User:Gatoclass/CTDN

Summary

The elephant in the room

Why is a new process needed?

Existing dispute resolution processes have proven a very blunt stick indeed. They have proven to be remarkably ineffective at reducing the amount of disputation in contentious topic areas, or in improving the quality of the end product in such areas. They have proven especially ineffective at removing so-called "civil POV-pushers": users who routinely violate content policies but are careful to always remain civil, knowing that as long as their POV-pushing is not too blatant, any case resulting from it is likely to be dismissed under the current dispute resolution paradigm as a "content dispute". Meanwhile, their good faith, policy compliant opponents eventually get themselves sanctioned by losing patience and resorting to incivility, leaving the POV-pushers to continue damaging the quality and credibility of the project with even less resistance.

But while the situation may seem hopeless to the veteran editor, there has I believe been a slow evolution in interpretation of existing dispute resolution processes which offers a way forward. What has been lacking, up to now, is a clear recognition of how the community's interpretation of DR processes has evolved, along with a clearly defined process for implementing the most recent interpretations. It is this gap between existing practice at one end and (often lagging) community attitudes at the other that this process is designed to bridge, and to do so in a way that I hope will prove highly advantageous both to the community at large—in the form of less disputation—and even more importantly, to the quality of our end product.

What's wrong with Arbcom Enforcement? Is this new process intended to replace it?

Arbcom Enforcement has a number of significant flaws in my view.

Firstly, penalties are handed out on a somewhat arbitrary basis, with no clear agreement on what penalties should be applied for given infractions. AE also works on the "escalating sanctions" principle, meaning heavier sanctions for repeated infractions. In effect, this means that a user can end up being topic banned indefinitely for the most minor infraction, like say a breach of 1RR or an honest mistake (something like the "three strikes and you're out" laws applied in some jurisdictions). This process tends to work against good faith users who are heavily involved in a given topic area, because it is only human to occasionally make technical errors or indeed errors of judgement, and their detractors can usually be relied upon to seize every such opportunity to try and rid themselves of a seasoned opponent.

Secondly, in the AE model, frequent filing of cases is likely to be interpreted as WP:BATTLEGROUND behaviour warranting sanction. But a good faith user working in a contentious topic area will probably encounter numerous examples of tendentious editing; so that under the AE model he may soon find himself facing a choice of either risking an indef ban for BATTLEGROUND conduct, or of foregoing dispute resolution altogether, neither of which are desirable outcomes.

Because of the aforementioned issues, many users are so intimidated by AE (or RFAR) that they only enter it as a last resort. This means that disputes can drag along for very long periods before somebody gets sufficiently frustrated to risk making an AE request. Long drawn out disputes are not only bad for content, they also leave users feeling increasingly aggrieved and helpless, adding to overall disillusionment with wikipedia. It is imperative in my view that good faith users are given support and encouragement through access to fast and effective dispute resolution processes.

AE has some other flaws, but these are some of the main issues the new process is designed to address.

This new process is not intended, at least initially, to replace AE, since in spite of its flaws, AE currently has a broader scope. Rather, the new process is intended to complement AE by introducing a venue for dealing relatively quickly with a number of common types of disputes.

But aren't you describing an attempt to adjudicate content disputes, and isn't that disallowed?

A major part of the problem with existing dispute resolution procedures has been that disputes perceived as content-related (as opposed to conduct-related) have often been considered off-limits to existing dispute resolution processes. However, there is often a fine line between a content dispute and a conduct issue, as user misconduct is frequently manifested only (or mainly) through violation of content policies.

What is required, therefore, is an effective method of dealing with users who routinely misapply content policies as a means of—or at least with the result of—subverting those very policies. There are indeed plenty of examples of such methods being advocated or utilized under existing DR processes, but they have often gone unrecognized.

Under the existing paradigm, ARBCOM itself is designed to resolve conduct rather than content disputes. However, even ARBCOM has long recognized the legitimacy of evaluating the content of mainspace edits as a means of assessing conduct. For example, it is not a content dispute, but a conduct issue, when somebody adds a demonstrable falsehood to an article, or misrepresents a source to say something it plainly does not, and continues to do so after the error has been brought to his attention. Perhaps the clearest example of this principle can be seen in this thread, where two sitting Arbs correct an administrator labouring under this very misapprehension (a not uncommon misapprehension, I think it fair to add).

