Ga m e Ta x on om ie s: A H igh Le ve l Fr a m e w or k for Ga m e
An a lysis a n d D e sign
I n 1999, Doug Church proposed t he use of form al abst ract design t ools for gam e design [ 3] . Part of Church's suggest ion w as t o develop a com m on design vocabulary. I t 's ironic t hat while t he gam e design com m unit y has st art ed t o develop t hese m ore r igorous design principles for gam es, t here is m uch confusion even about t he m ost basic of quest ions, such as what a gam e is, com pared t o a st ory or a sim ulat ion. This confusion only increases w hen w e st art t o consider new and em erging form s like m obile gam es, locat ion- based gam es and pervasiv e gam es. I t 's obvious t hat w e need som e basic dist inct ions and definit ions at t he highest level, so t hat m ore det ailed m et hods can be sort ed int o t heir appropriat e areas of applicat ion.
Developing a basic language for describing different t ypes of gam es requires different dim ensions of
dist inct ions. That is, we need ort hogonal t axonom ies: not everyt hing falls int o a sim ple hierarchical syst em of cat egories and subcat egories. Ort hogonal t axonom ies allow design concerns t o be separat ed. So w e can, for exam ple, consider w het her a gam e is a real- t im e st rat egy gam e or a w arfare sim ulat ion, irrespect ively of whet her it is creat ed for PCs, m obile devices, or t echnologically support ed physical environm ent s. The gam eplay pat t erns for an RTS m ay apply irrespect ively of t he im plem ent at ion st rat egy. Or at least , w e can t hese t hree funct ional and form al aspect s t o differing degrees, depending upon t he part icular gam e or gam e genre. Bey ond t he boundaries of gam es play ed only via com put ers and consoles w e ident ify furt her classificat ion dim ensions, from virt ual t o physical gam ing, and from fict ional t o non- fict ional gam ing.
This t axonom y has been developed w it hin t he Zero Gam e St udio of t he I nt eract ive I nst it ut e in Sw eden [ 9] . We developed t he t axonom y aft er m any long design discussions, and have found t he result ing fram ework t o be very useful, saving t im e and get t ing us past som e very basic quest ions and confusions. I t is, of course, im possible t o precisely classify m any specific gam es, since t heir different aspect s m ay belong t o m ult iple or am biguous classificat ions. Nevert heless, t his schem e provides a heurist ic and pract ical t ool for clarifying m any design issues, saving t im e in proposal w rit ing and design m eet ings, and providing higher level cat egories for ident ifying where m ore det ailed design m et hods m ay be applied.
Ga m e s a n d Ga m e Pla y
Com put er gam es encom pass a vast range of int eract ive m edia product ions. I n t he broadest possible sense w e call all of t hese t hings "gam es" . However, t his is not necessarily useful in underst anding dist inct ions am ong t he different creat ions t hat w e're considering. I t 's m uch m ore useful t o adopt a narrow er definit ion of "gam e" . So let us narrow t he definit ion a lit t le and st at e: a gam e is a goal- direct ed and com pet it ive act ivit y conduct ed w it hin a fram ework of agreed rules. This can be referred t o as t he ludic or ludological definit ion of gam e, t he k ind of definit ion at t he base of t radit ional gam e t heory in disciplines like econom ics.
Giv en t his definit ion of a gam e, it is oft en said t hat learning t o play a gam e involv es learning t he rules of t he gam e. Not ice how ever t hat our definit ion does not require t his. I t does require t hat act ivit y obeys t he rules, and t hat we im plicit ly or explicit ly agree t o t hose rules.
The rules est ablish what as a player you can or cannot do, and what t he behavioral consequences of act ions m ay be wit hin t he world of t he gam e. But , successful play does not necessarily require learning all of t he gam e rules - - only t hose necessary t o support a part icular playing st yle. Learning t o play a gam e, m aking progress w it hin a gam e, and, wit h persist ence and basic abilit y, event ually com plet ing or w inning a gam e are a m at t er of approaches t o progressing t hrough t he gam e and ( perhaps) event ually winning. I n general, it is a part icular w ay of t hinking about t he gam e st at e from t he perspect ive of a player, t oget her w it h a pat t ern of repet it ive percept ual, cognit iv e, and m ot or operat ions.
