasy o f the game. Fo r example, in a co mbat game, sho o ting is almo st always a key element. If the player’s freedo m to sho o t is heavily restricted, the player canno t live the fantasy. At the same time, the manipulability must be co ncise. To use the co mbat game exam- ple again, if the player is required to declare the amo unt o f gunpo wder to be expended o n each sho t, he may well find the manipulability a hindrance to the game. The manipulability must be meaningful to the fantasies o f the game. Finally, the manipulability must be fo cused: the o ptio ns fro m which the player cho o ses while manipulating the key element must be clo sely related. Fo r example, in the game GO SSIP, the key element ( statement o f affinity) assumes a linear sequence o f values ranging fro m hatred thro ugh lo ve. ENERGY CZAR vio lates this principle by requiring the player to cho o se fro m a large, disco nnected set o f o ptio ns. Menu structures and use o f the key- bo ard bo th arise fro m unfo cussed key elements. mo vement and sho o ting. This is no t necessarily bad; if bo th key elements are kept simple, o r if o ne key element retains primacy, the game can be successful. Ho wever, to o many key elements vio lating to o many o f these principles will ro b the game o f its fo cus. Yo ur main pro blem with creating the I/ O structure is o verco ming co nstraints; yo ur main pro blem with creating the game structure is realizing po ssibilities. Yo ur previo us wo rk with the I/ O struc- ture defines the limitatio ns o n the structure o f the game. Yo u can take mo re liberties with the internal structure because the player will no t directly enco unter it. Fo r example, fo r the game TAC- TICS I develo ped a very co mplex co mbat algo rithm that realistically calculates the effects o f armo r-piercing sho t. The co mplexity o f this algo rithm wo uld have co nfused the player had I tried to explain it. But the player do es no t need to understand the internal wo rkings o f the algo rithm; he need o nly grasp its effects. I therefo re did no t feel co nstrained to design a simple-minded and intuitively o bvio us algo rithm. Co ncentrate an pro viding eno ugh co lo r to guarantee that the game will co nvey the authentic feel o f reality. Keep yo ur sense o f pro po rtio n while adding details. It will do yo ur game no go o d to pro vide exquisite detail and accuracy in o ne sphere while o verlo o king the mo st fundamental ele- ments in ano ther sphere.
A very co mmo n mistake many designers make is to pile to o many game features o nto the game structure. In so do ing, they create an o verly intricate game, a dirty game. As I discussed in Chapter 4, dirt is undesirable; a game is a structure that must fit to gether cleanly and well, no t a brushpile. Dirt creates a seco nd pro blem no t mentio ned in Chapter 4: it gums up the I/ O structure o f the game. Fo r example, the lo ng-range scan feature o f STAR RAIDERS do es pro vide so me nice addi- tio nal capabilities, but it adds ano ther keystro ke to be memo rized by the player. That’s dirty input. Fo rtunately this pro blem is o verridden in STAR RAIDERS, because the fantasy puts the player at the co ntro ls o f a starship, and so the player finds the intricacy o f the co ntro l layo ut a suppo rting element o f the fantasy rather than a hindrance. In mo st games, yo u may well be fo rced to give up nice elements in the game structure in o rder to maintain the quality o f the I/ O structure. O n the o ther hand, yo u may be fo rced to go back and change the I/ O structure to inco rpo rate a game fea- ture yo u are unwilling to abando n. If yo u do so , do no t simply tack o n a no w co mmand; rethink the entire I/ O structure and mo dify it so that the new co mmand fits well with the rest o f the I/ O structure.
Designing the game structure is emo tio nally very different fro m designing the I/ O structure. While designing the I/ O structure, the designer must thread a precario us path between the Scylla o f expressive po wer and the Charybdis o f expressive clarity, even while the sto rms o f hardware lim- itatio ns to ss her design to and fro . While designing the game structure, the designer finds herself o n a placid sea stretching flat to the ho rizo n. The challenge taunting her no w is "Where do yo u go ?"
The pro gram structure is the third o bject o f yo ur design attentio ns. This structure is the vehicle which translates the I/ O structure and game structure into a real pro duct. O ne o f the mo st impo r- tant elements o f the pro gram structure is the memo ry map. Yo u must allo cate chunks o f memo - ry fo r specific tasks. Witho ut such safeguards, yo u may end up expending excessive quantities o f memo ry o n mino r functio ns, and having insufficient memo ry remaining fo r impo rtant tasks. Definitio ns o f critical variables and subro utines are also necessary. Finally, so me do cumentatio n o n pro gram flo w is impo rtant. Use flo w charts o r Warnier-O rr diagrams o r whatever suits yo ur fancy. This bo o k is no t primarily co ncerned with pro gramming; if yo u need guidance o n pro gram develo pment, co nsult any o f the many excellent bo o ks o n pro gram develo pment.
Evaluation of the Design
Yo u no w have three structures in hand: the I/ O structure, the game structure, and the pro gram structure. Yo u are satisfied that all three structures will wo rk and that they are co mpatible with each o ther. The next sto p in the design phase is to evaluate the o verall design fo r the mo st co m- mo n design flaws that plague games. The first and mo st impo rtant questio n is: do es this design satisfy my design go als? Do es it do what I want it to do ? Will the player really experience what I want him to experience? If yo u are satisfied that the design do es pass this crucial test, pro ceed to the next test. Examine the stability o f the game structure. Remember that a game is a dynamic pro cess. Are there any circumstances in which the game co uld get o ut o f co ntro l? Fo r example, if the game has mo ney in it, co uld a situatio n arise in which the player finds himself the o wner o f ridiculo usly large amo unts o f mo ney? In sho rt, do es the game structure guarantee reaso nable upper and lo wer bo unds o n all values? If no t, re-examine the game structure carefully with an eye to structural changes that will right the situatio n. If yo u have no o ther o ptio ns, yo u may be o bliged to put them in by brute fo rce ( e.g., "IF MO NEY > 10000 THEN MO NEY 10000") No w pro be the design fo r unanticipated sho rtcuts to victo ry. A player who can find a way to guar- antee victo ry with little effo rt o n his part will no t derive the full benefit o f yo ur game. Insure that all unintended sho rtcuts are blo cked so that the player must experience tho se pro cesses that yo u want him to experience. Any blo cks yo u place must be uno btrusive and reaso nable. The player must never no tice that he is being shepherded do wn the primro se path. An example o f o btrusive blo cking co mes fro m the game WAR IN THE EAST ( trademark o f Simulatio ns Publicatio ns, Inc) . This wargame deals with the Eastern Fro nt in Wo rld War 11. The Germans blitzed deep into Russia but their advance gro und to a halt befo re Mo sco w. To simulate this the designers gave the Germans an o verwhelming superio rity but also gave them a supply no o se who se length was care- fully calculated to insure that the Germans wo uld be jerked to a dead halt just o utside Mo sco w. The effect was co rrect, but the means o f achieving it were to o o bvio us, to o o btrusive. no w, befo re yo u co mmit to pro gramming the game. Do no t hesitate to abo rt the game no w; even if yo u abo rt no w yo u will still have I earned a great deal and can say that the effo rt was wo rth- while. A decisio n to give up at a later stage will entail a real lo ss, so give this o ptio n careful co n- sideratio n no w while yo u can still do it witho ut majo r lo ss. Abo rt if the game no lo nger excites yo u. Abo rt if yo u have do ubts abo ut its likeliho o d o f success. Abo rt if yo u are unsure that yo u can successfully implement it. I have in my files nearly a hundred game ideas; o f these, I have explo red at length so me 30 to 40. O f these, all but eight were abo rted in the design stage.
If the game has made it this far, yo u are no w ready to co mmit yo ur ideas to paper. Until no w yo ur do cumentatio n has been sketchy, mo re alo ng the lines o f no tes and do o dles than do cuments. No w yo u are ready to prepare yo ur co mplete game do cumentatio n. First, co mmit all o f yo ur design results fro m the previo us phase to paper. Define the I/ O structure and the internal game structure. The to ne o f this do cumentatio n sho uld emphasize the player’s experience rather than technical co nsideratio ns. Co mpare this first set o f do cuments with yo ur preliminary pro gram structure no tes; adjust the pro gram structure do cuments if necessary.
This is the easiest o f all the phases. Pro gramming itself is straightfo rward and tedio us wo rk, requiring attentio n to detail mo re than anything else. Seldo m has a game failed so lely because the pro grammer lacked the requisite pro gramming skills. Games have failed to live up to their po ten- tial because the pro grammer did no t expend eno ugh effo rt, o r rushed the jo b, o r didn’t bo ther to write in assembly language, but in few cases has talent o r lack o f it been the crucial facto r in the pro gramming o f a game; rather, effo rt o r lack o f it is mo st o ften the respo nsible facto r. If yo u place all o f yo ur self-respect eggs in the pro gramming basket, I suggest that yo u get o ut o f game design and wo rk in systems pro gramming. O therwise, write the co de and debug it.
PLAYTESTING PHASE Ideally, playtesting is a pro cess that yields info rmatio n used to po lish and refine the game design.
