Wikipedia:Wikipedia Signpost/Drafts/Reforming ArbCom



Reduce the workload

 * Firstly, and most importantly, we should remove ArbCom's responsibilities concerning privacy/legal issues, and turn them over to WMF professionals. All cases involving personal information, off-wiki harrassment, child protection, and other sensitive privacy/legal matters would no longer be handled by ArbCom, but rather by paid, specialized WMF staff. Untrained, anonymous, non-professional volunteers should never, under any circumstances, be actively handing down decisions in cases involving the law or people's sensitive real-life information. These cases can and do have real-world consequences, and such cases are far too sensitive and overwhelming to be handled by untrained encyclopedia volunteers.
 * Currently, according to the page of the now-disbanded BASC, ArbCom hears appeals from "(i) from editors who are subject to an OversightBlock or a Checkuserblock; (ii) from editors who are blocked for reasons that are unsuitable for public discussion; and (iii) from editors blocked or banned by Arbitration and Arbitration Enforcement decisions." Appeal procedures for block/ban type (iii) should be left as is, but in accordance with the proposal made above, blocks/bans of type (ii) should be handed over to the WMF professionals. Appeals for sanctions of type (i) could be heard by the general body of uninvolved functionaries, perhaps via the mailing list. Someone would submit an appeal to the relevant functionary mailing list, and the functionaries would then discuss whether to unblock the user. In this way, ArbCom would only hear appeals for sanctions that were imposed by ArbCom itself or by admins at AE. This is a more logical setup.
 * ArbCom's involvement in functionary issues should be completely removed. Functionary matters have no direct connection to resolving disputes. Instead, functionaries would be appointed yearly (perhaps in concurrence with the ArbCom election) by a simple community vote, in the same way that arbs themselves are elected. Allegations of functionary abuse should be exclusively handled by the global Ombudsman commission, which is dedicated to such matters.

If all these responsibilities were devolved to other bodies, ArbCom would be focusing 100% on what they should really be doing: resolving complex on-wiki disputes between editors. With their complete focus on this, they would have the time to pay more attention to individual cases and formulate better decisions. The more the workload is distributed, the less burnout there is. That would be a favorable result for everyone.

A new venue
I propose establishing a new intermediate DR venue to help resolve some disputes before they ever reach ArbCom. For now, it will be called IDR (intermediate dispute resolution). Currently, ArbCom requires "prior dispute resolution" to accept a case, but unfortunately, pretty much anything passes for "dispute resolution," and a commonly cited example of "prior DR" is some ANI thread. But ANI is not dispute resolution; we all know that ANI is a disaster, and much more often than not it simply worsens disputes instead of resolving them. This proposal suggests a real, general-purpose DR venue that would be used before going to ArbCom.

Basically, IDR would be overseen by coordinators with proven experience in DR. The coordinators would be chosen in a simple annual election, much like the Guild of Copy Editors or the Military History project selects their coordinators. One primary coordinator responsibility would be approving new mediation volunteers. To become a volunteer, an editor would submit a brief request to the coordinators, describing why they believe they would be a good mediator. After reviewing the editor's background for experience and maturity, the coordinators would vote by simple majority on whether to admit the editor. Once the editor is approved, there would be no term, although someone could be removed as a mediator in the event of gross misconduct or long-term inactivity.

But why is a selection process even needed? Even if it is needed, why can't the community do it directly? A selection process is necessary to ensure that the editor has the qualities to be a good mediator and effectively resolve disputes. Otherwise, some editor that is inexperienced, immature, foul-mouthed, etc., could simply sign up, start "mediating," and in the end cause more problems than they help solve. And if the community directly elected the mediators on a rolling basis, we would inevitably end up inflating the importance of the position and creating another process similar to RfA. An evaluation and vote by a small group of editors would be much more quiet, efficient, and far less brutal than an RfA-style process. It is easily demonstrable that it is far easier to reach an agreement among a group of a few people. And besides, it would be easier for the community to just annually elect a few coordinators and have them select new mediators, rather than handle all the applications on its own.

IDR would handle all disputes, both "content" and "conduct." Currently, though, content disputes are supposed to be handled via 30/DRN/MedCom (although MedCom is inactive, for all practical purposes), and conduct disputes through a poorly-defined ANI/ArbCom sequence. In my opinion, the current system of attempting to separate these two kinds of disputes does not give sufficient consideration to reality. The fact is that both elements can be fundamental elements of a dispute. Suppose a content dispute arises, and in the process of disputing the content issue, conduct devolves as well. This creates a content and conduct dispute, and trying to split up the different "parts" of the dispute into different venues will just create a mess. Disputes of this type are actually quite common, and in practice many disputes of this dual nature end up at ArbCom anyway, even though ArbCom is theoretically supposed to handle only conduct disputes. There is actually no real pre-ArbCom venue for resolving disputes involving conduct, although such a thing is sorely needed to peacefully resolve some disputes and avoid more than necessary ArbCom confrontations and potentially overblown sanctions.