A part icular gam eplay gest alt could be unique t o a person, a gam e, or even a play ing occasion. More generally t hough, recurrent gam eplay gest alt s can be ident ified across gam es, gam e genres, and players. Som e exam ples of gam eplay gest alt s include:
Act ion gam es: shoot while being hit , st rafe t o hiding spot , t ake healt h, repeat
St rat egy Gam es: order peasant s, send t o work, order soldiers, send t o perim et ers, repeat while slowly expanding t he perim et ers ( up t o t he point of cat ast rophic w in/ lose) ; OR: m ove x archers t o t ow er y every n m inut es t o head off t he enem y cam el m usket eers from t he east who ar r ive ever y n+ 1 m inut es
I n General: overcom e barrier, save if successful, reload and ret ry if unsuccessful
Such pat t erns m ay or m ay not be explicit ly designed for by t he creat ors of a gam e. They are not gam e design pat t erns in t he sam e sense t hat t he paper/ scissors/ rock syst em is, ie. t hey are not designed int o t he syst em of a gam e. I f designers do t ake t hem int o account , it is in support ing t he developm ent and em ergence of t hese pat t erns in play, nev er, in a good design, by forcing t hem on t he player.
N a r r a t iv e
St ories and narrat ives can be defined as broadly as gam e: everyt hing is a narrat ive/ st ory. Again, t his is not very useful. We can define a narrat ive as an experience t hat is st ruct ured in t im e. Different st ruct ures t hen represent different form s of narrat ive, and a narrat iv e is an experience m anifest ing a specific narrat iv e st ruct ure. A very com m on narrat ive st ruct ure used in com put er gam es, borrow ed from film script w rit ing, is t he t hree- act r est orat ive st ruct ure. The t hree act rest orat ive st ruct ure has a beginning ( t he first act ) in which a conflict is est ablished, follow ed by t he playing out of t he im plicat ions of t he conflict ( t he second act ) , and com plet ed by t he final resolut ion of t he conflict ( t he t hird act ) . This narrat ive st ruct ure also specifically includes a cent ral prot agonist , a conflict involving a dilem m a of norm at ive m oralit y, a second act propelled by t he false t he final resolut ion in act t hree are t ypically achieved by cut scenes, sequences of non- int eract ive video m at erial. level- specific conflict can be enhanced by increasing difficult y t hrough a level, or by an int ernal dram at ic st ruct ure t hat em phasizes t he point of com plet ing t he lev el, such as t he defeat of a lev el boss, t he big barrier conflict s and challenges wit hin a gam e level, which m ay include m onst ers t o be defeat ed or avoided, puzzles t o be solved, or t reasures, clues or keys t hat m ust be found in order t o progress in t he current or fut ure gam e levels, et c. Usually it is only t his low est level of t he act ion gam e plot t hat is highly int eract ive. The linear and non- int eract ive cut scenes fram ing gam e play are r evealed in a predefined order, and w it hin a level all players usually st art in t he sam e place and m ust have com plet ed t he sam e specific set of t asks in order t o com plet e t he
Given t hese definit ions, t he quest ion of t he relat ionship bet w een gam eplay and narrat ive can now be phrased m ore clearly. I n part icular, t he apprehension of an experience as a narrat ive requires t he cognit ive const ruct ion of a narrat ive gest alt , a cognit ive st ruct ure or pat t ern allowing t he percept ion and underst anding of an unfolding sequence of phenom ena as a unified narrat ive. The t hree- act rest orat ive st ruct ure is a very com m on, in fact t he dom inant , exam ple of a narrat ive gest alt in gam es and film s. I t is a pat t ern t hat people underst and and expect , and w ill oft en be disappoint ed if it is not sat isfied ( e.g. , if