In practice, playtesting o ften reveals fundamental design and pro gramming pro blems that require majo r effo rts to co rrect. Thus, playtesting is o ften interwo ven with a certain amo unt o f pro gram debugging. So metimes playtesting reveals that the game is to o serio usly flawed to save. A no nfatal, co rrectable flaw is usually a matter o f insufficiency o r excess: no t eno ugh co lo r, to o many pieces, no t eno ugh actio n, to o much co mputatio n required o f the player. A fatal flaw arises fro m a fundamental co n- flict between two impo rtant elements o f the game who se inco mpatibility was no t fo reseen. grammed can o nly achieve limited gains; if the game is badly defo rmed, abo rtio n is preferable to surgery. If playtesting reveals serio us but no t fatal pro blems, yo u must very carefully weigh yo ur o ptio ns. Do no t succumb to the temptatio n to fall back o n a quick and dirty patch jo b. Many times the pro blem that is disco vered in playtesting is really o nly a sympto m o f a mo re fundamental design flaw. Be analytical; determine the essence o f the pro blem. O nce yo u have determined the true nature o f the pro blem, take plenty o f time to devise a variety o f so lutio ns. Do n’t rush this pro cess; so metimes the ideal so lutio n co mes fro m an unexpected angle. Cho o se a so lutio n fo r its pro m- ise o f furthering the faithfulness o f the game to yo ur go als. Do no t o pt fo r the easiest so lutio n, but the so lutio n that best meets yo ur go als. Fo r example, while designing EASTERN FRO NT 1941, I ran into a severe pro blem with unit co unts: there were far to o many units fo r the player to co ntro l co nveniently. After wasting much time trying to devise ways to shrink the map o r directly reduce the number o f units, I eventually stumbled upo n zo nes o f co ntro l, a standard wargaming technique that extends the effective size o f a unit. The inclusio n o f zo nes o f co ntro l in the game no t o nly so lved the unit co unt pro blem; it also made the lo gistics rules mo re significant and gave the game a richer set o f strategies. I set o ut with the narro w go al o f reducing the unit co unt, but I fo und an impro vement with much bro ader implicatio ns. If yo ur initial design was well-develo ped ( o r yo u are just plain lucky) the game will no t face such crises; instead, the pro blems yo u will face will be pro blems o f po lish. All o f the little things that make a game go will be o ut o f tune, and the game will mo ve like a drunken dino saur instead o f the lithe leo pard yo u had envisio ned. Tuning the game will take many weeks o f wo rk. Fo r the sho rt term yo u can scrimp o n the tuning while yo u are wo rking o n o ther pro blems, fo r tuning the game requires delicate adjustments o f all the game facto rs; any o ther changes will o nly thro w o ff the tune. Therefo re, defer final tuning wo rk until the very end o f the po lishing stage. There are actually two fo rms o f playtesting. The first is yo ur o wn playtesting do ne in the final stages o f debugging. The seco nd fo rm co mes later when yo u turn o ver the game to o ther playtesters. The salient-difference between the two lies in the nature o f the bugs expo sed. Yo ur o wn playtesting sho uld reveal and eliminate all pro gram bugs ( arising fro m flaws in the pro gram structure) and many o f the game bugs ( arising fro m flaws in the game structure) . The game yo u give to the playtesters sho uld be free o f pro gram bugs; they sho uld disco ver o nly bugs in the game structure. There is no po int in sho wing an inco mplete game to playtesters, and indeed there is a danger in co ntaminating their o bjectivity by sho wing them a versio n o f the game to o early. But the time will co me when yo u feel that the game is very clo se to co mpletio n, and yo ur o wn sto ck o f ideas fo r impro vements is dwindling. This is the time to sho w the game to a few select playtesters.ask them what they think o f the game. Yo u need playtesters who po ssess a deep familiarity with games, playtesters who can analyze and criticize yo ur game with so me basis o f experience. Ideally the playtesters wo uld themselves be game designers, fo r they wo uld then share yo ur appreciatio n fo r the trade-o ffs essential to go o d game design. Yo u sho uld also kno w the player well, bo th his perso nality and his game taste. Yo u sho uld never use mo re than five o r six playtesters. A surplus o f playtesters o nly insures that yo u will no t be able to assess carefully the reactio n o f each playtester. A variety o f o ther systems have been used fo r playtesting. Mo st rely o n gathering large gro ups o f "real peo ple" and assessing their reactio ns to the game. I have little respect fo r such systems. Altho ugh they are scientific, o bjective, and demo cratic, they seldo m yield useful design info rma- tio n, fo r co nsumers make lo usy critics. The suggestio ns they make are inane and impractical; they do n’t kno w eno ugh abo ut co mputers o r games to make practical suggestio ns. Such metho ds may well wo rk with detergent and shaving cream, but I very much do ubt that any great mo vie, bo o k, o r so ng was created thro ugh market research o f this kind. I will co ncede that such metho ds can pro ve to be a useful way to guide the mass pro ductio n o f cheap games by designers o f limited tal- ents; this bo o k is no t directed to perso ns o f such a mentality. The playtesters will need a prelim- inary manual fo r the game. It need no t be a finished pro duct any mo re than the game itself---just eno ugh o rientatio n info rmatio n to get the playtester go ing with the game. Make sure that there is eno ugh in the manual that the playtester do esn’t waste time critiquing pro blems o f the game that will be so lved by the manual. Do no t sit do wn with the playtester in advance and co ach him thro ugh the game; yo u will o nly co ntaminate his o bjectivity. The playtester’s first reactio n to the game is yo ur best feedback o n the success o f the manual . Let the playtester experiment with the game fo r perhaps a week befo re yo u meet with him. Do no t ask the playtester to keep lengthy written reco rds o f play perfo rmance; he wo n’t do it. Instead, include in the manual a few sugges- tio ns abo ut po tential pro blems that wo rry yo u. The mo st fo r which yo u sho uld ask in writing is a simple reco rd o f game o ptio ns selected and subsequent sco res. Schedule alo ng interview with the playtester after he has had eno ugh time to digest the game. Co me to the interview prepared with a set o f standard questio ns that yo u ask all playtesters. Do no t lead the playtester’s answers and do n’t so licit praise. Yo ur jo b is to find flaws; acco lades co me later. While it is mo re scientific to use a third perso n to co nduct the interview ( thereby assuring mo re ho nest answers) , this impo ses a middleman between yo u and yo ur playtesters. I prefer to get the info rmatio n directly fro m the playtester. I also prefer to take a very negative tack during the interview, enco uraging the playtester to criticize the game alo ng with me and to suggest means o f impro ving it. Playtesters’ criticisms are difficult to evaluate. Mo st criticisms must be rejected fo r a variety o f rea- so ns. So me are inco mpatible with yo ur go als; so me are no t achievable in the-memo ry space remaining. So me are reaso nable, but wo uld require majo r so ftware surgery inco mmensurate with right; waste no time implementing them. Ho w do yo u tell the go o d 10% ? This is the stuff o f wis- do m; I certainly do n’t kno w. The final stage o f the design cycle is devo ted to po lishing the game. The po lishing stage is actual- ly co ncurrent with the later stages o f playtesting and may invo lve several iteratio ns with the playtesters. This stage is critical; the designer has been wo rking o n the game fo r a lo ng time by no w and the luster o f the new design has wo rn o ff. It is no w o nly a big jo b that sho uld have been finished mo nths ago . The playtesters lo ve it, the publisher lo ves it and wants it right no w, and the designer is sick o f it. The urge to dump the damn thing is o verpo wering. Resist this urge; press o n relentlessly and po lish, po lish, po lish. Keep testing the game, fine-tuning it, and adding tiny embellishments to it. O nce it’s o ut the do o r, it’s go ne fo rever. Every single game I have do ne has fo llo wed the same pattern: I po lished the game until I was sick o f it and never wanted to see it again. When at last I sent the game o ut, I rejo iced; I was free o f that do g at last. Within a mo nth I was regretting my impatience and wishing I co uld have a chance to clean up that o ne embar- rassing bug that I had never no ticed. Within three mo nths my regret had turned into shame as I disco vered o r was to ld o f many mo re bugs. I have pro grams o ut there who se patrimo ny I ho pe never beco mes widely kno wn. O ne o f the last tasks yo u must perfo rm befo re releasing the game is the preparatio n o f a game manual. Manuals are frequently given sho rt shrift by just abo ut everybo dy asso ciated with co m- puter games. This is a serio us mistake, fo r the manual is a vital element in the o verall game pack- age. A co mputer has many limitatio ns; so me can be o verco me with a go o d manual. Much o f the static info rmatio n asso ciated with a game can be presented in a manual. The manual is also an excellent place to add fantasy suppo rt elements such as pictures and backgro und sto ries. Finally, a well-written manual will clear up many o f the misunderstandings that o ften arise during a game. Yo u must write yo ur o wn manual fo r the game, no matter ho w po o r a writer yo u are, and even if a pro fessio nal writer will prepare the final manual. The attempt to write yo ur o wn manual will increase yo ur respect fo r the skills o f the pro fessio nal writer, making it mo re likely that yo u will have a pro ductive relatio nship with the writer. Writing yo ur o wn manual will also pro vide feed- back o n the cleanliness o f the game design. Clumsy designs are hard to describe, while clean designs are easier to describe. Finally, yo ur o wn manual will be a useful so urce do cument fo r the pro fessio nal writer. Yo u sho uld be prepared fo r the writer to thro w o ut yo ur manual and start all o ver---a go o d writer wo uld rather create a new manual than po lish an amateur's crude effo rts. Yo u must cater to the writer’s needs, answering all his questio ns as co mpletely as po ssible. O nly a clo se and suppo rtive relatio nship between designer and writer can pro duce an excellent manual.
O nce the pro gram is o ut, brace yo urself fo r the critics. They will get their filthy hands o n yo ur lo vely game and do the mo st terrible things to it. They will play it witho ut reading the rules. they will find it intellectually deficient. They will divine imaginary technical flaws and speculate inco rrectly o n yo ur deep psycho lo gical hang-ups that led yo u to pro duce such a game. O ne critic o f mine co ncluded that TANKTICS was o bvio usly slapped to gether o n a rush schedule; actually, the time between first effo rts and final publicatio n was five years and two mo nths. Ano ther ro ast- ed ENERGY CZAR ( an energy eco no mics educatio nal simulatio n) because it wasn’t as exciting as his favo rite arcade game. Do n’t let these critics affect yo u. Mo st critics are far less qualified to crit- icize pro grams than yo u are to write them. A very few critics with the larger publicatio ns are quite tho ughtful; yo u sho uld pay attentio n to their co mments. With mo st critics, tho ugh, yo u sho uld pay heed o nly to views shared by three o r mo re independent critics. Remember also that even a go o d critic will ro ast yo u if yo ur go al is no t to his taste.
The public is ano ther matter. If they do n’t buy yo ur game, yo u lo se two ways: first, yo u o r yo ur emplo yer make little mo ney o n the game; and seco nd, yo u do n’t reach as many peo ple with yo ur message. It do esn’t matter ho w beautiful yo ur message is-if no bo dy listens to it, yo u have failed as an artist. O ne failure is no thing to wo rry abo ut; every artist bo mbs o ccasio nally. Two failures in a ro w are bad; three sho uld initiate a serio us reco nsideratio n o f artistic values. Are yo u willing to be a no ble and starving artist, o r a so mewhat wealthier artisan? Lo o k within yo ur heart, lo ng and hard. If deep do wn inside yo u kno w that yo u met yo ur go als, then igno re the critics and the public. C H A P T E R S I X D e s ig n Te c h n iq u e s a n d I d e a ls
very artist develo ps her o wn special techniques and ideals fo r the executio n o f her art. The painter wo rries abo ut brush stro kes, mixing o f paint, and texture; the musical co mpo ser learns techniques o f o rchestratio n, timing, and co unterpo int. The game designer also
acquires a variety o f specialized skills, techniques, and ideals fo r the executio n o f her craft. In this chapter I will describe so me o f the techniques that I use.
BALANCING SOLITAIRE GAMES
A so litaire game pits the human player against the co mputer. The co mputer and the human are very different creatures; where human tho ught pro cesses are diffuse, asso ciative, and integrated, the machine’s tho ught pro cesses are direct, linear, and arithmetic. This creates a pro blem. A co m- puter game is created fo r the benefit o f the human, and therefo re is cast in the intellectual terri- to ry o f the human, no t that o f the co mputer. This puts the co mputer at a natural disadvantage. Altho ugh the co mputer co uld easily whip the human in games invo lving co mputatio n, so rting, o r similar functio ns, such games wo uld be o f little interest to the human player. The co mputer must play o n the human’s ho me turf, so mething it do es with great difficulty. Ho w do we design the game to challenge the human? Fo ur techniques are available: vast reso urces, artificial smarts, limited info rmatio n, and pace.
This is by far the mo st heavily used technique fo r balancing a game. The co mputer is pro vided with immense reso urces that it uses stupidly. These reso urces may co nsist o f large numbers o f o ppo nents that o perate with a rudimentary intelligence. Many games use this plo y: SPACE
INVADERS, MISSILE CO MMAND, ASTERO IDS, CENTIPEDE, and TEMPEST are so me o f the mo re po pular games to use this technique. It is also po ssible to equip the co mputer with a small number o f o ppo nents that are themselves mo re po werful than the human player’s units, such as the supertanks in BATTLEZO NE. The effect in bo th cases is the same: the human player’s advan- tage in intelligence is o ffset by the co mputer’s material advantages. This appro ach has two benefits. First, it gives the co nflict between the human and the co mputer a David versus Go liath air. Mo st peo ple wo uld rather win as apparent underdo g than as equal. Seco nd, this appro ach is the easiest to implement. Pro viding artificial intelligence fo r the co m- puter’s players can be difficult, but repeating a pro cess fo r many co mputer players takes little mo re than a simple lo o p. O f co urse, the ease o f implementing this so lutio n carries a disadvantage: everybo dy else do es it. We are knee-deep in such games! Laziness and lack o f determinatio n have far mo re to do with the prevalence o f this technique than game design co nsideratio ns.
The o bvio us alternative to the use o f sheer numbers is to pro vide the co mputer player with intel- ligence adequate to meet the human o n equal terms. Unfo rtunately, artificial intelligence tech- niques are no t well eno ugh develo ped to be useful here. Tree-searching techniques have been develo ped far eno ugh to allo w us to pro duce passable chess, checkers, and O thello players. Any o ther game that can be expressed in direct tree-searching terms can be handled with these tech- niques. Unfo rtunately, very few games are appro priate fo r this treatment.
An alternative is to develo p ad-ho c artificial intelligence ro utines fo r each game. Since such ro u- tines are to o primitive to be referred to as "artificial intelligence", I instead use the less grandio se term "artificial smarts". This is the metho d I have used in TANKTICS, EASTERN FRO NT 1941, and LEGIO NNAIRE, with varying degrees o f success. This strategy demands great effo rt fro m the game designer, fo r such ad-ho c ro utines must be reaso nable yet unpredictable.
O ur first requirement o f any artificial smarts system is that it pro duce reaso nable behavio r. The co mputer sho uld no t drive its tanks o ver cliffs, crash spaceships into each o ther, o r pause to rest directly in fro nt o f the human’s guns. In o ther wo rds, o bvio usly stupid mo ves must no t be allo wed by any artificial smarts system. This requirement tempts us to list all po ssible stupid mo ves and write co de that tests fo r each such stupid mo ve and precludes it. This is the wro ng way to handle the pro blem, fo r the co mputer can demo nstrate unanticipated creativity in the stupid- ity o f its mistakes. A better ( but mo re difficult) metho d is to create a mo re general algo rithm that o bviates mo st absurd mo ves.
