Managing conflicts between users in Wikipedia

B. Ja quemin et al. , Conits in Wikip e dia Managing onits b et w een users in Wikip edia Bernard Ja quemin 1 , Aurélien La uf 1 , Céline Poud a t 2 , Martine Hura ul t-Plantet 1 et Niolas A ura y 2 1 LIMSI CNRS UPR 3251, Orsa y (F rane) 2 ENST, P aris (F rane) {Bernard.Jaquemin,Aurelien.Lauf,Martine.Hurault-Plan tet}limsi.fr {Celine.P oudat,Niolas.Aura y}enst.fr Abstrat Wikip edia is no w ada ys a widely used enylop edia, and one of the most visible sites on the In ternet. Its strong priniple of ollab orativ e w ork and free editing sometimes generates disputes due to disagreemen ts b et w een users. In this artile w e study ho w the wikip edian omm unit y resolv es the onits and whi h roles do wikip edian  ho ose in this pro ess. W e observ ed the users b eha vior b oth in the artile talk pages, and in the Arbitration Committee pages sp eially dediated to serious disputes. W e rst set up a users t yp ology aording to their in v olv emen t in on- its and their publishing and managemen t ativit y in the enylop edia. W e then used those user t yp es to desrib e users b eha vior in on tributing to artiles that are tagged b y the wikip edian omm unit y as b eing in on- it with the oial guidelines of Wikip edia, or on v ersely as b eing w ell featured. Keyw ords: So ial net w ork, Wikip edia, W eb omm unit y , Conit, Col- lab orativ e w ork 1 In tro dution The Wikip edia enylop edia pro jet has b eome a referene informational re- soure, and one of most visible sites on the In ternet. Amazing and far remo v ed from the Enligh tenmen ts spirit  where the exp ert and his signature onstitute the text qualit y guaran tee , Wikip edia is based on a v ery dieren t editorial pro ess. The whole pro jet is based on a few strong ideologial priniples, also alled pil lars , oial guidelines or fundamental priniples in Wikip edia. First, the goal is learly to b e a generalist enylop edia pro jet with sev eral linguisti in- stanes that are indep enden tly managed. Then, the Wikip edia on ten ts also ha v e to b e ob jetiv e. Wikip edians re k on that the b est w a y to gran t the ob- jetivit y is to set out a neutr al p oint of view (NPO V) 1 . Moreo v er, texts are 1 The artiulation b et w een b oth is p erformed as follo w: "What p eople b e- liev e is a matter of ob jetiv e fat, and w e an presen t that quite eas- ily from the neutral p oin t of view." (Jim b o W ales, o-founder of Wikip edia, So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. 1 B. Ja quemin et al. , Conits in Wikip e dia freely edited and redistributed, and the enylop edia has b een dev elop ed with free and op en soure soft w are. The en tire editorial pro ess, from the writing artiles to the marostruture organization, is olletiv ely managed. Finally , the wikip edians ha v e to resp et elemen tary go o d manners. So, ev en if the Wikip edia editorial pro ess totally diers from the traditional enylop edia one, the goals of enylop edi relev ane and ob jetivit y are in fat v ery lose [5 , 7℄. Sev eral formal and informal w a ys to regulate and on trol the enylop edia ha v e progressiv ely b een in tro dued b y the wikip edian omm unit y in order to ob ey and to mak e users ob ey the pil lars . The ommon wikip edian philosoph y mak es it p ossible to gather together a large p opulation of users writing ab out an unlimited n um b er of themes or domains, to share their inomplete kno wledge, to represen t the v arious w a ys of thinking, and to delete errors thanks to suessiv e users retiations [15 , 3℄. Ho w ev er, this philosoph y also generates disputes and onits link ed to inevitable disagreemen ts b et w een on tributors. What pro esses do es the wikip edian omm unit y use to resolv e the onits, and what roles do the wikip edians  ho ose in this pro ess? In this artile, b y analyzing the on tributors b eha vior in plaes where on- its are resolv ed, w e pro vide elemen ts to help answ er these questions. The users b eha vior is observ ed b oth in the artiles that are tagged as b eing