t he st ory ends before t he cent ral conflict is resolved, or if t he hero dies perm anent ly during t he st ory) . I n playing a com put er gam e, one m ust learn and t hen perform a gam eplay gest alt in order t o progress t hrough t he event s of t he gam e. To experience t he gam e as a narrat ive also requires t he creat ion of a narrat ive gest alt unifying t he gam e experiences int o a coherent narrat ive st ruct ure. The t ension bet w een gam eplay and narrat ive can now be view ed as a com pet it ion bet w een t hese respect ive gest alt form at ion and perform ance processes for percept ual, cognit ive, and m ot or effort . Wit hin t he range of effort required for im m ersion and engagem ent , if gam eplay consum es m ost of t he player's available cognit ive resources, t here will be lit t le scope left for perceiving com plex narrat ive pat t erns ( e.g., w e forget t he m ot ivat ion behind t he charact er's bat t les, and what was t he uber- villain's nam e again?) . More t han t his, t he narrat ive adds lit t le t o player im m ersion and engagem ent ( who cares, it 's fun anyw ay) . Conversely , focusing on t he developm ent of t he sense of narrat ive ( e.g., in t he case of m ult ipat h m ovies) reduces t he play er's need and capacit y for a highly engaging gam eplay gest alt .
Good gam e design achieves bet t er int egrat ion of t he gam eplay and narrat ive st ruct ures of t he gam e. This can be done by m et hods like cont inuously but unobt rusively rem inding t he player of t he narrat ive cont ext ( rat her t han having a few perfunct ory cut scenes) , and using cut scenes and cinem at ic sequences as rew ards at appropriat e m om ent s wit hin t he rhyt hm ic pat t erns of gam e play ( so t hey nat urally fall wit hin pauses and rest s, and are not perceiv ed as int errupt ions) .
w it hin cognit ive, em ot ive, and perform at ive effort . I s it wort h t rying t o j um p over a ravine at t he risk of falling and having t o reload a past gam e st at e for t he sake of a healt h pack t hat m ay help m e t o get past t he t ough enem y ahead wit hout t hen having t o reload and ret ry w hen t he enem y defeat s m e? The conflict is an ergonom ic one w it hin in t erm s of perform ing gam eplay gest alt s. And t his has not hing t o do wit h t he higher-level narrat iv e cont ext . So t he t ension bet w een gam eplay and narrat ive is even m ore fundam ent al t han being a sim ple com pet it ion for cognit ive and perform at ive resources: t he player's invest m ent in t he low level conflict as an act ive part icipant is disconnect ed from any deep narrat ive significance underst ood in t erm s of t he shape of t he higher level narrat ive gest alt . Underst anding t his explains t he perceived t ension bet w een narrat ive and gam e play and suggest s st rat egies for overcom ing t his t ension by developing gam e play m echanics t hat are fundam ent ally dram at ic, in t hat t heir consequences do affect t he higher lev el narrat iv e pat t erns of t he gam e.
Sim u la t ion
Much has been m ade over t he last couple of years of t he view of gam es as sim ulat ions. But what exact ly is a sim ulat ion, such t hat it 's different from a narrat iv e or a gam e? A sim ulat ion can be defined as: a represent at ion of t he funct ion, operat ion or feat ures of one process or syst em t hrough t he use of anot her.
Hence a sim ulat ion m ay involve no specific repet it ive and goal- orient ed act ivit ies ( t here m ay be no obvious end st at e, ot her t han t he player get t ing bored) , and no specific predefined pat t erns in t im e. Tim e pat t erns em erge over t he course of running a sim ulat ion, and can be com plet ely different for different runs. Repet it ive act ion m ay be used t o operat e a sim ulat ion, but m ay not be direct ed t o any specific overall goal.