Some other examples where ARBCOM has judged given edits to be in violation of content policies:

WP:V (specifically, misuse of sources): [1][2]
WP:OR: [3][4]
WP:UNDUE (or WP:FRINGE): [5][6][7][8]

ARBCOM itself only adjudicates conduct issues, but if necessary is prepared to evaluate users' adherence to content policies when doing so, as can be seen in the above diffs. In short, it is not in fact forbidden for administrators to evaluate whether or not the content of a given mainspace edit, or set of edits, is legitimate under the existing content policies and guidelines. What has been lacking up to now is a systematic method by which admins can apply such evaluations, under a process that protects both the interests of the encyclopedia as well as the rights of individual users. It is this shortcoming in the existing policy framework, for contentious topic areas, that this proposed process seeks to address.

But wouldn't this proposed process violate wikipedia's consensus model?

The whole point about topic areas under discretionary sanctions is that they are topic areas where the normal consensus model has broken down, requiring intervention by administrators. That is why ARBCOM and AE exist. Consensus is fine in areas where reasonable people are working together collaboratively, but where topics are dominated by agenda-driven factions, there is often little if any prospect of consensus being achieved, sometimes even on the most minor issues, and certainly not within any practicable timeframe.

What is the scope of this process?

As indicated above, the process will initially cover alleged breaches of WP:V, WP:OR and WP:UNDUE. WP:NPOV itself (apart from UNDUE) will not initially be covered by this process as NPOV is broader in scope than the other policies and may prove more difficult to adjudicate. Concerns about chronic POV-pushing should instead be addressed at other venues such as WP:AE. Should this process prove effective, NPOV could be included in the above list in a trial of its own.

Typical issues handled by this process involving WP:V might include, for example, misrepresentations of source, insertion of demonstrable falsehoods into articles, or of claims unsupported by a source or only supported by dubious sources; and so on.

I am of the opinion that disputes concerning V, OR or UNDUE cover the majority of breaches in Wikipedia mainspace, though obviously there are some breaches that are not covered by these guidelines. I am open to suggestions as to what other areas of policy the process might conceivably be used to cover, either initially or at some future time.

Won't this process compromise the independence of participating administrators?

No more so IMO than under any other dispute resolution process. Administrators are already subject to frequent accusations of bias when adjudicating disputes, and will surely be subject to similar accusations under this process, but there is no more reason to treat such accusations seriously here than under other processes. What this process does include that is lacking in other processes, however, is accountability for participating admins, in that they may face bans from adjudicating cases under this process for making poor or unsound decisions.

Won't this process just increase the burden on administrators?

The burden on admins may at first increase with an initial flurry of cases, but as bans are handed out the number of new cases would be expected to drop sharply. I am also hopeful that the relatively narrow scope of this process, together with the tight restrictions on evidence, will encourage more admins to participate.

What advantages does this process have over existing processes?

Accountability for all involved parties, including adjudicating administrators.
Defined and limited sanctions for offending parties. Sanctions are large enough to discourage misconduct but not so great as to encourage sockpuppetry. Sanctions do not increase over time as it is assumed that all users are capable of making honest errors of judgement from time to time; in short this process protects good faith editors who occasionally make errors of judgement, at the same time as it provides an effective method of keeping chronic offenders out of circulation.
Eliminates the risk of a "false consensus" on content, where a temporary (perhaps manipulated) majority allows a group to trump policy through sheer weight of numbers. This is achieved through the fact that adjudications under this process are never binding on the content of the article itself; they only make a judgement as to which user has applied policy more appropriately at a given moment. The judgement may be wrong of course, but since it is not binding on content, and the penalty to be applied is limited, a wrong decision has a limited effect. The process is based on the assumption that experienced administrators will get it right more often than they get it wrong, and that over time the more responsible editors will therefore benefit to the detriment of their opponents.
Users can resort to this process an unlimited number of times without fear of being branded vexatious or of having a WP:BATTLEGROUND mentality. Users are encouraged to resort to this process as a means of breaking deadlocks over certain types of adjudicable disputes. The discouragement to vexatious use of this process is built in per the accountability section.
Because of the requirement to give offenders an opportunity to self-revert before resorting to this process, it is anticipated that many minor disputes that might otherwise snowball in long edit wars involving multiple users, can be nipped in the bud without ever coming to dispute resolution.
It is anticipated that this process will result in substantial improvement in the quality of content under discretionary sanctions, firstly because users now have a venue for getting many typical disputes resolved quickly, meaning they can move on to other improvements rather than being stuck for weeks on end trying to get a single issue resolved; secondly because agenda-driven editors should find that their typical methods of subverting content policies are no longer effective, so that they will have to either improve the quality of their editing, or else leave the project.

How do you know this process will work?

I don't. That's why it will be trialed for six months, reviewed, and possibly trialed an additional six months after that, before being adopted.

CTDN

This page describes a process for resolving certain kinds of disputes in contentious topic areas only. A contentious topic area is a topic area covered by Arbcom general or special sanctions. These are areas of the encyclopedia that in many cases have proven chronically resistant to improvement, and that have typically generated a volume of disputation quite out of proportion to their size.