A seco nd requirement o f an artificial smarts ro utine is unpredictability. The human sho uld never be able to seco nd-guess the behavio r o f the co mputer, fo r this wo uld shatter the illusio n o f intel- ligence and make victo ry much easier. This is may seem to co ntradict the first requirement o f rea- so nable behavio r, fo r reaso nable behavio r fo llo ws patterns that sho uld be predictable. The appar- ent co ntradictio n can be reso lved thro ugh a deeper understanding o f the nature o f interactio n in a game. Three realizatio ns must be co mbined to arrive at this deeper understanding. First, reac- tio n to an o ppo nent is in so me ways a reflectio n o f that o ppo nent. A reaso nable player tries to anticipate his o ppo nent’s mo ves by assessing his o ppo nent’s perso nality. Seco nd, interactiveness is a mutual reactio n---bo th players attempt to anticipate each o ther’s mo ves. Third, this interac- tiveness is itself a measure o f "gaminess". We can co mbine these three realizatio ns in an analo gy. A game beco mes analo go us to two mirro rs aligned to wards each o ther, with each player lo o king o ut fro m o ne mirro r. A puzzle is analo go us to the two mirro rs being unreflective; the player sees a static, unrespo nsive image. A weakly interactive game is analo go us to the two mirro rs being weakly reflective; each player can see and interact at o ne o r two levels o f reflectio n. A perfectly interactive game ( the "gamiest game") is analo go us to the two mirro rs being perfectly reflective; each o f the two players recursively exchanges places in an endless tunnel o f reflected anticipa- tio n’s. No matter ho w reaso nable the behavio r, the infinitely co mplex pattern o f anticipatio n and co unter-anticipatio n defies predictio n. It is reaso nable yet unpredictable. puter is to anticipate human mo ves interactively, it must be able to assess the perso nality o f its o ppo nents---a ho peless task as yet. Fo r the mo ment, we must rely o n mo re primitive guidelines. Fo r example, my experience has been that algo rithms are mo st predictable when they are "partic- ular". By "particular" I mean that they place an emphasis o n single elements o f the o verall game pattern. Fo r example, in wargames, algo rithms alo ng the lines o f "determine the clo sest enemy unit and fire at it" are particular and yield predictable behavio r.
I have fo und that the best algo rithms co nsider the greatest amo unt o f info rmatio n in the bro ad- est co ntext. That is, they will facto r into their decisio n-making the largest number o f co nsidera- tio ns rather than fo cus o n a small number o f particular elements. To co ntinue with the example abo ve, a better algo rithm might be "determine the enemy unit po sing the greatest co mbinatio n o f threat and vulnerability ( based o n range, activity, facing, range to o ther friendly units, co ver, and sighting) ; fire o n unit if pro bability o f kill exceeds pro bability o f being killed".
Ho w do es o ne implement such principles into specific algo rithms? I do ubt that any all purpo se system. can ever be fo und. The best general so lutio n I have fo und so far fo r this pro blem utilizes po int systems, field analysis, and changes in the game structure. First, I establish a po int system fo r quantifying the merit o f each po ssible mo ve. This is a time- ho no red technique fo r many artificial intelligence systems. A great deal o f tho ught must go into the po int system. The first pro blem with it is o ne o f dynamic range: the designer must insure that the pro bability o f two accessible mo ves each accumulating a po int value equal to the maximum value allo wed by the wo rd size ( eight bits) appro aches zero . In o ther wo rds, we can’t have two mo ves each getting a sco re o f 255 o r we have no way o f kno wing which is truly the better mo ve. This pro blem will diminish as 16-bit systems beco me mo re co mmo n. A seco nd pro blem with the po int system is the balancing o f facto rs against each o ther. In o ur hypo thetical tank game used abo ve, we agree that climbing o n to p o f a hill is go o d, but we also agree that mo ving o nto a ro ad is go o d. Which is better? If a hillto p po sitio n is wo rth 15 po ints, what is a ro ad po sitio n wo rth? These questio ns are very difficult to answer. They require a deep familiarity with the play o f the game. Unfo rtunately, such familiarity is impo ssible to attain with a game that has yet to be co mpleted. The o nly alternative is bro ad experience, intimate kno wledge o f the situatio n being represented, painstaking analysis, and lo ts o f experimenting. A seco nd element o f my general appro ach to artificial smarts is the use o f field analysis. This is o nly applicable to games invo lving spatial relatio nships. In such games the human relies o n pat- tern reco gnitio n to analyze po sitio ns and plan mo ves. True pattern reco gnitio n o n the level o f human effo rt is beyo nd the abilities o f a micro co mputer. Ho wever, so mething appro aching pat- tern reco gnitio n can be attained thro ugh the use o f field analysis. The key effo rt here is the cre- atio n o f a calculable field quantity that co rrectly expresses the critical info rmatio n needed by the co mputer to make a reaso nable mo ve. Fo r example, in several o f my wargames I have made use o f safety and danger fields that tell a unit ho w much safety o r danger it faces. Danger is calculated are very dangero us and small distant units are o nly slightly dangero us. A similar calculatio n with friendly units yields a safety facto r. By co mparing the danger value at its po sitio n with the safety value at its po sitio n, a unit can decide whether it sho uld exhibit bo ld behavio r o r timid behavio r. O nce this decisio n is made, the unit can lo o k aro und it and measure the net danger minus safe- ty in each po sitio n into which the unit co uld mo ve. If it is feeling bo ld, it mo ves to wards the dan- ger; if it is feeling timid, it mo ves away. Thus, the use o f fields allo ws a unit to assess a spatial array o f facto rs.
Ano ther technique fo r co ping with artificial smarts pro blems is so simple that it seems like cheat- ing: change the game. If an element o f the game is no t tractable with artificial recko ning, remo ve it. If yo u can’t co me up with a go o d way to use a feature, yo u really have no cho ice but to delete it. Fo r example, while designing TANKTICS, I enco untered a pro blem with lakes. If a lake was co n- cave in shape, the co mputer wo uld drive its tanks to the sho re, back up, and return to the sho re. The co ncave lake created a trap fo r my artificial smarts algo rithm. I wasted a great deal o f time wo rking o n a smarter artificial smarts ro utine that wo uld no t be trapped by co ncave lakes while retaining desirable eco no mies o f mo tio n. After much wasted effo rt I disco vered the better so lu- tio n: delete co ncave lakes fro m the map.
Ideally, the experienced game designer has eno ugh intuitive feel fo r algo rithms that she can sense game facto rs that are intractable and avo id them during the design stages o f the game. Mo st o f us must disco ver these things the hard way and retrace o ur steps to mo dify the design. Experiencing these disasters is part o f what pro vides the intuitio n.
A special pro blem is the co o rdinatio n o f mo ves o f many different units under the co ntro l o f the co mputer. Ho w is the co mputer to assure that the different units mo ve in a co o rdinated way and that traffic jams do n’t develo p? O ne way is to use a sequential planning system co upled with a simple test fo r the po sitio n o f o ther units. Thus, unit #1 mo ves first, then #2, then #3, with each o ne avo iding co llisio ns. I can assure yo u fro m my o wn experience that this system replaces co lli- sio ns with the mo st frustrating traffic jams. A better way uses a virtual mo ve system in which each unit plans a virtual mo ve o n the basis o f the virtual po sitio ns o f all units. Here’s ho w it wo rks: we begin with an array o f real po sitio ns o f all co mputer units. We create an array o f virtual po sitio ns and initialize all virtual values to the real values. Then each unit plans its mo ve, avo iding co lli- sio ns with the virtual po sitio ns. When its mo ve is planned, it places its planned final po sitio n into the virtual array. O ther units then plan their mo ves. After all units have planned o ne virtual mo ve, the pro cess repeats, with each unit planning its mo ve o n the basis o f the interim virtual mo ve array. This huge o uter lo o p sho uld be co nvergent; after a sufficient number o f iteratio ns the ro u- tine terminates and the virtual po sitio ns fo rm the basis o f the mo ves made by the co mputer’s units. This technique sho uld be useful fo r co o rdinating the mo ves o f many units and preventing traffic jams.
No matter ho w go o d an algo rithm is, it has a limited regime o f applicability. The o dds are that a specific algo rithm will wo rk best under a narro w range o f co nditio ns. A go o d game design must ate a number o f algo rithms and switch fro m o ne to ano ther as co nditio ns change. The transitio n fro m o ne algo rithm to ano ther is fraught with peril, fo r co ntinuity must be maintained acro ss the transitio n. I well remember a frustrating experience with algo rithm transitio ns with LEGIO N- NAIRE. The co mputer-barbarians had three algo rithms: a "run fo r safety" algo rithm, an "appro ach to co ntact" algo rithm, and an "attack" algo rithm. Under certain co nditio ns a barbar- ian o perating under the "appro ach to co ntact" algo rithm wo uld decide o n bo ld behavio r, dash fo rward to make co ntact with the human, and make the transitio n to the "attack" algo rithm, which wo uld then declare an attack unsafe. The barbarian wo uld thus balk at the attack, and co n- vert to the "run fo r safety" algo rithm, which wo uld direct it to turn tail and run. The human play- er was treated to a spectacle o f fero cio usly charging and frantically retreating barbarians, no ne o f who m ever bo thered to actually fight. I eventually gave up and re-designed the algo rithms, merg- ing them into a single "advance to attack" algo rithm with no transitio ns.
The artificial smarts techniques I have described so far are designed fo r use in games invo lving spatial relatio nships. Many games are no n-spatial; o ther artificial smarts techniques are required fo r such games. O ne o f the mo st co mmo n types o f no n-spatial games uses systems that behave in co mplex ways. These games o ften use co upled differential equatio ns to mo del co mplex systems. LUNAR LANDER, HAMMURABI, ENERGY CZAR, and SCRAM are all examples o f such games. The primary pro blem facing the designer o f such games is no t so much to defeat the human as to mo del co mplex behavio r. I advise the game designer to be particularly careful with games invo lv- ing large systems o f co upled differential equatio ns. HAMMURABI uses three co upled first-o rder differential equatio ns, and mo st pro grammers find it tractable. But the co mplexity o f the pro b- lem rises very steeply with the number o f differential equatio ns used. ENERGY CZAR used the fantastic sum o f 48 differential equatio ns, a feat made believable o nly by the fact that many co n- straints were impo sed o n them. In general, be wary o f mo re than fo ur co upled differential equa- tio ns. If yo u must use many differential equatio ns, try to use parallel differential equatio ns, in which the same fundamental equatio n is applied to each element o f an array o f values. To help keep the system balanced, each differential equatio n sho uld have a damping facto r that must be empirically adjusted: new value = o ld value + ( driving facto r / damping facto r)
A small damping facto r pro duces lively systems that bo unce aro und wildly. A large damping fac- to r yields sluggish systems that change slo wly. Unfo rtunately, reco urse to simple damping facto rs can backfire when a relatio nship o f negative feedback exists between the "new value" and the "driving fo rce". In this case, large damping inhibits the negative feedback, and o ne o f the vari- ables go es wild. The behavio r o f systems o f differential equatio ns is co mplex; I suggest that designers interested in these pro blems study the mathematics o f o verdamped, underdamped, and critically damped o scillato ry systems. Fo r mo re general info rmatio n o n so lving systems o f differ- ential equatio ns, any go o d textbo o k o n numerical analysis will serve as a useful guide.
The applicatio n o f all o f these metho ds may well pro duce a game with so me intelligence, but o ne’s expectatio ns sho uld no t be to o high. Even the expenditure o f great effo rt is no t eno ugh to pro duce truly intelligent play; no ne o f my three effo rts to date play with an intelligence that is adequate, by itself, to tackle a human player. Indeed, they still need fo rce ratio s o f at least two to o ne to stand up to the human player.
Ano ther way to make up fo r the co mputer’s lack o f intelligence is to limit the amo unt o f info r- matio n available to the human player. If the human do es no t have the info rmatio n to pro cess, he canno t apply his superio r pro cessing po wer to the pro blem. This technique sho uld no t be applied to excess, fo r then the game is reduced to a game o f chance. It can, nevertheless, equalize the o dds. If the info rmatio n is withheld in a reaso nable co ntext ( e.g., the player must send o ut sco uts) , the restrictio ns o n info rmatio n seem natural.