I t 's int erest ing t o regard single- play er st rat egy gam es from t he sim ulat ion perspect iv e. During com pet it iv e play, t here is an obvious goal. But m any gam es w ill allow us t o cont inue playing aft er all of t he enem ies are defeat ed. Unt il resources run out , t hese gam es m ay t hen chug along indefinit ely sim ulat ing a sim ple econom ic syst em . There is no m ore gam eplay by our st rict ludic definit ion, and t he narrat ive aft er w inning has no int erest ing t em poral ( dram at ic) st ruct ure. Sim ulat ions like flight sim ulat ors are oft en int erest ing from t he perspect iv e of skill dev elopm ent ; t hey are not int erest ing as gam es or st ories, but for underst anding how a part icular syst em funct ions in different circum st ances.
A Un ifie d Cla ssifica t ion Pla n e
Taking t hese t hree form s, t he ludic gam e, narrat ive and sim ulat ion, w e can const ruct a classificat ion plane as a t riangle wit h one form at each point , as shown on Figure 1. I t is t hen possible, as a heurist ic ( ie. a useful working t ool) for com paring different gam es and genres, t o place gam es and genres on t hat plane, em phasizing t he relat ive degree t o w hich t hey em body elem ent s of ludic gam ing, sim ulat ion and narrat ive.
I n t his schem e w e can place avat ar w orlds and vehicle sim ulat ors at t he sim ulat ion ext rem e. Early avat ar worlds w ere t hree dim ensional virt ual spaces in which a user could be represent ed by a m ov able avat ar. These worlds rarely present ed m uch t o do, how ever, since t hey lacked any ludic or narrat ive cont ent .
Figur e 1 . A 2 - dim e n sion a l cla ssifica t ion pla n e sh ow s t h e com pa r a t iv e de gr e e s t o w hich a pa r t icu la r ga m e or ge nr e is lu dic, n a r r a t iv e , or sim u la t ion - ba se d.
At t he narrat ive ext rem e w e place t he fixed narrat ive st ruct ures of digit al linear m ovies. Mult ipat h m ovies hint at gam e- like int eract ion by present ing choices for t he view er, while hypert ext advent ures provide a high degree of int eract ion in t he play er's creat ion of specific narrat iv e experiences.
Act ion gam es, st rat egy gam es and RPGs incorporat e prom inent feat ures of all form s, being gam es, sim ulat ors and narrat ives. RPGs generally have m ore narrat ive cont ent t han act ion gam es, and st rat egy gam es have m ore sim ulat ion t han narrat ive.
Ga m blin g a n d A Th r e e - D im e n sion a l Cla ssifica t ion Spa ce
Gam ing is oft en also underst ood in t he sense of gam bling. The world of com put er gam ers usually appears t o be v ery separat e fr om t he world of gam bling, alt hough gam bling com panies are cert ainly gam e com panies t hat deliver m any gam bling product s as gam es. To cont inue w it h our definit ion fet ish, w e can define gam bling as:
decisions of gain or loss m ade by chance w it hin a fram ework of agreed rules.
Chance is cent ral t o t he idea of gam bling. Of course, m any form s of gam bling have scope for skill; but t hese can be placed som ewhere bet ween gam bling and ludic gam ing by t he definit ions present ed here. I n fact , w e can add anot her point t o our classificat ion syst em and ext end our t wo dim ensional classificat ion plane t o produce a t hree- dim ensional classificat ion space, as shown on Figure 2.
Figu r e 2 . A 3 - dim e n siona l cla ssifica t ion spa ce in t r odu ce s pur e st och a st ic, or pr oba bilist ic, de cision pr oce sse s a s a n e w e le m e n t of for m .
The different point s wit hin t his space represent different degrees by which a product ion represent s a gam e, a narrat ive, a sim ulat ion, or a gam bling syst em . For exam ple, t he gam e of poker has elem ent s of pure gam ing and also elem ent s of gam bling, since it present s a w in/ lose scenario played according t o a rule set , in which chance has a significant im pact upon t he out com e but w it hin which skill can also have a m aj or role. I f we look at t he dim ension from gam bling t o sim ulat ion, we ent er a v ery undev eloped zone of virt ual econom ies, while t he dim ension heading t owards narrat ive suggest s experiences st ruct ured in t im e but significant ly det erm ined by chance.