in par- tiular aordane ( go o d or fe atur e d artiles ), or on v ersely not in aordane, with the main guidelines of Wikip edia ( r elevan e dispute artiles , NPO V dispute artiles . . . ), and in pages sp eially dediated to serious p ersonal onits, the A rbitr ation Committe e [16 , 13 ℄. As a result, w e presen t the follo wing on tribu- tions: First, w e mak e a users t yp ology aording to parameters that bring to ligh t their in v olv emen t in onits and their publishing and managemen t ativit y in the enylop edia. In partiular, w e establish relationships b et w een the n um b er of app earanes b efore the Arbitration Committee, the initiation of a request to the Arbitration Committee, and the n um b ers of on tribution to artiles and talk pages of Wikip edia. W e sho w that ma jor on tributors are often in v olv ed in arbitration, and mostly as the initiating part y . Then, w e analyse the distribution of those t yp es of users among the on- tributors to artiles that do not resp et a neutral p oin t of view, giv en that it is one of the most imp ortan t priniples of Wikip edia. W e nd that all the ma jor on tributors who tak e their onit b efore the Arbitration Committee are also on tributors to NPO V artiles, against only one half for the minor on tributors. Finally , b y analysing the distribution of those wikip edians in v olv ed in se- rious disputes, among the on tributors to tagged artiles, w e nd that ma jor on tributors who are often in v olv ed in arbitration, are m u h more frequen tly on tributing to proteted artiles (sub jet to disputes or v andalism), than to featured artiles. http://en.wikip edia.o rg/wiki/Wikip edia_talk:A ttr ibut ion/Role _of_trut h ). Th us the Wikip edia's aim at the ob jetivit y is only p erformed at an opinion in v en tory lev el, despite their unev en qualit y on the same page [8℄. 2 So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. B. Ja quemin et al. , Conits in Wikip e dia 2 Related w ork A n um b er of authors study onits in Wikip edia in relation with o ordina- tion and o op eration underlying ollab orativ e w ork. F or instane, [9℄ dev elop quan titativ e measures of the osts in v olv ed b y ollab orativ e w ork, using the onepts of diret (i.e. writing artile) and indiret w ork (i.e. disussion or an ti-v andalism). A t the artile lev el, the history of the revisions is often used to mo del and iden tify onit or o ordination p erio ds [9 , 14 ℄. The aim of the presen t study is rather to analyse the b eha vior of wikip edians, who are in v olv ed in onits, faed with the main to ols wikip edians use to resolv e onits. Studies of onit managemen t and so ial on trol in virtual omm unities sho w that su h so ial systems ha v e the same kind of problems as real so ial systems. In partiular, [10 ℄ sho w that the so ial dilemma b et w een individual and olletiv e in terest in the problem of o op eration remains, ev en if it tak es other forms. F urthermore, [ 4 ℄ observ es that metho ds using b oth mediation and arbitration b etter manage onits than p o w er strategies of so ial on trol, as it do es in the real w orld. Indeed, the w a y a omm unit y manages its onits rev eals its go v ernane mo de [ 2 , 9 , 14 ℄. In the F ren h Wikip edia, mediation tak es plae in talk pages of artiles whi h ha v e a template message at the top of the page, and arbitration tak es plae in the Arbitration Commitee pages. In fat, template messages at the top of artile pages are strongly link ed to the oial guidelines of Wikip edia. Indeed, these priniples pla y an imp ortan t role in the managemen t and resolution of onits. [ 15 ℄ analysed the on ten t of the artile talk pages, and found that 7.9% of the ativit y in those pages onsists in referenes to Wikip edia oial guidelines. The b eha vior of wikip edians has b een studied either from their motiv ations p oin t of view [11 ℄, either onsidering the t yp e [ 12 ℄ or the ev olution of their partiipation [3℄. Our analysis of the b eha vior of wikip edians is based on quan- titativ e data as w ell as in [ 12 ℄, but is restrited to those wikip edians who are in v olv ed in onits. 