Limited info rmatio n pro vides a bo nus: it can tickle the imaginatio n o f the player by suggesting witho ut actually co nfirming. This o nly happens when the limitatio ns o n the info rmatio n are art- fully cho sen. Rando mly assigned gaps in info rmatio n are co nfusing and frustrating rather than tantalizing. A naked wo man can be beautiful to the male eye, but an artfully dressed wo man can co nceal her charms suggestively and thus appear even mo re alluring. The same wo man rando m- ly co vered with miscellaneo us bits o f clo th wo uld o nly lo o k silly.
Ano ther way to even balance between human and co mputer is thro ugh the pace o f the game. The human may be smart, but the co mputer is much faster at perfo rming simple co mputatio ns. If the pace is fast eno ugh, the human will no t have eno ugh time to apply his superio r pro cessing skills, and will be befuddled. This is a very easy technique to apply, so it co mes as no surprise that it is very heavily used by designers o f skill and actio n games. I do no t enco urage the use o f pace as an equalizing agent in co mputer games. Pace o nly succeeds by depriving the human player o f the time he needs to invest a larger po rtio n o f himself into the game. Witho ut that investment, the game can never o ffer a rich challenge. Pace do es fo r co mput- er games what the o ne-night stand do es fo r ro mance. Like o ne-night stands, it will never go away. We certainly do no t need to enco urage it.
These fo ur techniques fo r balancing co mputer games are never used in iso latio n; every game uses so me co mbinatio n o f the fo ur. Mo st games rely primarily o n pace and quantity fo r balance, with very little intelligence o r limited info rmatio n. There is no reaso n why a game co uld no t use all amo unts o f each metho d, the game wo uld no t have to strain the limitatio ns o f each. The design- er must decide the appro priate balance o f each fo r the go als o f the particular game.
RELATIONSHIPS BETWEEN OPPONENTS
Every game establishes a relatio nship between o ppo nents that each player strives to explo it to maximum advantage. The fundamental architecture o f this relatio nship plays a central ro le in the game. It defines the interactio ns available to the players and sets the to ne o f the game. Mo st co m- puter games to date utilize very simple player-to -player relatio nships; this has limited their range and depth. A deeper understanding o f player-to -player relatio nships will lead to mo re interesting games.
The simplest architecture establishes a symmetric relatio nship between the two players. Bo th po s- sess the same pro perties, the same strengths and weaknesses. Symmetric games have the o bvi- o usly desirable feature that they are auto matically balanced. They tend to be much easier to pro - gram because the same pro cesses are applied to each player. Finally, they are easier to learn and understand. Examples o f symmetric games include CO MBAT fo r the ATARI 2600, BASKETBALL, and DO G DAZE by Gray Chang.
Symmetric games suffer fro m a variety o f weaknesses, the greatest o f which is their relative sim- plicity. Any strategy that pro mises to be truly effective can and will be used by bo th sides simul- taneo usly. In such a case, success is derived no t fro m planning but fro m executio n. Alternatively, success in the game turns o n very fine details; chess pro vides an example an advantage o f but a single pawn can be parlayed into a victo ry.
Because o f the weaknesses o f symmetric games, many games attempt to establish an asymmetric relatio nship between the o ppo nents. Each player has a unique co mbinatio n o f advantages and disadvantages. The game designer must so meho w balance the advantages so that bo th sides have the same likeliho o d o f victo ry, given equal levels o f skill. The simplest way o f do ing this is with plastic asymmetry. These games are fo rmally symmetric, but the players are allo wed to select ini- tial traits acco rding to so me set o f restrictio ns. Fo r example, in the Avalo n-Hill bo ardgame WIZ- ARD’S Q UEST, the players are each allo wed the same number o f territo ries at the beginning o f the game, but they cho o se their territo ries in sequence. Thus, what was initially a symmetric rela- tio nship ( each perso n has N territo ries) beco mes an asymmetric o ne ( player A has o ne co mbina- tio n o f N territo ries while player B has a different co mbinatio n) . The asymmetry is pro vided by the players themselves at the o utset o f the game, so if the results are unbalanced, the player has no o ne to blame but himself. games establish an asymmetric relatio nship between the co mputer player and the human player because the co mputer canno t ho pe to co mpete with the human in matters o f intelligence. Thus, the human player is given reso urces that allo w him to bring his superio r planning po wer to bear, and the co mputer gets reso urces that co mpensate fo r its lack o f intelligence.
The advantage o f asymmetric games lies in the ability to build no ntransitive o r triangular rela- tio nships into the game. Transitivity is a well-defined mathematical pro perty. In the co ntext o f games it is best illustrated with the ro ck-scisso rs-paper game. Two players play this game; each secretly selects o ne o f the three pieces; they simultaneo usly anno unce and co mpare their cho ices. If bo th made the same cho ice the result is a draw and the game is repeated. If they make differ- ent cho ices, then ro ck breaks scisso rs, scisso rs cut paper, and paper enfo lds ro ck. This relatio n- ship, in which each co mpo nent can defeat o ne o ther and can be defeated by o ne o ther, is a no n- transitive relatio nship; the fact that ro ck beats scisso rs and scisso rs beat paper do es no t mean that ro ck beats paper. No tice that this particular no ntransitive relatio nship o nly pro duces clean results with three co mpo nents. This is because each co mpo nent o nly relates to two o ther co mpo nents; it beats o ne and lo ses to the o ther. A ro ck-scisso rs-paper game with binary o utco mes ( win o r lo se) canno t be made with mo re than three co mpo nents. O ne co uld be made with multiple co mpo - nents if several levels o f victo ry ( using a po int system, perhaps) were admitted. No ntransitivity is an interesting mathematical pro perty, but it do es no t yield rich games so lo ng as we hew to the strict mathematical meaning o f the term. The value o f this discussio n lies in the generalizatio n o f the principle into less well-defined areas. I use the term "triangular" to describe such asymmetric relatio nships that extend the co ncepts o f no ntransitivity beyo nd its fo rmal def- initio n. A simple example o f a triangular relatio nship appears in the game BATTLEZO NE. When a saucer appears, the player can pursue the saucer instead o f an enemy tank. In such a case, there are three co mpo nents: player, saucer, and enemy tank. The player pursues the saucer ( side o ne) and allo ws the enemy tank to pursue him unmo lested ( side two ) . The third side o f the triangle ( saucer to enemy tank) is no t directly meaningful to the human---the co mputer maneuvers the saucer to entice the human into a po o r po sitio n. This example is easy to understand because the triangu- larity assumes a spatial fo rm as well as a structural o ne. Triangularity is mo st o ften implemented with mixed o ffensive-defensive relatio nships. In mo st co nflict games, regardless o f the medium o f co nflict, there will be o ffensive actio ns and defensive o nes. So me games co ncentrate the bulk o f o ne activity o n o ne side, making o ne side the attacker and the o ther side the defender. This is a risky business, fo r it restricts the o ptio ns available to each player. It’s hard to interact when yo ur o ptio ns are limited. Much mo re entertaining are games that defend. What is mo re impo rtant, players can trade o ff defensive needs against o ffensive o ppo rtu- nities. Triangular relatio nships auto matically spring fro m such situatio ns. The essence o f the value o f triangularity lies in its indirectio n. A binary relatio nship makes direct co nflict unavo idable; the antago nists must appro ach and attack each o ther thro ugh direct means. These direct appro aches are o bvio us and expected; fo r this reaso n such games o ften degenerate into tedio us exercises fo llo wing a narro w script. A triangular relatio nship allo ws each player indirect metho ds o f appro ach. Such an indirect appro ach always allo ws a far richer and subtler interactio n.
Actors and Indirect Relationships
Indirectio n is the essence o f the value o f triangularity to game design. Indirectio n is itself an impo rtant element to co nsider, fo r triangularity is o nly the mo st rudimentary expressio n o f indi- rectio n. We can take the co ncept o f indirectio n further than triangularity. Mo st games pro vide a direct relatio nship between o ppo nents, as sho wn in the fo llo wing diagram: Since the o ppo nent is the o nly o bstacle facing the player, the simplest and mo st o bvio us reso lu- tio n o f the co nflict is to destro y the o ppo nent. This is why so many o f these direct games are so vio lent. Triangularity, o n the o ther hand, pro vides so me indirectio n in the relatio nship: With triangularity, each o ppo nent can get at the o ther thro ugh the third party. The third party can be a passive agent, a weakly active o ne, o r a full-fledged player. Ho wever, it’s to ugh eno ugh get- ting two peo ple to gether fo r a game, much less three; therefo re the third agent is o ften played by a co mputer-generated acto r. An acto r, as defined here, is no t the same as an o ppo nent. An acto r fo llo ws a simple script; it has no guiding intelligence o r purpo se o f its o wn. Fo r example, the saucer in BATTLEZO NE is an acto r. Its script calls fo r it to drift aro und the battlefield witho ut actively participating in the battle. Its functio n is distractio n, a very weak ro le fo r an acto r to play. The acto r co ncept allo ws us to understand a higher level o f indirectio n, diagrammatically repre- sented as fo llo ws: In this arrangement, the players do no t battle each o ther directly; they co ntro l acto rs who engage in direct co nflict. A go o d example o f this scheme is sho wn in the game RO BOTWAR by Muse So ftware. In this game, each player co ntro ls a killer ro bo t. The player writes a detailed script ( a sho rt pro gram) fo r his ro bo t; this script will be used by the ro bo t in a gladiato rial co ntest. The game thus remo ves the players fro m direct co nflict and substitutes ro bo t-acto rs as co mbatants. Each player is clearly identified with his o wn ro bo t. This fo rm o f indirectio n is unsuccessful because the co nflict itself remains direct; mo reo ver, the player is remo ved fro m the co nflict and fo rced to sit o n the sidelines. I therefo re see this fo rm o f indirectio n as an unsuccessful transi- tio nal stage.
The next level o f indirectio n is sho wn in a very clever bo ardgame design by Jim Dunnigan, BAT- TLE FO R GERMANY. This game co ncerns the invasio n o f Germany in 1945. This was o bvio usly
Anglo -Americans in the west. Uneven struggles make frustrating games. Dunnigan’s so lutio n was to split bo th sides. O ne player co ntro ls the Russians and the west-fro nt Germans; the o ther co n- tro ls the Anglo -Americans and the east-fro nt Germans. Thus, each player is bo th invader and defender: Neither player identifies directly with the invaders o r the Germans; the two co mbatants have lo st their identities and are no w acto rs. The highest expressio n o f indirectio n I have seen is Dunnigan’s RUSSIAN CIVIL WAR game. This bo ardgame co vers the civil war between the Reds and the Whites. Dunnigan’s brilliant appro ach was to co mpletely disso lve any identificatio n between player and co mbatant. Each player receives so me Red armies and so me White armies. During the co urse o f the game, the player uses his Red armies to attack and destro y o ther players’ White armies. He uses his White armies to attack and destro y o ther players’ Red armies. The end o f the game co mes when o ne side, Red o r White, is annihilated. The winner is then the player mo st identifiable with the victo rio us army ( i.e., with the largest pile o f lo ser’s bo dies and the smallest pile o f winner’s bo dies) .
The indirectio n o f this game is truly impressive. The two co mbatants are in no way identifiable with any individual until very late in the game. They are acto rs; Red and White battle witho ut human manifestatio n even tho ugh they are played by human players. There is o nly o ne limita- tio n to this design: the system requires mo re than two players to wo rk effectively. Nevertheless, such highly indirect player-to -player architectures pro vide many fascinating o ppo rtunities fo r game design. Direct player-to -player relatio nships can o nly be applied to direct co nflicts such as war. Direct co nflicts tend to be vio lent and destructive; fo r this reaso n, so ciety disco urages direct co nflicts. Yet co nflict remains in o ur lives, taking mo re subtle and indirect fo rms. We fight o ur real-wo rld battles with smiles, distant allies, pressure, and co -o peratio n. Games with direct play- er-to -player relatio nships canno t ho pe to address real human interactio n. O nly indirect games o ffer any po ssibility o f designing games that successfully explo re the human co nditio n.