Fr om Fict ion t o N on - Fict ion Ga m in g
Figu r e 3 . Va r ia t ion s in de gr e e of fict ion a l con t e n t a r e inde pe n de n t of t h e
lu dic/ n a r r a t ive / sim u la t ion cla ssifica t ion of a ga m e , a nd so a r e r e pr e se n t e d a long t h e t h ir d dim e n sion of a cla ssifica t ion pr ism .
Milit ary vehicle sim ulat ors lie st rongly at t he sim ulat ion ext rem e, but com bine elem ent s of bot h real and fict it ious worlds. The fict ion is realized by im aginary ( ie. sim ulat ed) com ponent s lik e enem y vehicles and bat t lefields, while t he non- fict ion elem ent s include accurat e funct ional m odeling of real syst em s, and t he use of physical vehicle m odels as int eract ion and st aging t echnology.
Liv e act ion role- play ing, or LARP, gam es involve perform ances of gam e charact ers in phy sical space. LARPing m ay be m ore or less gam e- lik e, depending upon t he degree t o which play ers use rule set s. But m ost of t he experience is a form of im provisat ional t heat re in which t he players ar e t he audience. Hence LARPing t ends t o be highly fict ional, but lies bet ween sim ulat ion and narrat iv e.
Fr om V ir t u a l t o Ph ysica l Ga m in g
Figu r e 4 . Re pr e se n t in g t he cont inu u m fr om ph y sica l t o v ir t u a l ga m ing for m s a ga m e cla ssifica t ion pr ism cle a r ly diffe r e n t ia t in g liv e a ct ion ga m in g fr om scr e e n - a n
d-k e y boa r d/ con t r olle r - ba se d com pu t e r ga m e s.
Sport s gam es by t his definit ion are very m uch at t he physical ext rem e, while current com put er gam es are predom inant ly virt ual. New form s of locat ion based and m obile gam ing com bine bot h virt ual and physical gam ing, oft en using a com put at ional and m obile infrast ruct ure t o support gam e play act ion in t he real world. Only a sm all num ber of t echnology based gam es have been developed t hat use real- w orld locat ion as a significant fact or in gam eplay. Perhaps t he m ost fam ous exam ple is Bot fight ers, developed by t he sm all Sw edish m obile- gam es st udio I t 's Alive! [ 7] . The gam e t racks GSM- cell locat ion and allow s players w it hin range of each ot her t o score kills and gat her resources t o buy upgrades. Port ugese com pany Ydream s have recent ly launched a Bot fight er- like ant i- t errorist gam e int roducing t he concept of physical sanct uary in cert ain locat ions, m alls and rest aurant s. The proj ect s Can You See Me Now and t he recent Uncle Roy All Around You, creat ed by t he UK m ixed- realit y perform ance group Blast Theory [ 8] , bot h use handheld com put ers, GPS locat ion t racking, and inv isible online play ers t o const ruct gam es where fast physical m ovem ent and dev ice- m ediat ed t eam work are cent ral t o gam eplay.
Use s of Ga m e Cla ssifica t ion Spa ce s
So, w e have a bunch of definit ions, and w e can use t hese t o define som e classificat ion planes and spaces. Of what use is t his in pract ical gam e design?
One use is as a high- level road m ap for m apping out where ot her design t echniques can be applied. I t is very im port ant t o hav e syst em at ic principles for know ing where m ore det ailed t echniques, such as abst ract form al design t ools and gam e design pat t erns [ 4, 6] , should be applied. The dist inct ions of t he t axonom y also allow us t o see where t echniques from ot her fields can be applied. For exam ple, acknowledging t he narrat ive elem ent s of a gam e indicat es w here m et hods for t he const ruct ion of narrat ives, heavily developed for film script w rit ing, can be applied w it hin gam es.