3 Corpus Wikip edia is a generi term for the free m ultilingual and ollab orativ e online enylop edia 2 as w ell as a referene to ev ery instane of this enylop edia. Ea h instane refers to a dieren t oun try and/or language. The instane w e are in terested in for this artile is the F ren h v ersion of Wikip edia 3 . The orpus w e used w as extrated from the Wikip edia ba kup of 2006/04/02: more than 600,000 pages inluding 370,000 artile pages and 40,000 talk pages (aording to Wikip edia's in ternal ar hiteture, ea h artile page an b e link ed to a talk page). A to ol alled Wiki2T ei 4 w as then used in order to on v ert the wikitext syn tax to a TEI-omplian t XML syn tax (TEI standing for T ext En o ding Initiative ). The artiles of Wikip edia are written b y v olun tary on tributors w orking with ea h other via a wiki. Sine an y one an freely edit an y artile, man y virtual 2 A v ailable at http://www.wikip edia.o rg/ . 3 A v ailable at http://fr.wikip edia.o rg/ . 4 Op en soft w are a v ailable at http://wiki2tei.sourefo rge.net/ and freely distributed aording to the terms of the BSD liense ( http://www.op ensoure.o rg/lien es/ bsd - liense .php ). So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. 3 B. Ja quemin et al. , Conits in Wikip e dia plaes are pro vided to a v oid or settle onits that ma y arise in the pro ess. First of all, ea h artile is link ed to a disussion page where on tributors an ex hange and justify their assertions, and th us rea h ompromises aording to Wikip edia's netiquette and neutralit y p oliy . F urthermore, users an insert sp e- i tags 5 on top of artiles whi h do not resp et Wikip edia's oial guidelines (su h as neutralit y or relev ane dispute) [ 11 , 6 ℄ or, on the on trary , to rew ard an exemplary artile (alled fe atur e d or go o d artiles 6 ). Theses tags are used to highligh t for the omm unit y the fat that some artiles need impro v emen t and th us an b e used as p oin ts of referene for users. Finally , when disputes degenerate in to p ersonal onits and get out of hand, ea h user an register a omplain t to the Arbitration Committee. The Arbitration Committee is a group omp osed of sev en on tributors to Wikip edia, eleted b y the rest of the omm unit y for six mon ths. Delib erations and v otes of the Arbitration Commit- tee are publi and usually tend to rea h unanimit y , whi h implies onsensus, as it is the rule for the artiles. The role of Arbitrators is not to express an opinion ab out the sien ti righ tness or the editorial p oliy of an artile but to ensure that Wikip edia's oial guidelines are resp eted: neutral p oin t of view (NPO V), the need to ite general soures, netiquette (alled wikilove b y the wikip edian F ren h omm unit y), the resp et of the la w, et. They ha v e the righ t to imp ose santions on users su h as temp orary or denitiv e artile probation (meaning that the user annot on tribute an ymore to one or more artiles) or, less often, general restrition (meaning that the user is literally banned from all Wikip edia). Th us, there are three virtual plaes to manage a onit, in order of serious- ness: the disussion pages link ed to an artile, the disussion pages link ed to an NPO V dispute artile and the pages of the Arbitration Committee. W e fo us on the last t w o b eause they orresp ond to op en onits. The rst orpus w e olleted is omp osed of ab out 1,000 artiles that ha v e (or ha v e had) the NPO V tag. Ea h artile is asso iated, when p ossible, to its disussion page (some artiles are not link ed to a disussion page b eause the disussion ma y ha v e started after w e extrated the orpus). Ab out 1,600 on- tributors in terv ened in these pages. W e automatially added seman ti tags to this orpus in order to extrat ea h on tribution and its size, who wrote it and when  whi h tells us whi h on tributions w ere written during the onit and whi h w ere not  and, when p ossible, to whom it answ ers. Ho w ev er, it is imp os- sible to kno w who wrote a on