SMOOTH LEARNING CURVES
As a player wo rks with a game, s/ he sho uld sho w steady and smo o th impro vement. Beginners sho uld be able to make so me pro gress, intermediate peo ple sho uld get intermediate sco res, and experienced players sho uld go t high sco res. If we were to make a graph o f a typical player’s sco re as a functio n o f time spent with the game, that graph sho uld sho w a curve slo ping smo o thly and steadily upward. This is the mo st desirable case. A variety o f o ther learning curves can arise; they reveal a great deal abo ut the game. If a game has a curve that is relatively flat, we say that the game is hard to learn. If the curve is steep, we say the game is easy to learn. If the curve has a sharp jump in it, we say that there is just o ne trick to the game, mastery o f which guarantees co mplete mastery o f the game. If the game has many sharp jumps, we say that there are many tricks. A particularly bad case arises when the player’s sco re falls tradicto ry elements that co nfuse o r distract the player at a certain level o f pro ficiency. The ideal always slo pes upward smo o thly and steadily. Games witho ut smo o th learning curves frustrate players by failing to pro vide them with reaso n- able o ppo rtunities fo r bettering their sco res. Players feel that the game is either to o hard, to o easy, o r simply arbitrary. Games with smo o th learning curves challenge their players at all levels and enco urage co ntinued play by o ffering the pro spect o f new disco veries. A smo o th learning curve is wo rked into a game by pro viding a smo o th pro gressio n fro m the beginner’s level to an expert level. This requires that the game designer create no t o ne game but a series o f related games. Each game must be intrinsically interesting and challenging to the level o f player fo r which it is targeted. Ideally, the pro gressio n is auto matic; the player starts at the beginner’s level and the advanced features are bro ught in as the co mputer reco gnizes pro ficient play. Mo re co mmo nly, the player must declare the level at which he desires to play.
THE ILLUSION OF WINNABILITY
Ano ther impo rtant trait o f any game is the illusio n o f winnability. If a game is to pro vide a co n- tinuing challenge to the player, it must also pro vide a co ntinuing mo tivatio n to play. It must appear to be winnable to all players, the beginner and the expert. Yet, it must never be truly winnable o r it will lo se its appeal. This illusio n is very difficult to maintain. So me games main- tain it fo r the expert but never achieve it fo r the beginner; these games intimidate all but the mo st determined players. TEMPEST, fo r example, intimidates many players because it appears to be unwinnable. The mo st successful game in this respect is PAC-MAN, which appears winnable to mo st players, yet is never quite winnable.
The mo st impo rtant facto r in the creatio n o f the illusio n o f winnability is the cleanliness o f the game. A dirty game intimidates its beginners with an excess o f details. The beginner never o ver- co mes the inhibiting suspicio n that so mewhere in the game lurks a "go tcha". By co ntrast, a clean game enco urages all players to experiment with the game as it appears.
Ano ther key facto r in maintaining the illusio n o f winnability arises fro m a careful analysis o f the so urce o f player failure. In every game the player is expected to fail o ften. What trips up the play- er? If the player believes that his failure arises fro m so me flaw in the game o r its co ntro ls, he beco mes frustrated and angry with what he rightly judges to be an unfair and unwinnable situa- tio n. If the player believes that his failure arises fro m his o wn limitatio ns, but judges that the game expects o r requires superhuman perfo rmance, the player again rejects the game as unfair and unwinnable. But if the player believes failures to be attributable to co rrectable erro rs o n his o wn part, he believes the game to be winnable and plays o n in an effo rt to master the game. When the player falls, he sho uld slap himself gently and say, "That was a silly mistake! " SUMMARY
In this chapter I have described a number o f design metho ds and ideals that I have used in devel- o ping several games. Metho ds and ideals sho uld no t be used in grab bag fashio n, fo r taken to gether they co nstitute the elusive element we call "technique". Technique is part o f an artist’s signature, as impo rtant as theme. When we listen to Beetho ven’s majestic Fifth Sympho ny, o r the rapturo us Sixth, o r the ecstatic Ninth, we reco gnize in all the identifying stamp o f Beetho ven’s masterful technique. If yo u wo uld be a co mpute game designer, yo u must establish and develo p yo ur o wn technique. C H A P T E R S E V E N T h e F u t u r e o f C o m p u t e r G a m e s
n this bo o k, I have explo red co mputer games fro m a number o f angles. I have presented my claim that co mputer games co nstitute an as-yet untapped art fo rm. Implicit in this claim is the ho pe that this art fo rm will so meday be tapped. Unfo rtunately, histo ry bears o ut the fears
o f cynics mo re o ften than the ho pes o f dreamers. I must therefo re separate ho pes fro m predic- tio ns. Where are co mputer games go ing? Ho w will they change in the years to co me? Will we see them emerge as a true art fo rm? There are a number o f divergent trends apparent no w; analysis o f them is co mplicated by co nflicting interpretatio ns o f the current state o f co mputer game design. I shall begin by addressing the mo st co mmo nly cited arguments, and pro ceed to the framewo rk I prefer.
FAD OR FIXTURE?
The first and mo st impo rtant questio n co ncerns the very survival o f the co mputer games indus- try. O ne scho o l o f tho ught maintains that co mputer games are merely a fad, a tempo rary infatu- atio n that will quickly pass when their no velty value is exhausted. Pro po nents o f this view co m- pare the co mputer game to o ther fads that swept into so ciety with equal fo rce. They maintain that co mputer games lack sufficient fundamental appeal to insure any staying po wer. Eventually, these peo ple say, co mputer games will go the way o f the hula ho o p.
This line o f tho ught is breezily rejected by all members o f the industry, but I fear that the co nfi- dence peo ple express is little mo re than the Titanic syndro me---the co nfidence that arises fro m mere size. They tend to blindly extrapo late into the future the asto unding gro wth rates we have experienced in the past. It is certainly hard to give credence to do o msayers when the curve o f gro wth slo pes upward so steeply. Ho wever, few industry o ptimists can pro vide justificatio n fo r their extrapo latio ns. Just because the industry do ubled in 1982 do es no t mean that it will do uble in 1983 o r 1984. Indeed, it canno t co ntinue to annually do uble much lo nger; if it did, o nly eleven years’ time wo uld be needed fo r Atari alo ne to engulf the entire Gro ss Natio nal Pro duct like so me mo nstro us PAC-MAN. Furthermo re, size alo ne generates negative fo rces that will certainly reduce the gro wth rate. In the simple days o f the seventies, when co mputer games were co unted by the tho usands rather than the millio ns, no bo dy much cared abo ut their effects because they were a mino r co mpo nent o f o ur so ciety. But no w, they are everywhere. They are such a po werful fo rce that they are affecting so ci- ety in such a way as to generate negative feedback. We no w have a backlash develo ping against co mputer games, with o rdinances against arcades po pping up all o ver the co untry. Parents are beginning to restrict their children’s access to the games. Edito rialists warn against the dire effects o f playing the games. Already several preliminary studies have been undertaken to determine the able, but the day will certainly co me when the crap game we call research co mes up snakeyes, and a blo ckbuster repo rt is issued demo nstrating that co mputer games cause cancer in labo rato ry rats. Bigger critters than Atari have bitten the dust; bigger industries than o urs have shriveled and died. Size and past success are no guarantee o f permanence. We need substantive reaso ns fo r co nfi- dence in the future rather than simple extrapo latio ns o f past histo ry. I am co nvinced that sub- stantive reaso ns fo r o ptimism exist; the full presentatio n o f my reaso ning will co me later in this chapter. Fo r no w let me say that co mputer games satisfy a fundamental desire fo r active recre- atio n, and as such are assured o f a bright future.
THE TECHNOLOGICAL EXTRAPOLATION The mo st co mmo nly cited future fo r co mputer games is the techno lo gical extrapo latio n.
Adherents o f this scho o l po int to the undeniably steady march o f techno lo gy and the rapid impro vements that we have seen in the hardware fo r delivering games. They then extrapo late these trends directly to pro ject a future po pulated by superco mputers with fabulo us games cho ck- full o f unbelievable graphics and incredibly realistic experiences. These peo ple emphasize tech- no lo gical facto rs as the primary agents o f change. They claim that the big breakthro ughs will co me with the use o f bigger and faster pro cesso rs, megabytes o f RAM, new languages, and better display hardware. Ho lo graphy, trackballs, laserdisks, bo dy senso rs-these are the co in o f the realm amo ng the techno lo gical extrapo lato rs.
I cast a jaded eye o n such predictio ns. This is the same line o f tho ught that extrapo lated co mput- er develo pment in the late 60’s to predict ever-larger, ever-faster mainframes as the primary avenues o f develo pment in the co mputer industry fo r the 70’s. Co mputers did indeed beco me larger in that decade, but the develo pment o f larger co mputers was no t the do minant event o f the 70’s. Instead, the maturatio n o f minico mputers and the genesis o f micro co mputers were the majo r develo pments o f the 70’s. The extrapo lato rs never fo resaw the co ming o f micro co mputers, because micro s didn’t fit into their "bigger and better" extrapo latio ns. I do no t deny that techno lo gy will impro ve; it will. The real issue is no t whether o r no t techno l- o gy will impro ve, but whether o r no t techno lo gical limitatio ns are the primary co nstraints o n the game designer. I do no t deny that techno lo gical limitatio ns do impo se severe co nstraints o n all co mputer games, and I readily ackno wledge that techno lo gical advances will remo ve many o f these co nstraints. Thus, techno lo gical immaturity, the weakness o f current 8-bit, 64K, 1 MHz sys- tems---is a crippling limitatio n. Yet I maintain that artistic immaturity is an even mo re crippling limitatio n. Co nsider two extreme hypo thetical future wo rlds. The first wo rld has no techno lo gical develo p- ment and the seco nd wo rld has no artistic develo pment. In the first wo rld I am stuck with an Atari 800 as my so le medium fo r game design. This do es no t wo rry me to o much; I co uld explo re the po ssibilities o f this machine fo r five o r ten years befo re beginning to feel trapped. The seco nd
RAIDERS and BREAKO UT, with mo re co lo rful explo sio ns, snazzier so unds, and 3-D pho to n to r- pedo es, but never anything new o r different. I wo uld feel trapped immediately. Neither o f these wo rlds will happen; we will have bo th techno lo gical develo pment and artistic develo pment. Yet, we must remember that the techno lo gical develo pment, while entirely desir- able, will never be the driving fo rce, the engine o f change fo r co mputer games. Artistic maturatio n will be the dynamo that drives the co mputer games industry. The relative impo rtance o f techno lo gical develo pment and artistic maturity is made clear by a co mpariso n o f mo dern mo vies with the silent mo vies. The mo dern mo vies bo ast gigantic tech- no lo gical advantages---so und, co lo r, and fabulo us special effects. When used with skill and artistry, the new techno lo gies are indeed magnificent. Yet, all these advantages canno t make up fo r a lack o f artistic quality: the co mputer-graphics blo ckbuster TRO N co mpares po o rly with any o f Charlie Chaplin’s mo vies. if Chaplin co uld do so much with black and white film and no so und, why canno t we do go o d wo rk with 8 bits and 48K?
ASSESSMENT: TECHNOLOGICAL REVOLUTION
To explain my o wn assessment, I must present so me backgro und abo ut ho w I view techno lo gical revo lutio ns. The first great techno lo gical revo lutio n I will draw o n is the revo lutio n in trans- po rtatio n that swept American so ciety in the first half o f the twentieth century. The auto mo bile was invented in the late 1800’s; by the turn o f the century it was available as a co nsumer pro duct. Ho wever, many pro blems plagued the auto mo bile. It was expensive and unreliable. It lacked the so ftware ( suppo rt services such as service statio ns and appro priate ro ads) to make it truly practi- cal. It required co nsiderable skill and dedicatio n to o perate. Furthermo re, it was unnecessary; American culture had develo ped quite successfully witho ut it, so there was little existing need fo r it. Thus, the auto mo bile was no t a practical to o l; it was a plaything o f the wealthy.
With the passage o f time, these pro blems with the auto mo bile lessened in severity. Mass pro duc- tio n lo wered the co st and increased the reliability; mo re service statio ns and better ro ads became available. Mo re and mo re auto mo biles were purchased; by the late twenties the auto mo bile was a co mmo n fixture o f American life.