The classificat ion dim ensions also allow us t o separat e concerns. A good exam ple of t his is t he previously described t ension bet w een gam e play and narrat ive. Using definit ions of gam e and narrat ive t hat clearly separat e t hem as form s m akes it clear why t here is oft en a perceived t ension bet w een t hem . The dist inct ion also suggest s a m ore clear- headed approach t o resolving t hose t ensions. I f w e clearly ident ify which aspect s of t he gam e experience are t o have narrat iv e st ruct ure and which are t o be pat t erned gam ing, we can apply narrat ive t echniques at t he right level and consider det ailed m echanics for int egrat ing narrat ive w it h gam e play. We can also ret hink som e m ore fundam ent al quest ions, for exam ple, can w e define gam e m echanics t hat do
seriously advance t he higher- level narrat ive?
The classificat ion dim ensions also support brainst orm ing for gam e ideas. I f a new gam e is placed in a part icular place in t he classificat ion syst em , designers can ask t hem selves about different possible t echniques for int egrat ing t he different form al aspect s of t he gam e. More t han t his, if w e look for areas of t he planes and spaces t hat are em pt y, we can explore new t ypes, form s and genres of gam es. For exam ple, in t he
The m ost obvious use of t he kinds of definit ions present ed here is t o follow Doug Church's suggest ion of developing a com m on design vocabulary. This m ust begin at t he highest level, and can save m uch t im e and confusion in high levels discussions about w hat a gam e proj ect is going t o be. The dist inct ions present ed here cam e out of pract ical experiences in discussing gam e design, and discussions t hat oft en suffered from confusion due t o t he lack of a w ell est ablished design vocabulary at t he highest level. This happens a lot in discussions about where gam es are going, what w e can expect t o see over a t im e fram e ext ending five or t en years int o t he fut ure. New t echnical possibilit ies for locat ion- based and m obile gam ing present m any new possibilit ies for gam e form and experience. We need clear languages for discussing and m aking decisions about t hese possibilit ies.
High level t axonom ies are also a crucial precondit ion for defining t he scope of gam e design pat t erns [ 4] . While a num ber of pat t erns have been ident ified [ 6] , t his is very prelim inary work, and t he m ost useful form s of design pat t erns m ust be regarded as a t opic of ongoing explorat ion. I n fact , t his work will be endless, j ust as t he scope of possible gam es is endless. Our t axonom ies m ust also cont inue t o evolve, as will t he kind of heurist ic design rules com prising Hal Barwood's " 400 Design Rules" [ 1,2] . All of t hese t ools represent com plem ent ary and evolving m et hods for gam e design. They cannot y et be regarded as st able and fully validat ed, but a high level classificat ion syst em can nevert heless save m uch t im e and confusion in gam e design, and provide a cont ribut ion t o t he event ual developm ent of com prehensiv e and syst em at ic t ools for designing gam es of ever increasing com plexit y.
Re fe r e n ce s
[ 1] Hal Barwood "Four of t he Four Hundred 2001" , GDC lect ure, 2001.
[ 2] Hal Bar w ood and Noah Falst ein " Mor e of t he 400: Discovering Design Rules 2002" , GDC lect ure, 2002.
[ 3] Doug Church, " Form al Abst ract Design Tools" Gam asut ra, July 16, 1999.
[ 4] Bernd Kreim eier, " The Case For Gam e Design Pat t erns", Gam asut ra, Decem ber 12, 2002.
[ 5] Craig Lindley 2002 "The Gam eplay Gest alt , Narrat iv e, and I nt eract iv e St ory t elling", Com put er Gam es and Digit al Cult ures Conference, June 6- 8, Tam pere, Finland, 2002.
[ 6] ht t p: / / ww w.gam edesignpat t erns.org/
[ 7] ht t p: / / ww w.it salive.com / page. asp/
[ 8] ht t p: / / ww w.uncleroyallaroundyou. co.uk/