tribution when users do not sign it, delib erately or not. This is the reason wh y b et w een 2% and 5% of the on tributions ma y ha v e b een improp erly tagged. The seond orpus is omp osed of ab out 80 pages from the Arbitration Com- mittee. These pages are relativ ely w ell formed and homogeneous, allo wing us again to automatially tag them so as to learly mak e their essen tial ar hite- ture stand out: the onit desription, who registered the omplain t and when, the parties in v olv ed, if the omplain t is admissible or not, and the v erdit of the arbitrators. F urthermore, ea h user is asso iated to his messages, and ea h arbitrator to his on tributions and, of ourse, his v ote. Finally , the v erdit is omp osed of at least one v erdit prop osal and a v ote; there are as man y 5 Dened in Wikip edia as "a frame t yp e in artiles indiating a piee of information or a link" http://en.wikip edia.o rg/wiki/Wikip edia:T emplate . 6 http://en.wikip edia.o rg/wiki/Wikip edia:Go o d_a rtile s . 4 So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. B. Ja quemin et al. , Conits in Wikip e dia oun terprop osals and v otes as needed un til the arbitrators are able to rea h an agreemen t. Ea h prop osal is learly iden tied and asso iated to the righ t ar- bitrator and ea h v ote is asso iated to its arbitrator and to the prop osition it refers to. 4 T yp ology of users in onit The Arbitration Committee is therefore a formal plae for the resolution of onits. Though rather rare  only ab out one h undred users among 31 000 wikip edians w ere implied in an arbitration within a 5-y ear p erio d , arbitra- tions represen t an imp ortan t to ol for Wikip edia go v ernane. Indeed, eleted arbitrators an imp ose p enalties against Wikip edia users who transgressed the pil lars . F or instane, p enalt y ma y onsist in blo  king a user in order to k eep the user from writing within artiles during a ertain p erio d of time. It therefore giv es strong means for on trolling publiation. Among the h undred arbitrations whi h to ok plae from the b eginning of Wikip edia-F rane to 2006 april, some user names app ear more often, either as the initiating p arty , or as the other involve d p arty . Those t w o topis, frequeny of app earane and role in the omplain t, allo w us to dra w up an initial t yp ology of users engaged in a dispute. W e rst distinguished three kind of protagonists dep ending on the frequeny of their app earanes: very r e gular ones who ha v e b et w een 3 et 14 app earanes 7 , r e gular ones who ha v e t w o app earanes, and o   asional ones who ha v e only one app earane. Conerning their role in the omplain t, w e then distinguished three ategories, the initiating p arty , that is to sa y those who are most often the initiator of the omplain ts, the other in- volve d p arty , and nally those who app ear in a more balaned w a y , sometimes as initiating part y and sometimes as other in v olv ed part y . W e an see on T a- ble 1 that among the wikip edians who often app ear, the very r e gular ones , are the initiating part y for most part, ev en though o   asional ones , who app eared only one, are mainly other in v olv ed part y . W e also note that most of those who app eared t wie to ok one the initiating part y p osition, and one the other in v olving part y p osition. T able 1: App earanes b efore the Arbitration Committee App earanes Users Initiating part y Other part y Both 314 (v ery regular ones) 10 50% 30% 20% 2 (regular ones) 17 12% 29% 59% 1 (o asional ones) 74 30% 70% 0% W e then added to that t yp ology the w a y users on tribute to Wikip edia. W e onsidered the n um b er of their on tributions in editing artiles, either in ar- tile pages, or in the disussion pages, b eause it is mainly in this plae that onits b egin 8 . Conerning this p oin t, w e noted big dierenes b et w een users. W e drew up four ategories, the major  ontributors whose n um b er of on tribu- tions extends from ab out 12,000 to 40,000 during the studied p erio d, the L ar ge 7 14 is an yw a y a sort of reord, then there are t w o of them ha ving 7, another ha ving 4, the other ones ha ving 3 app earanes 8 W e did not onsider for instane on tributions in the bistr ots of Wikip edia. So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. 