The third stage became o bvio us in the 1950’s. The auto mo bile changed the face o f American so ci- ety. Ho using patterns began to change. Co mmuting became practical. Urban sprawl sprawl. Drive-in restaurants and theaters became co mmo n. The techno lo gy changed the so ciety. The fo urth stage began asserting itself at abo ut the same time. As the auto mo bile changed American so ciety, so to o did so ciety change the auto mo bile. O riginally designed as a device to transpo rt peo ple and pro perty fro m po int A to po int B as quickly, safely, and reliably as po ssible, it was transfo rmed into a fo rm o f self-expressio n, a recreatio nal device, and ultimately an end in itself. Co uld Henry Fo rd have anticipated dune buggies, vans with waterbeds, lo w-riders, and naked-lady ho o d o rnaments? I do ubt it. no lo gy was initially desirable to o nly a small part o f the public. With time, co nditio ns impro ved and the techno lo gy co nquered so ciety. Then it began to change so ciety. In the pro cess, so ciety began to change the techno lo gy. The directio n o f this change was away fro m the pragmatic and to wards the recreatio nal.
Let us no w examine the seco nd great revo lutio n o f this century, the entertainment revo lutio n sparked by the televisio n. When televisio n first became available in the late 1940’s, it was expen- sive, unreliable, and lacking sufficient so ftware ( pro grams) to make it anything mo re than a to y fo r the wealthy. With time, these pro blems were o verco me. Televisio ns became cheaper, mo re reli- able, and o ffered mo re pro gramming. They swept into so ciety with great fo rce. In the pro cess, they dramatically changed the lifestyles o f the American peo ple. Nighttime entertainment was no w readily available. Leisure time activities changed acco rdingly. But the public wo rked its will o n tel- evisio n. It evo lved fro m "visible radio ", o r a means o f presenting lectures, plays, and speeches, into a medium with its o wn perso nality. Thus, the same fo ur stages o utlined fo r the auto mo bile o ccurred with televisio n: pio neer, co nquest, transfo rmatio n o f so ciety by the techno lo gy, and transfo rmatio n o f the techno lo gy by so ciety. The same sequence o f stages is o ccurring with co mputers. At the mo ment, perso nal co mputers are still expensive, unreliable, hard to use, and lacking so ftware. The situatio n is changing rapidly; prices are failing, machines are beco ming friendlier, and so ftware availability impro ves daily. All o bservers agree that perso nal co mputers will take so ciety by sto rm. The o nly differences o f o pin- io n are tho se o f magnitude. Will 1990 see 5 millio n co mputers in American ho mes, o r 10 mil- lio n, o r 20 millio n? No o ne kno ws, but everyo ne agrees that the figure will be large.
We therefo re expect that perso nal co mputers will change the face o f American so ciety. We expect that netwo rking will allo w mo re Americans to participate in eco no mic activities fro m the ho me, decreasing the lo ad o n transpo rtatio n and accelerating the pace o f eco no mic life. The ease o f manipulating info rmatio n will give info rmatio n an even mo re pro minent ro le in o ur so ciety. O ur financial system will beco me less dependent o n currency. O ur lives will be changed by these machines.
But we o urselves will no t be changed. The co mputer will change o ur habits and o ur leisure time, but it will no t change o ur perso nalities, fo r emo tio nally we are still the same peo ple who built the pyramids, fo ught the Crusades, and co lo nized the New Wo rld. O ur analysis o f the two previ- o us revo lutio ns leads us to expect that the relatio nship between so ciety and the co mputer will be o ne o f recipro cal transfo rmatio n. We further expect that the nature o f this transfo rmatio n will be a shift fro m the pragmatic to ward the recreatio nal, fro m the functio nal to the frivo lo us. This leads us to suspect games as the primary vehicle fo r so ciety to wo rk its will o n co mputers. Ten years ago , even five years ago , this suggestio n wo uld have seemed ridiculo us. Co mputers were the aweso me creatures o f man’s cleverness, the intelligent pro geny o f the machine age. They were with co mputers was whether they wo uld be man’s slave o r his master. The po ssibility that they might be his playmate never cro ssed anyo ne’s mind. We were wro ng, fo r the co mputer game has already established itself as a primary fo rm o f use o f the co mputer. By any number o f measures, co mputer games are already a majo r po rtio n o f the wo rld o f co mputers. Co nsider, fo r example, the number o f co mputer games in existence. What is the mo st repro duced pro gram in human histo ry, the o ne pro gram with mo re co pies in existence than any o ther pro gram in the wo rld? At the mo ment, the #1 pro gram is undo ubtedly CO MBAT, the game cartridge supplied with every ATARI 2600. Millio ns and millio ns o f co pies o f this car- tridge have been distributed. Perhaps yo u o bject that this measure is unfair because no bo dy buys the pro gram by itself. Very well, then, co nsider PAC-MAN, ASTERO IDS, SPACE INVADERS, and MISSILE CO MMAND, each o f which has so ld millio ns o f co pies. Indeed, were we to co mpile a "To p Fo rty" lit o f the best-selling pro grams o f all time, I very much do ubt that Visicalc ( trademark o f Visico rp) o r any serio us piece o f so ftware wo uld make the list. Games do minate. Perhaps yo u o bject that numbers alo ne do no t adequately measure so cial significance. Perhaps yo u wo uld pre- fer to measure eco no mic significance. Very well, let’s try a co mpariso n. Visicalc, the mo st success- ful perso nal co mputer serio us package, has so ld, say, 400,000 co pies at, say, $200 apiece. That amo unts to $80 millio n gro ss. By co ntrast, if Atari sells, say, 5 millio n co pies o f PAC-MAN at $30 apiece, that’s $150 millio n. And that’s just o ne title; there are many o ther games generating large sales figures. Thus, games are already a primary fo rm o f use o f co mputer techno lo gy. They have established themselves as a majo r co mpo nent in the wo rld o f co mputers. In the accelerated wo rld o f the 80’s, the fo urth stage ( transfo rmatio n o f techno lo gy by so ciety) is upo n us even as the seco nd phase ( co nquest) is beginning.
THE NATURE OF CHANGE
Games are the vehicle with which so ciety will change the co mputer. Ho w will the games them- selves be changed by so ciety? We can expect two pro cesses to affect games: the mass market and the flo wering o f hetero geneity. In so me ways, these pro cesses wo rk against each o ther.
The Mass Market
As co mputer games beco me a mass market item, they will fall prey to the ho mo genizing fo rces o f the mass market. The emphasis will no t be o n o riginality o r creativity, but rather o n adhering to the time-ho no red fo rmulas. Just as mo vies and televisio n fell prey to the fo rmulas o f sex and vio - lence, co ps and ro bbers, sitco ms, and the o ther mechanical incantatio ns o f the mass media, so to o will games fall victim to the tyranny o f the mass market. ( Are my biases sho wing?) We will are already caught in the grip o f this fo rce, fo r we are pro ducing little mo re than variatio ns o n a single theme: "blast the mo nsters! ". This has so ld well, so we stick with it. This cynical view o f the mass market is co untered by the realizatio n that the mass market is o cca- sio nally capable o f sustaining a real blo ckbuster. Ho llywo o d may grind o ut an army o f so ulless clo nes, but every no w and then so mething really interesting co mes o ut. When this happens, the mass market respo nds fabulo usly. 2001-A SPACE O DYSSEY, STAR WARS, and RAIDERS O F THE LO ST ARK are examples o f o riginal, creative ideas co ming o ut fo r the mass market and enjo ying success. Just because so mething wo rks in the mass market do es no t mean that it must be junk.
The Flowering of Heterogeneity
The games market differs fro m the mo vie market and the televisio n market in that it is less cen- tralized and has fewer eco no mies o f scale. In this respect it is clo ser to the bo o ks market and the reco rds market. Fo r this reaso n, I expect the games market to exhibit a greater degree o f hetero - geneity and less slavish o beisance to mass tastes.
I therefo re expect a ho st o f baby markets fo llo wing in the train o f the mass market. While the baby markets will never be as lucrative as the mass market, they perfo rm two valuable services. First, they pro vide a testing gro und fo r new ideas that, if successful, will be swallo wed up by the vo ra- cio us mass market. Beyo nd, the baby markets will always pro vide a haven fo r the refugees fro m medio crity and a playgro und fo r tho se who se tastes aren’t average. Yo u may ask why baby markets have no t yet develo ped very far to date. I answer the questio n with a little sto ry. Suppo se that yo u were the first astro naut to land o n a newly disco vered planet, and there yo u fo und a civilizatio n every bit the equal o f o urs, but fo r a single exceptio n: they had no literature. No no vels, no po etry, no children’s bo o ks, no textbo o ks, no magazines, no thing that we have, with o ne exceptio n: they did have co mic bo o ks. O n further study, yo u disco vered the rea- so n fo r this o ddity. Reading was a new disco very o nly recently po pularized by teenagers and shunned by the majo rity o f adults who felt intimidated by this newfangled skill. Thus, literature was used by teenagers to express the fantasies and interests they enjo yed: co nfro nting autho rity, vio lent reso lutio n o f co nflict and so fo rth. Hence co mic bo o ks. Co uld yo u no t lo o k o n this situ- atio n and reco gnize the seeds o f the future in it? Wo uld no t the flo wering o f o ther fo rms o f liter- ature be expected as the kids gro w up and develo p new interests? Wo uld no t no vels, sho rt sto ries, westerns, go thic ro mances, po etry, and o ther genres be incipient in the situatio n yo u fo und? So it is with co mputer games. Until no w the preserve o f teenage males, these games are bursting into so ciety at large. While they have satisfied until no w the fantasies o f twisted co mputer-nerd minds, they will so o n blo sso m into a much richer array o f fantasies. We will have co untry-western acco untant games, and sno b games. The so ciety that invented the ho t tub, CB radio , and dune buggies will have no reservatio ns abo ut impressing its character o n co mputer games. Eventually, games will be reco gnized as a serio us art fo rm. The explo ratio n o f games as a serio us art fo rm will be restricted to a tiny fractio n o f the to tal activity. Mo st o f the effo rt will always be mo re alo ng the lines o f po p-art. Yet this tiny gro up o f games-artists will be respo nsible fo r creat- ing the future classics o f games, the games that endure.
To co nclude: I see a future in which co mputer games are a majo r recreatio nal activity. I see a mass market o f co mputer games no t to o different fro m what we no w have, co mplete with blo ckbuster games, spin-o ff games, remake games, and tired co mplaints that co mputer games co nstitute a vast wasteland. I even have a term fo r such games---cyberschlo ck. I also see a much mo re exciting lit- erature o f co mputer games, reaching into almo st all spheres o f human fantasy. Co llectively, these baby market games will pro bably be mo re impo rtant as a so cial fo rce than the ho mo genized clo nes o f the mass market, but individual games in this arena will never have the eco no mic suc- cess o f the big time games.
By 1985 so ftware sto res will be as co mmo n as reco rd sto res; by 1990 they will be as co mmo n as bo o ksto res. O n entering the so ftware sto re, yo u will be co nfro nted by racks and racks o f games, with serio us so ftware o ccupying a smaller po rtio n o f the flo o rspace. Just as in a bo o ksto re o r reco rd sto re, yo u will see aisles devo ted to particular tastes in games. Yo u can bro wse thro ugh co l- lectio ns o f co wbo y games as yo ur co mpanio n explo res the latest space games. Perhaps yo u will lo o k fo r the latest pro duct o f yo ur favo rite autho r, all o f who se wo rks are co llected in alphabeti- cal o rder. O n the walls yo u will see po sters anno uncing the latest smash hit games by so ftware superstars. After evaluating a number o f games yo u will make yo ur cho ices and purchase them. Then yo u’ll go o ut to the parking lo t to disco ver that so me idio t has dented the fender o f yo ur car. So me things never change. C h a p t e r E ig h t T h e D e v e lo p m e n t o f E x c a lib u r
In Chapter 5, I presented an idealized game design sequence. I attempted to describe a general purpo se metho d that pro perly reco gnized the co ncepts develo ped thro ugho ut this bo o k. It is a sad truth that the practicality o f the schemes we devise is inversely pro po rtio nal to the idealism they embo dy. I have never designed a game in co mplete acco rdance with the system described in Chapter 5. My real designs have fo llo wed co nsiderably ro ckier co urses. In this chapter, I will describe the develo pment o f EXCALIBUR, a recent design. The co ntrast between the real pro cess, jerky and mistake-pro ne, and the ideal pro cess sho uld help the reader bridge the gap between the- o ry and practice.