IDR would fill these gaps and become the primary intermediate step before going to ArbCom. We could merge DRN into IDR, although we could keep third opinion as a starting point for basic disputes; for consistency, however, we might also expand it to include third opinions on conduct matters. To summarize all this, third opinion would become the starting point for basic disputes, IDR would be the intermediate step for mediating disputes, and ArbCom the final, binding, sanction-imposing last resort for all disputes. The false boundary between content and conduct disputes would be eliminated. This a much more direct process than the disorganized, divided one we have now.

So, how would IDR cases operate? Case procedures would borrow from DRN, with some modifications. The parties would first make their statements, describing the dispute and their view of the matter. Making the decision as to whether to accept a case would be the responsibility of the coordinators. Any coordinator could either accept or decline a case; in other words, for efficiency purposes, individual coordinators could act independently in case acceptance matters. Before accepting a case, a coordinator should ensure that more basic resolution (such as third opinion or a substantial talk page discussion) has been attempted. If no such thing has occurred, the case will be declined and the parties will be advised to pursue talk page discussion or a third opinion before returning. If the case is accepted, the case will be assigned a mediator. Through a discussion, the mediator will attempt to diffuse the dispute by encouraging civil negotiation between the parties and by actively proposing compromises. If the parties reach an agreement, the case will be closed as successful. However a coordinator may close the case as failed if it becomes clear that no compromise will ever be reached. The case will then be referred to ArbCom. Although participation would technically be voluntary, editors would have a strong incentive to participate due to the inattractiveness of the alternative: mandatory arbitration and sanctions by ArbCom. To strengthen this incentive, an IDR case closed as failed because of refusal to participate would count as attempted DR for the purpose of filing ArbCom cases; this would ensure that a party could not escape eventual dispute resolution by simply refusing to participate in IDR.

Cases and enforcement
The best way to streamline the case procedures would be to simply make cases more focused on the parties, instead of leaving the door open for making every case a community dramafest. Here is my proposal for new case procedures, step by step:
 * 1) Case request. The filing party files a case request, describing the issue being disputed. They would also list any prior attempts at real dispute resolution (ANI does not count). The other listed parties are then notified, and they are given a certain amount of time to file their own statement describing their own perspective on the case. If they do not respond within the prescribed amount of time, the arbs simply vote on whether to accept the case. If the case is accepted, it simply proceeds to evidence phase. There must be some deadlines—a common problem in the current ArbCom is that arbs will wait for a very long time before all parties make their statements, and this clogs the system. Flat deadlines will also ensure equality for all parties in all cases. Certain editors will not get more time to make a statement in one case than another editor does in another case.
 * 2) *Statements from non-party community members would not be permitted during the request phase. It would be best if ArbCom could decide that issue for themselves, free from the overwhelming confusion that must arise as a result of the many thousands of words from the plethora of drive-by statements from random people. Besides, uninvolved Wikipedians really have no direct stake in whether or not ArbCom accepts a case. ArbCom and the parties are the ones most affected by that decision.
 * 3) Evidence. If the case is accepted and opened, the evidence phase begins. Involved parties would present evidence for their side of the case, such as diffs and elaborations on those diffs, such as how the conduct shown violates policy. Evidence from uninvolved parties would not not allowed. In real-life arbitration venues, only the parties present evidence—not just anyone can walk into the room and begin giving their own evidence. Allowing any random editor to post their own lengthy list of evidence is an unnecessary time sink, and contributes to the inefficiency of the process.
 * 4) Discussion. However, there would be a page on which editors could comment on the evidence and give their opinions on how that evidence should influence the outcome of the case. It would open about midway through the evidence phase, and would close a few days after the closure of the evidence phase.
 * 5) Decision. After the comments page is closed, the arbitrators post, and vote on, the proposed decision. The PD would be very straightforward and solely based on the main issues presented by the parties. In other words, ArbCom would not inject new issues that were never mentioned by the parties. It has no business doing that—its purpose should simply be to arbitrate the dispute described to them. For instance, if one party accuses the other of certain misconduct, ArbCom would vote on whether that misconduct actually occurred, but it should not look for additional misconduct on its own despite the fact that no one else ever mentioned it.

Concerning sanctions themselves, ArbCom ought to use outright site bans less liberally. When the dispute is plainly centered around specific editors or a particular article/content topic, interaction bans or topic bans should be used. Site bans should only be used if there is a broad behavioral problem (such as sockpuppetry or habitual personal attacks on any dissenting editors), or if someone intentionally and repeatedly violates lesser sanctions previously imposed on them.

Finally, all AE requests should be left open for a minimum amount of time (24-48 hours, perhaps), so that multiple admins can give their opinion on the issue presented. The purpose of this would be to facilitate more balanced enforcement by integrating more viewpoints into the process and by preventing the unilateral, knee-jerk imposition of a practically irreversible AE block. Unfortunately, under the current system, any admin can impose an immediate, irrevocable block without any discussion. No admin should have the power to do that unilaterally just because they claim "arbitration enforcement." If there is no consensus among the admins, or if admins simply neglect to discuss the request, the request would be closed with no action.

Better arbitrators
''Biblioworm is an administrator who has edited Wikipedia since 2014. He is a strong advocate for reforming Wikipedia processes; he founded the RFA2015 reform project and the new WikiProject Reforming Wikipedia. He has previously written for the Signpost.''