5 B. Ja quemin et al. , Conits in Wikip e dia  ontributors , b et w een 2,800 and 12,000 on tributions, the midd le  ontributors b et w een 600 and 2,800, and the minor  ontributors , b et w een one and 600 on- tributions. Finally , w e onsidered the t yp e of their on tributions aording to whether they on tribute to artile pages or disussion pages. W e therefore dis- tinguished three ategories aording to whether they on tribute more often to artiles or to disussions, or to b oth of them in a balaned w a y . T able 2: The on tributions of the protagonists b efore the Arbitration Committee Con tributions Users Artile orien t. Disussion orien t. Both 12,00040,000 (Ma jor on trib.) 7 100% 0% 0% 2,80012 000 (Large on trib.) 23 96% 0% 4% 6002,800 (Middle on trib.) 31 81% 0% 19% 1600 (Minor on trib.) 40 70% 5% 25% T able 2 sho ws that users who get in v olv ed in disputes in Wikip edia on- tribute more to artiles than to the asso iated talk pages, despite their onits. Nev ertheless, it also sho ws that the less they on tribute to artiles, the more they ha v e a tendeny to disuss. Comparing the n um b er of on tributions and the frequeny of app earanes (T able 3), w e realize that parties of the Arbitration Committee who are very r e gular are for the most part big  ontributors , while o   asional ones are more often smal l  ontributors . T able 3: Categories of on tributors in omplain ts App earanes Con tributors Ma jor Large Middle Minor 314 (v ery regular ones) 10 20% 50% 30% 0% 2 (regular ones) 17 13% 29% 29% 29% 1 (o asional ones) 74 4% 18% 31% 47% Comparing the n um b er of on tributions and the role in the omplain t (T a- ble 4 ), w e note that the big  ontributors are more often the initiating part y and that the smal l  ontributors are more often the other in v olv ed part y . Indeed w e note an inrease of the prop ortion of other involve d p arty and a derease of the prop ortion of initiating p arty as the n um b er of on tributions dereases. P art of protagonists who are sometimes the initiating part y and sometimes the other in v olv ed part y is marginal for ea h ategory of on tribution size. T able 4: Role in the omplain t b y size of on tribution Con tributions Users Initiating part y Other part y Both 12,00040,000 (Ma jor on trib.) 7 57% 29% 14% 2,80012,000 (Large on trib.) 23 39% 44% 17% 6002,800 (Middle on trib.) 31 32% 58% 10% 1600 (Minor on trib.) 40 15% 75% 10% The analysis of these tables ev ok es that the big on tributors assimilated the pil lars of Wikip edia, and really are ab out enforing them [1, 6 ℄. Indeed, the emerging trend is that the more they on tribute to artiles, the more they arry 6 So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. B. Ja quemin et al. , Conits in Wikip e dia out publiation on trol at the same time. They exerise this on trol in the framew ork of the Arbitration Committee through their role as initiating part y . They exerise this on trol mainly o v er midd le and smal l on tributors. In the follo wing setion, w e study whether w e an omplete this t yp ology of on tributors b efore the Arbitration Committee with the t yp es of artile they on tribute to, in v olving the pil lars of Wikip edia. Indeed, w e sa w that users put dieren t tags within artiles in order to w arn other users ab out brea hes of the rules of Wikip edia. W e used those tags to ategorize artiles as fe atur e d artiles , NPO V dispute artiles , r elevan e dispute artiles , and pr ote te d artiles . 