BEGINNINGS In December o f 1981, I began wo rking fo r Alan Kay in his new Co rpo rate Research unit at Atari.
Given to tal creative freedo m, I reso lved to do a game wo rthy o f the vast faith that Dr. Kay had invested in me. I wanted this game to be grand and glo rio us, a game so lo fty in its go als and play that it wo uld put all o thers to shame. Since marketing co nsideratio ns were no t significant to the game, I reso lved that this game wo uld run in a 48K disk-based enviro nment. This affo rded me plenty o f co mputer reso urce with which to wo rk. My backgro und is in wargames, and I naturally tho ught in terms o f a wargame. War is the mo st extreme expressio n o f human co nflict, the greatest evil o f human existence, and the highest tragedy o f o ur species; it is therefo re an o bvio us starting po int fo r a serio us artist. I wanted to break away fro m the co nventio nal treatment o f war in wargames, which either glo rifies war as an expressio n o f misco nceived hero ism, o r trivializes war as a fascinating intellectual exercise. I want- ed so mething mo re than a wargame, so mething that placed war in a meaningful co ntext. My game wo uld include war as a viable o ptio n that must so metimes be exercised, but no t frivo lo us- ly. I wanted a game that warmo ngers wo uld inevitably lo se, because I deeply believe that peace- ful strategies are o ften the mo st practical o nes. This game wo uld address statecraft as a human enterprise; as such it wo uld necessarily fo cus o n leadership. Ano ther fundamental go al I estab- lished was that the game wo uld actually co nsist o f a number o f games linked to gether. This wo uld allo w me to sho w po licy, statecraft, and war at a variety o f scales, fro m the mo st strategic and indi- rect level to the mo st tactical and direct level. My next task was to determine the fantasy co ntext fo r the game. I bo iled the po ssibilities do wn to two co ntenders: a game dealing with the USA after a majo r nuclear war, and a game abo ut Britain in the Dark Ages after the co llapse o f Ro man autho rity. Bo th co ntexts deal with so cieties attempt- ing to reo rganize themselves after a calamity. I decided that the first fantasy was to o mo rbid fo r my purpo ses. Furthermo re, the seco nd fantasy co ntext was shro uded in the legends o f King Arthur, an intrinsically interesting subject. I therefo re cho se the Arthurian co ntext. peace to the tro ubled land. The challenge o f the game wo uld arise fro m the unwillingness o f the o ther kings to submit to Arthur's autho rity. The player wo uld be required to use a variety o f tech- niques to establish his autho rity, o nly o ne o f which wo uld be military actio n. Indeed, I reso lved that o veruse o f military metho ds wo uld brutalize the natio n and result in endless insurrectio ns and anarchy. With these no ble go als established, I began serio us design wo rk o n the game.
EARLY WORK: JANUARY-APRIL, 1982
I first turned to the questio n, what is leadership? The answer to this questio n is central to the game. It was essential fo r me to determine the essence o f leadership at the natio nal level and reduce this essence to a fo rm manageable in a game. I needed to extract the central decisio ns o f leadership and design a fo rm fo r expressing them. The military aspects o f leadership are the mo st o bvio us and easiest to wo rk with. I wo uld have had no difficulty designing a game in which the player must make all the co rrect military decisio ns. Yet, this was no t satisfacto ry to me: I wanted to address wider issues. My game had to address the so cial, diplo matic, and interperso nal aspects o f leadership. Ho w was I to represent and manipulate these facto rs in the co urse o f the game? These pro blems vexed me fo r mo nths. I quickly grew impatient with the struggle with such fundamental pro blems. The child in me wanted immediate gratificatio n. To satiate these impatient impulses, I wro te the title and ending scenes fo r the game. These were no t crucial to the structure o f the game, but they. gave me an o ppo rtunity to explo re so me interesting graphics techniques witho ut co mpro mising the integrity o f my design. The ending scene po sed so me interesting pro blems. It sho ws the swo rd Excalibur twirling thro ugh the air o ver a lake, falling into a hand that abruptly rises o ut o f the water to catch it, and then recedes beneath the waves. I spent a great deal o f time trying to add the lo nely so und o f the wind whistling against the blade o f the swo rd, but I was never able to o btain satisfacto ry results. I therefo re turned to the idea o f acco mpanying the title and ending scenes with so me appro priate music. I cho se as my two prime candidates a sectio n fro m Siegfried's death and funeral in Wagner's Siegfried, and a po rtio n o f Dvo rak's Seventh Sympho ny. I also determined the fundamental structure o f the game at this time. There were to be fo ur fun- damental nested games. The first, CAMELOT, wo uld co ncern Arthur's activities within his castle. These wo uld include the management o f his o wn kingdo m, the co nduct o f diplo macy, and the preparatio n o f the army. The seco nd game mo dule, BRITAIN, wo uld allo w Arthur to travel aro und the island o f Britain with his army and engage in strategic military activity. The third game mo d- ule, BATTLE, wo uld allo w Arthur to fight battles with enemy armies. If Arthur himself managed to enco unter an enemy king o n the battlefield, then he wo uld enter the fo urth mo dule, JO UST. This last mo dule was intended to be a simple skill-and-actio n game in which Arthur attempted to unho rse his o ppo nent. The game wo uld use a full first-perso n view o f an advancing ho rseman, lance leveled, with the who le scene bo uncing up and do wn with the gallo ping o f Arthur's o wn ho rse. I entertained m yself b y d evising clever grap hics algo rithm s that wo uld generate the JO UST game wo uld take o nly a few seco nds to play and wo uld no t pro vide much challenge. So I started o ver with a new idea: a swo rdfight game. The first pro blem I faced was, ho w can I sim- ulate the mo tio n o f a swo rd thro ugh jo ystick co mmands? I go t o ut a yardstick and spent ho urs in my living ro o m, swinging the yardstick, trying to divine so me so rt o f pattern to it that co uld be represented cleanly with a jo ystick. My difficulties aro se fro m the fact that the mo tio n o f a swo rd in a swo rdfight is a very co mplex mo tio n, and a jo ystick simply canno t adequately express all the intricacies o f such mo tio n. I eventually fo und a reaso nable system. The side-to -side mo tio n o f the jo ystick co ntro lled the angle o f attack o f the swo rd, fro m ho rizo ntal swing fro m the left, thro ugh a vertical swing o ver the player's head, to a ho rizo ntal swing fro m the right. Backward mo tio n o n the jo ystick swung the swo rd backwards in preparatio n fo r a stro ke; fo rward mo tio n o f the jo y- stick sent the swo rd fo rward in its stro ke.
This pro blem so lved, I began wo rk o n so me new graphics ro utines that wo uld sho w an o ppo sing swo rdsman in first-perso n graphics. This pro ved to be a very difficult task. I eventually gave up o n the swo rdfight game fo r much o f the same reaso ns that had led me to abando n the jo ust game. Besides, I didn't want Arthur to be able to hack his way to victo ry. If swo rdfights canno t assure success, what's the po int o f having them in the game? By no w it was March. I began wo rk o n the BRITAIN mo dule. This was a .scro lling map with a number o f embellishments thro wn in. I had earlier do ne .scro lling maps in EASTERN FRO NT 1941 and LEGIO NNAIRE, so the implementatio n o f this mo dule was easy fo r me. Since I had lo ts mo re memo ry fo r this game, I decided to splurge and make a gigantic scro lling map. I ended up with a 6K map o f Britain that is quite large. Slo wly the design was taking shape in my head, but a fundamental questio n remained unan- swered: was this to be a histo rical game o r a fictio nal game? That is, was this a game abo ut Britain in the sixth century AD o r was this a game abo ut King Arthur? I read every bo o k I co uld lay my hands o n abo ut bo th subjects. This research led me to co nclude that Britain in the sixth century was a chao tic and depressing place. The native Celts were defending their ho meland against invading Anglo -Saxo ns landing o n the eastern co ast o f the island. Fo r two centuries the Anglo - Saxo ns slo wly pushed the Celts westward. King Arthur was actually a Celtic general who led a brief co untero ffensive against the Anglo -Saxo ns, winning the battle o f Mo unt Bado n and halting the Anglo -Saxo n o ffensive fo r abo ut 50 years. But Arthur's success was o nly a brief respite; in the end, the Celts lo st. Thus, the histo rical reco rd do es no t suppo rt my needs fo r a so ciety struggling to reo rganize itself. Instead, the sto ry o f Britain in the Dark Ages is the sto ry o f o ne peo ple being relentlessly driven o ut by ano ther.
Yet, fro m the dreams o f the vanquished aro se the legend o f the co nquering King Arthur, a legend that passed thro ugh the ages and agreeably mo lded itself to suit the needs o f any sto ryteller. As I read the many incarnatio ns o f these legends, I was struck by their surpassing flexibility. Each artist gio us inspiratio n, ribald tales, o r expo sitio ns o f the chivalric ideal. Even Mark Twain turned them to go o d use fo r his characteristic blistering so cial co mment. A majo r turning po int in the design pro cess came when I watched the mo vie EXCALIBUR. This is a magnificent film that beautifully captures the best elements o f the Arthurian legends yet makes its o wn statement. I watched it o ver and o ver, reveling in the richness o f the tale. This mo vie shamed me with its excellence. I realized that I had been co mpro mising the impo rtant artistic issues in my game in o rder to play with cute graphics. I climbed a lo nely hill and spent a day med- itating. I rededicated myself to the lo fty artistic go als I had earlier set fo r myself. I also knew that I co uld no t realize them alo ne; I had to go t help. I enlisted the aid o f Larry Summers, and hired Valerie Atkinso n to help me. With new determinatio n, we set to wo rk.
THE LONG HAUL: MAY-DECEMBER 1982
Here is where we sto o d in May, 1982: I had established the bro ad design but had left many details unfinished. A number o f disparate chunks o f co de had been written, but they did no t fit to geth- er at all. There was no o verall design do cument. Faced with so many things to do , I fo o lishly o pted to finish so me o f the mo re o bvio us mino r things. I wro te the CALIG mo dule that draws Go thic characters o nto the screen. Valerie set to wo rk preparing the bit map tables fo r the ro utine. Larry wo rked o n finishing the title scene by adding the music and the disso lve ro utines. This wo rk, never intended as mo re than flashy windo w-dressing, unfo rtunately co nsumed nearly two mo nths. In June we began wo rk o n the CAMELOT mo dule, with Valerie taking primary pro gramming respo nsibility. This mo dule was actually a set o f illustrated menus. Each ro o m ( menu) had fo ur o ptio ns described by a single-wo rd entry. A vertical band allo wed the player to mo ve his cro wn- curso r to the menu selectio n. To the right o f the vertical band we placed a graphic windo w fo r sho wing so me critical bit o f info rmatio n. Fo r example, in the Ro und Table Ro o m, we sho wed a circle depicting the Ro und Table and a set o f shields representing the knights o f the Ro und Table. Their spatial po sitio ns in the ro o m indicated their so cial relatio nships. In the Treasury Ro o m we had intended to sho w piles o f co ins; we had to delete that feature later o n to sho w mo re detailed eco no mic data. We had also intended to use a kernelled display that wo uld have allo wed much mo re co lo r o n the screen; later o n we gave up o n that idea, fo r it wo uld have co nsumed to o much executio n time. As Valerie set to wo rk o n this sizable jo b, I began wo rking o n the so cial game asso ciated with the Ro und Table. I plunged into the task witho ut realizing the magnitude o f what I was attempting. I wanted to pro duce a small game that wo uld require Arthur to manage a so cial gro up. I quickly realized that the mo st interesting features o f such a situatio n lay no t the radial relatio nships ( the relatio nships between Arthur and the o ther knights) but in the circumferential relatio nships amo ng the knights. Altho ugh Arthur must perfo rce deal with knights radially, the circumferential relatio nships may well be the deciding facto rs. I fo und this system fascinating and wo rked inten-
I was so pleased with the algo rithms that I threw to gether a sho rt BASIC pro gram that turned them into a stand-alo ne game. This game seemed very pro mising to me; particularly impressive was my wife's reactio n. A wo man who takes a dim view o f silly games, she to o k an instant liking to this game. Surprised and gratified that I had finally pro duced so mething she co uld enjo y, I decided to pursue this new game, o riginally a study fo r EXCALIBUR, as a co mpletely new pro ject. Aric Wilmunder was hired to execute the design, called GO SSIP. In July we entered a lo ng and slo w perio d o f frustrating pro gress. I began devo ting a larger share o f my time to the writing o f this bo o k. O ther duties further distracted me. Witho ut my active daily participatio n, the pro ject began to flo under. Larry and Valerie plugged away at their wo rk, mak- ing the best o f a weak situatio n. Fo r mo nths they slo wly built o n the system we had created, flesh- ing o ut the skeletal system I had so briefly described. Since I had so little time to devo te to the pro ject, I did a great deal o f designing by the seat o f my pants. In o ur regular weekly meetings, they wo uld present me with the latest design flaw they had unco vered. Having no clear memo ries o f previo us decisio ns, I wo uld hack to gether an ad ho c so lutio n. My intuitio ns are fairly go o d, and many times I go t away with these deplo rable techniques. Ho wever, many o f my o n-the-fly decisio ns fell apart and wro ught havo c with the o verall design. Po o r Valerie put features into the CAMELOT mo dule, o nly to have have them stripped o ut, then later re-installed. O ur reco rds fo r this perio d indicate a great deal o f wasted effo rt. We had intended that the treas- ury ro o m in Camelo t wo uld be illustrated with piles o f co ins indicating quantities o f wealth. A great deal o f time was expended writing co in-drawing ro utines. In the end, we realized that we didn't have eno ugh screen space to sho w these piles o f co ins, so we had to use simple numbers drawn o nto the screen. Indeed, the list o f things we designed, pro grammed, and later dro pped is a revealing measure o f my o wn failure to plan ahead. The list includes declaratio ns o f war ( dro pped but later incarnated as "Attack") , alliances, sieges, demands fo r tribute, armies mo ving aro und in Britain, and a ho st o f mino r patches.
Six mo nths were co nsumed in this muddle. These six mo nths were no t a to tal lo ss; indeed, much pro gress was made: Larry co mpleted the eco no mics pro cessing, the BRITAIN mo dule, disk swap- ping o f mo dules, the presentatio n o f diplo matic news, and a number o f majo r co nso lidatio ns o f the ever-burgeo ning co de. Valerie to o k the CAMELOT mo dule much further, linking it to the new features and making it the largest and mo st co mplex mo dule in the entire game. Yet, all o f this co uld have been co mpleted in half the time had I been mo re o rganized and devo ted mo re ener- gy to the pro ject. By Christmas, everybo dy was tired o f the pro ject, demo ralized, and despairing that the pro ject wo uld ever be co mpleted. Tho se were dark days indeed.
RENEWED EFFORT (JANUARY - APRIL 1983)
In January 1983 I was able to return EXCALIBUR to its rightful place as my highest prio rity pro j- ect. I plunged into the pro ject with a co ld determinatio n to get this pro ject do ne and o ut the do o r. Go ne were the grand and lo fty feelings o f 1982, the misty-eyed visio n o f a truly grandio se game.
Larry and Valerie. Ruthlessly I slashed at the design, ripping o ut vaguely defined o r no n-essential sectio ns. The design discipline that I had so ught to escape by co nsuming vast co mputer reso urce was fo rced o n me by my inability to co mplete the pro ject. At ho me, I wo rked o n the artificial intelligence ro utines fo r the knights in the Ro und Table Ro o m. This to o k a few weeks. Then I tack- led the BATTLE scene. During February and March I wro te, debugged, and playtested this mo dule. I was po ssessed, driven to co mplete the game by my self-impo sed deadline o f April 1. My reco rds indicate that I averaged 300 bytes o f debugged co de per day. Industry averages are 75-100 bytes per day. Larry and Valerie were caught up in the frenzy. They wo rked furio usly o n integrating all the pieces o f the pro gram to gether and reso lving the myriad inco nsistencies thereby pro duced. Entire mo dules handling Merlin's ro o m, eco no mics, vassalage, tithes, and swapping co de were designed, co ded, and debugged.
Despite this, we failed to make o ur April 1 deadline. We mo ved it back to April 15. Even this became impo ssible to meet. Nevertheless we made April 15 an impo rtant milesto ne -- all co ding wo uld be co mpleted by this date. The first two weeks o f April were co nsumed in a wild o rgy o f effo rt. Meeting every day, so metimes fo r fo ur ho urs at a stretch, we hammered o ut what was undo ubtedly the to ughest part o f the design: the artificial intelligence algo rithms. I had reserved this task fo r last, fo r the AI ro utines must reflect every aspect o f the design. The design must therefo re be co mplete, and all variables co mpletely defined, befo re AI algo rithms can be designed. Mo reo ver, the creatio n o f the AI ro utines tends to freeze the design, since significant design changes after the AI is do ne can ruin the entire AI design.
The AI fo r EXCALIBUR is easily the mo st difficult I have ever attempted. It must co nsider the per- so nalities o f the different kings, eco no mic facto rs, military facto rs, and geo metric facto rs. The sys- tem we develo ped uses intermediate variables that express co ncepts such as the amo unt o f mili- tary prestige a king has, ho w much prestige he has an eco no mic manager, and ho w well-liked he is. Perso nality traits facto red into the algo rithms include ambitio n, stupidity, and defensiveness.
FINAL WORK (MAY - JUNE 1983)
We almo st succeeded in meeting o ur milesto ne o f having all co de co mpleted by April 15. The co de remaining was quite trivial. We all to o k a break fo r two weeks. In May we began final wo rk o n EXCALIBUR. Larry and Valerie began ro o ting o ut and eliminating all the bugs in the pro gram. As I write this, they are still wo rking o n the task. In June, we will begin tuning and po lishing the game. I wo uld like to spend mo re time po lishing this game, but it is lo ng o verdue. It will have been in develo pment fo r 18 mo nths, and will have co nsumed 3 pro grammer-years o f effo rt. In these days o f six-week develo pment times o f quicky games, EXCALIBUR may well be o ne o f the mo st sweated-o ver games ever do ne. It is certainly o ne o f the mo st ambitio us designs ever attempted. It may no t be successful, but if it fails, it will no t be fo r want o f effo rt.
Crawfo rd's 1998 no te: We shipped EXCALIBUR in July. I n t e r v ie w w it h C h r is C r a w f o r d , F if t e e n Ye a r s A f t e r E x c a lib u r a n d T h e A r t o f C o m p u t e r G a m e D e s ig n By Su e Peab o d y , Asst . Pro f. o f Hist o ry, Wash in g t o n St at e Un iversit y Van co u ver Date: Tue, 17 Jun 1997
Prof. Peabody asks:
I'm interested in what yo u think o f the changes that have o ccurred in the last decade since yo u wro te this -- what did yo u co rrectly anticipate? What was o bscured in yo ur crystal ball? Is there anything that yo u wo uld like to add to the piece no w that yo u co uldn't o r didn't when yo u o rig- inally wro te it? Gee, it's actually been fifteen years since I wro te that in 1982, so I can be even less humiliated by its erro rs. I will no t try to evaluate specific statements, but rather respo nd to the o verall to ne. I was pretty much o n the mark in guessing the appro ximate rate o f gro wth o f revenues in entertainment so ftware. The industry is indeed much bigger and better-funded than back in the early 80s. Where I was way o ff the mark was my o ptimism abo ut the bro adening o f the marketplace. I believed that by this time we'd be seeing a wide range o f entertainment so ftware addressing a wide range o f tastes. That has no t happened; co mputer games no w are co mpletely unchanged in terms o f their basic appeal. They are precisely the same fast-actio n sho o t-em-ups o r nerdy strategy games that we were dishing o ut 15 years ago . What became o f Excaliber? ( I gather that it was very successful.) Indeed no t. It came o ut just as Atari co llapsed and was lo st in the dust o f the disaster. Tho se few peo ple who saw it, tho ugh, seem to have been impressed. I kno w that mo st designers regard it as a mino r landmark in game design. Do yo u think that the co mputer game lends itself better to certain kinds o f histo ry? Abso lutely! And this is bo th its strength and its weakness. Every fo rm o f histo rical examinatio n has biases built into it. The stuff and substance o f histo ry -- do cuments -- has a built-in bias to wards big sho ts. We kno w lo ts o f details abo ut Charlemagne, but damn little abo ut the few mil- lio n peasants who lived under his rule. We kno w so me things abo ut the Bro nze Age better than the Iro n Age, because bro nze do esn't rust away. O f co urse, co mputer games aren't evidence, but they are a prism thro ugh which we can lo o k at the evidence, and they bias o ur view, to o . This bias can be a strength, especially when it fo rces us to take an o peratio nal view o f histo ry rather than a mytho lo gical view. By this I mean that histo - ry can be "wo ndro us sto ries" o r it can be "natural pro cesses." Thus, the mytho lo gical style wo uld tell us that Napo leo n wo n so many battles because he was a brilliant strategist -- ho o ray fo r Napo leo n! But we can also wargame o ut his battles, fo llo w what he actually did and why he did
Napo leo n was a lo t mo re ruthless than his o ppo nents in terms o f "living o ff the land" ( taking all the peasants' fo o d) . I suspect that the written wo rd is weaker fo r o peratio nal thinking than it is fo r mytho lo gical think- ing. Mo st written descriptio ns o f the Battle o f Midway lo ve to tell o f that dramatic mo ment when the Japanese admiral lo o ked up and saw the American dive bo mbers o verhead, and in that o ne instant, the battle was lo st. But ho w did they get to that juncture? Yes, written acco unts do man- age to co mmunicate the intricate sequence o f events that led to such a pro fo und reversal o f fo r- tune, but the written explanatio ns are either impo ssible to fo llo w o r have a mytho lo gical feel, as if this battle were so me grand Greek drama acted o ut in the Pacific O cean. When yo u actually play o ut the thing, yo u get a greater sense o f ho w micro sco pically lo gical pro cesses can lead to macro - sco pically asto unding results. A co mputer game, like any histo ry, can be used to emphasize so me aspect o f histo ry. Fo r exam- ple, I designed a game so me years back that I called Guns & Butter, in which I presented the the- sis that techno lo gical develo pment arises auto matically fro m eco no mic gro wth. Mo st histo ries o f techno lo gy have a "great man" flavo r to them, so I presented the alternative view that new tech- no lo gies arise auto matically as so o n as an eco no my is large eno ugh to utilize them. ( By the way, wo uld this be termed a Marxist view o f techno lo gical histo ry?) I wo n't claim that this thesis is nec- essarily co rrect, but it certainly o ffered a different view o f histo rical pro cesses. The tendency o f po lities to agglo merate at ever-larger levels came thro ugh quite clearly in the game. O bvio usly, there's plenty o f ro o m fo r abuse here, and the relative o pacity o f the designer's assumptio ns and biases ( co mpared with print) co uld make co mputer games a greater so urce o f mischief than enlightenment. Go ebbels was so frightening because he had a pretty go o d grip o n ho w to use mo dern media fo r pro paganda purpo ses. Right no w, we're all to o dumb to figure it o ut. So meday we'll have o ur interactive Go ebbels. O ne way to characterize the difference between the "thesis" o f a histo rical game and the "thesis" o f a bo o k o r article is that the game thesis can be written in present tense ( e.g. "the French Revo lutio n resulted fro m a go vernment fiscal crisis, an eco no mic emergency and a lessening o f mo narchical autho rity" ) whereas a co nventio nal textual thesis is in past tense. Interestingly, I was just clearing o ut so me o ld paper wargames fro m SPI days, and they all spo rt, acro ss the to p o f the bo x, the legend "The time is: 0600 ho urs, Thursday, May 21st, 1476" o r so me such. The sense o f being in the present is vital to simulatio n -- and o ne o f its mo st po werful attrac- tio ns. Isn't the who le idea o f histo ry to make the past accessible to the present? I'm caught up in the midst o f a so ftware deadline... Go tta go no w. Chris W E B L I N K S T H E A R T O F C O M P U T E R G A M E D E S I G N :C h r is C r a w f o r d E m a il
: ch riscraw fo rd @w ave.n et P r o f. S u e P e a b o d y, D e p a r t m e n t o f H is t o r y, W a s h in g t o n S t a t e U n iv e r s it y Va n c o u v e r : E m a il
: p eab o d y@van co u ver.w su .ed u e m a il
M a r io C r o t e a u : : kalid @sym p at ico .ca