5 Users in onit and pil lars of Wikip edia The NPO V dispute tag is the rst tangible evidene of a disagreemen t b et w een wikip edians. Th us w e studied  harateristis of on tributors who partiipated in artiles with the NPO V tag, and partiularly the ones who are also parties of arbitration b y the Arbitration Committee. This analysis rev eals sev eral b e- ha vior trends. In T able 5 , w e study the b eha vior of the on tributors, shared out in ategories follo wing the n um b er of their on tributions. W e ompare on trib- utors in artiles with a NPO V tag to all the on tributors in Wikip edia. The seond olumn indiates for ea h setion the n um b er of on tributors in NPO V artiles. The third olumn sho ws the n um b er of app earanes b efore the Arbi- tration Committee for the on tributors in NPO V artiles in omparison with all the protagonists b efore the Arbitration Committee, for ea h ategory (see T able 2). In T able 6, w e study the b eha vior of the on tributors who app ear b efore the Arbitration Committee, onsidering on the one hand the app earane fre- queny , and on the other hand their role in the omplain t. The seond olumn indiates, for ea h ategory of frequeny and of role, the n um b er of on trib- utors in Wikip edia who app ear b efore the Arbitration Committee. The third olumn indiates for ea h ategory the n um b er of on tributors in NPO V arti- les who app ear b efore the Arbitration Committee, and the prop ortion of these on tributors to all the on tributors of the same ategory who app ear b efore the Arbitration Committee. T able 5 sho ws that 77% of the protagonists b efore the Arbitration Committee app ear among the 1600 on tributors partiipating to at least one artile with the NPO V tag. It suggests that a lot of onits arise from an ob jetivit y on tro v ersy . T able 5: Protagonists who app ear b efore the Arbitration Committee (A C) among the on tributors in NPO V artiles, b y on tributions size Con tributors ategories # NPO V on tributors NPO V on tributors b efore the A C Ma jor on tributors 30 7 (100% of 7) Large on tributors 151 21 (91% of 23) Middle on tributors 335 27 (84% of 31) Minor on tributors 1121 23 (57% of 40) T otal 1637 78 (77% of 101) W e also notie (T able 5 ) a v ery mark ed presene of the protagonists who app ear b efore the Arbitration Committee among the most v erb ose on tributors of our sample. W e also note (T able 6 ) that the very r e gular pr otagonists b efore So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. 7 B. Ja quemin et al. , Conits in Wikip e dia T able 6: Protagonists who app ear b efore the Arbitration Committee (A C) among the on tributors in NPO V artiles, b y app earanes t yp e Protagonists ategories Before the A C In NPO V artiles V ery regular 10 10 (100%) Regular 17 12 (70%) Oasional 74 56 (76%) Initiating part y 29 26 (90%) Other part y 60 44 (73%) Both 12 8 (67%) the Arbitration Committee and the initiating parties on tribute more in NPO V pages than r e gular and o   asional pr otagonists , or than other in v olv ed parties. The very r e gular pr otagonists and initiating parties are partiularly presen t in NPO V disussions. In order to study further the b eha vior of the on tributors in onit, w e no w onsider their partiipation in other artiles with a partiular tag, indiating either a brea h of relev ane or ob jetivit y priniples, or a partiular ageemen t with the oial guidelines of Wikip edia. These tags are the neutral p oin t of view (NPO V) dispute tag, the relev ane dispute tag and the proteted artile tag, that tak es plae when the on tro v ersy degenerates in to onit in order to prev en t the artile from b eing mo died, and the featured artile tag, that indiates its partiular qualit y , aording to the pil lars . Figure 1: Con tributors in proteted artiles and protagonists In Figures 1, 2, 3 and 4, the sample omprises only on tributors in NPO V artiles, who sometimes also on tribute in artiles with another tag. The urv es in these gures presen t in desending order the n um b er of on tributions for the 20 most v erb ose on tributors, resp etiv ely in proteted artiles, in featured 8 So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. B. Ja quemin et al. , Conits in Wikip e dia artiles, in NPO V artiles and in non-relev an t artiles. F or ea h on tributor, the n um b er of his app earanes b efore the Arbitration Committee (v ertial line) and the n um b er of his omplain ts (small irle) are also indiated, orresp onding to the righ t sale. Figure 2: Con tributors in featured artiles and protagonists W e observ e sev eral in teresting dierenes in these gures. In partiular, among the 20 most v erb ose on tributors in proteted artiles (Figure 1), 7 are protagonists b efore the Arbitration Committee, namely 35% of the ma jor on- tributors on these artiles. F urthermore, their b eha vior b efore the Arbitration Committee is disparate: some of them initiate the pro edure and the others are other in v olv ed parties, some are v ery regular or regular protagonists and the others are o asional ones. On the other hand, Figure 2 sho ws that, among the most v erb ose on tributors in featured artiles, only 3 app eared b efore the Arbi- tration Committee, all of them as initiating parties. Nonetheless their apparen t aggressiv eness m ust b e put in to p ersp etiv e: as none of these protagonists is a regular one, the omplain ts are few. The b eha vior of the ma jor on tributors in NPO V and relev ane dispute ar- tiles is b et w een these t w o trends. Among the 20 most proli on tributors in NPO V artiles indeed (Figure 3), 25% app eared b efore the Arbitration Com- mittee. And 4 of the 20 ma jor on tributors in non-relev an t artiles, ie 20%, also app eared in arbitrations (Figure 4 ). In all these gures, the wikip edians with a partiular status 9 are starred (*). It is in teresting that most of the ma jor on tributors in the onsidered artiles ha v e also a partiular status. 9 Some partiular status exists in the wikip edian omm unit y , e.g. administrator, stew- ard, arbitrator, bureaurat. . . Su h a status is onferred b y the omm unit y to a on tributor through an eletion pro ess. This status gran ts him/her extended righ ts in prosp et of man- aging the enylop edia. So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. 9 B. Ja quemin et al. , Conits in Wikip e dia Figure 3: Con tributors in NPO V artiles and protagonists Figure 4: Con tributors in non-relev an t artiles and protagonists This observ ation onrms the previously men tioned orrelation b et w een a strong in v olv emen t of the on tributors in the Wikip edia pro jet, denoted b oth b y the n um b er of on tributions and b y the partiular status [ 1 , 6 ℄, and their in terv en tion where and when the oial guidelines need to b e proteted. 10 So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. B. Ja quemin et al. , Conits in Wikip e dia 6 Conlusion The Wikip edia enylop edia is mainly based on ollab orativ e w ork. This oial guideline yields to o op eration patterns, inluding disussions and information sharing in order to realize the ommon goal. But su h an extended ollab oration also engenders onits. Disagreemen ts whi h degenerate in to serious p ersonal disputes, with p ossible insults or systemati rev erts, are nally not so frequen t. They only in v olv ed one h undred users among 30,000 wikip edians o v er a p erio d of v e y ears. Oial guidelines, the Wikip edia pil lars , are lear, and there are not man y of them. They onstitute strong bases for onit resolution. T o ols and pro edures ha v e b een dev elop ed step b y step in order to enfore those priniples. W e studied onit ev olution through the b eha vior of users who app ear b efore the Arbitration Committee, and through their on tributions to those artiles that are tagged su h as fe atur e d artiles , NPO V artiles , non-r elevant artiles , and pr ote te d artiles . As exp eted, users app earing b efore the Arbitra- tion Committee are more n umerous on artiles sub jet to a NPO V or relev ane on tro v ersy , and m u h more on proteted artiles, than on featured artiles. The presene of in v olv ed parties b efore an Arbitration Committee has dif- feren t meanings dep ending on whether one is the initiating part y or the other in v olv ed part y . W e note that ma jor and large on tributors, also often in v olv ed as Wikip edia administrators, do most of the job of publiation on trol. They are more often the ones who initiate arbitrations, and moreo v er the ones who on tribute the most to featured artiles. T ables 2, 3 , 4 of Setion 4 learly sho w the ev olution of the relativ e sizes resp etiv ely b et w een initiating parties and other in v olv ed parties, b et w een on tribution to artiles and on tribution to disussions, b et w een regular and o asional in v olv ed parties b efore Arbitration Committee, aording to the size of on tributions. As a result, w e ma y sa y that onits in Wikip edia are resolv ed b oth b y means of a strong ommitmen t to lear oial guidelines, through sp ei plaes dev oted to managing them, and b y in terv en tions of some atten tiv e users. A  kno wledgemen ts The resear h rep orted here w as supp orted b y a gran t from the F ren h National Resear h Ageny (ANR), within the framew ork of the Autograph pro jet ANR- 05-RNR T-03002 (S0604108 W). Referenes [1℄ An thon y , D., Smith, S., Williamson, T.: Explaining Qualit y in In ternet Colletiv e Go o ds: Zealots and Go o d Samaritans in the Case of Wikip edia. Dartmouth College, Hano v er (2005) [2℄ Aura y , N., P oudat, C., P ons, P .: Demo ratizing sien ti vulgarization. The balane b et w een o op eration and onit in F ren h Wikip edia. Observ atorio 3, 185199 (2007) [3℄ Bry an t, S.L., F orte, A., Bru kman, A.: Beoming Wikip edian: transfor- mation of partiipation in a ollab orativ e online enylop edia. In: A CM So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99. 11 B. Ja quemin et al. , Conits in Wikip e dia SIGGR OUP Conferene on Supp orting Group W ork, pp. 110. A CM Press, New Y ork (2005) [4℄ DuV al Smith, A.: Problems of Conit Managemen t in Virtual Comm u- nities. In: Smith, M., K ollo  k, P . (eds.) Comm unities in Cyb erspae, pp. 134166. Routledge, London (1999) [5℄ Endrezzi L.: La omm unauté omme auteur et éditeur: l'exemple de Wikip édia. In: Jounée d'étude des URFIST "Év aluation et v alidation de l'information sur In ternet" (2007) [6℄ F orte, A., Bru kman, A.: Wh y Do P eople W rite for Wikip edia? Inen- tiv es to Con tribute to Op en-Con ten t Publishing. In: GR OUP 05 W orkshop: Sustaining Comm unit y: The Role and Design of Inen tiv e Me hanisms in Online Systems (2005) [7℄ Giles, J.: In ternet enylopaedias go head to head. Nature 438(7070), 900 901 (2005) [8℄ Gourdain, P ., O'Kelly , F., Roman-Amat, B., Soulas, D., v on Droste zu Hül- sho, T.: La Rév olution Wikip édia. Les enylop édies v on t-elles mourir? Mille et Une Nuits, P aris (2007) [9℄ Kittur, A., Suh, B., P endleton, B.A., Chi, E.H.: He sa ys, She Sa ys: Conit and Co ordination in Wikip edia. In: SIGCHI Conferene on Human F ators in Computing Systems, pp. 453462, A CM Press, New Y ork (2007) [10℄ K ollo  k, P ., Smith, M.,: Managing the virtual ommons: o op eration and onit in Computer omm unities. In: Computer-Mediated Comm uniation: Linguisti, So ial and Cross-Cultural P ersp etiv es, Susan Herring (ed.), pp. 109128. John Benjamins, Amsterdam (1996) [11℄ Kuznetso v, S.: Motiv ations of on tributors to Wikip edia. A CM SIGCAS Computers and So iet y 36(2), 17 (2006) [12℄ Ortega, F., Gonzalez-Barahona, J.M.: Quan titativ e Analysis of the Wikip edia Comm unit y of Users. In: WikiSym'07, pp. 7586, Mon treal, Canada (2007) [13℄ Stvilia, B., T widale, M., Gasser, L., Smith, L.: Information Qualit y Dis- ussions in Wikip edia. T e hnial Rep ort, Univ ersit y of Illinois at Urbana- Champaign (2005) [14℄ Viégas, F.B., W atten b erg, M., Da v e, K.: Studying Co op eration and Con- it b et w een Authors with history o w Visualizations. In: SIGCHI Confer- ene on Human F ators in Computing Systems, pp. 575582. A CM Press, New Y ork (2004) [15℄ Viégas, F.B., W atten b erg, M., Kriss, J., V an Ham, F.: T alk Before Y ou T yp e: Co ordination in Wikip edia. In: 40th Ha w aii In ternational Conferene on System Sienes (2007) [16℄ Zlati, V., Bozievi, M., Stefani, H., Domazet, M.: Wikip edias: Collab- orativ e w eb-based enylop edias as omplex net w orks. Ph ysial Review E, 74(1) 611 (2006) 12 So ial Asp ets of the W eb 08, Innsbru k, Ma y 6 2008, pp. 87-99.