If your dispute does not involve a contentious topic area, it does not belong on this page.

Policies covered by this process

Alleged breaches of the following content policies or guidelines can initially be adjudicated at this page:

WP:V
WP:OR
The following aspects of WP:NPOV:

Other aspects of the WP:NPOV policy not listed above will not initially be adjudicated under this process as they may arguably involve a greater degree of subjectivity; these other aspects may be included at a later date.

Typical issues handled by this process involving WP:V might include, for example, misrepresentations of source, insertion of demonstrable falsehoods into articles, or of claims unsupported by a source or only supported by dubious sources; and so on.

The following behavioural policy may also be adjudicated here:

WP:CIV.

Concerns involving any of the above policies should as a rule be submitted to this page rather than to AE.

The process

Opening a case

Any user account more than three months old and with more than 500 mainspace edits can bring a content dispute related to a contentious topic area to this noticeboard. The user, known as the complainant, will explain how he believes his opponent, known as the respondent, has violated a policy or aspect of a policy with a specific edit or group of edits.

Before opening a case over a given issue, the complainant must warn the potential respondent on his talk page that he intends to do so, thus giving the latter an opportunity to self-revert. If the latter refuses to do so, or if he fails to respond to the warning after 24 hours and has since resumed editing, then and only then can a case be opened.

Once a case has been opened, the respondent must be notified of the case on his talk page by the complainant. Once the respondent has been so notified, he has 48 hours to respond following his next edit on Wikipedia after the notification. That is, an edit made to Wikipedia by the respondent after notification will be regarded as proof that they have seen the notification and that they know they have 48 hours to respond. The respondent may request extra time to prepare a case if he feels it necessary, but if he fails to respond altogether per this clause, the case can proceed without him.

The complainant must supply one or more diffs, stating precisely how he believes these diffs violate policy, and which policies they violate.

The respondent may not introduce material not directly related to the issue or issues brought forward by the complainant. This is to ensure that cases remain focussed and do not become about the entire history of each user's contributions. For example, the respondent may not bring up edits by the complainant in an attempt to demonstrate the complainant's editing is "just as bad" or worse. He may not bring up his protagonist's past history of sanctions, if any (and neither may the complainant). If the respondent wants to present evidence against the complainant, he must start his own case, subject to all the other conditions outlined on this page. The respondent's evidence, in short, will be limited to defending only the diffs presented as evidence by the complainant. In cases of doubt, an adjudicating admin may decide what evidence is and is not relevant to the current case.

Adjudicating a case

Once both parties have put their side of the dispute, uninvolved administrators will decide upon the merits of the complainant's case. Any number of admins may comment on the case and any number may vote on whether or not to sanction the respondent. A minimum of two admins must vote in order to close a case; a minimum of three if the first two admins vote differently. A 60% majority is necessary to ensure that a case is upheld. A case may not be closed until 24 hours after the latest vote has been cast, to allow other admins a chance to respond to the case before it is closed. It may then be closed immediately by any administrator, including an admin who participated in the vote.

The adjudicating administrators may not decide to sanction both parties. The purpose of this restriction is to remove the temptation on administrators to apply collective punishment.

The burden of proof in this process is solidly upon the shoulders of the complainant. If the adjudicators feel that neither party is at fault, that both parties are equally at fault, or that fault cannot be determined, then the complainant and not the respondent will be sanctioned. This is to discourage frivolous cases and thus ease the burden on adjudicating admins.

At any time before the admins start voting, the case may be closed if both the complainant and respondent come to an agreement about the content under dispute.

Adjudicating administrators may only adjudicate every other dispute involving either the same issue or the same respondent or complainant. That is, if they adjudicate over a case involving a given issue, they should stand aside for the next case involving that issue, and similarly for the next case involving one or more of the named parties to the case.

Sanctions

If the complainant's case is upheld, the respondent will be topic banned for a period of three months. If the complainant's case is rejected, the complainant will be topic banned for a period of three months.

Users who violate a ban will have their ban reset. After a second violation of the same ban, the user will have their ban reset and be blocked for the duration of the ban.

In additional to the original complainant and respondent, other users may add statements to the case in support of either the complainant or respondent. However, every user who adds a supporting statement will be subject to a one month topic ban should the party they supported lose the case. This is to ensure that cases do not turn into lengthy exercises in mud-slinging - a common problem on many dispute resolution pages.

The complainant may name more than one party to the dispute if he chooses. However, for every respondent he names, he must prove the case against them or be subject to an additional sanction himself, of three additional months of ban for every additional user against whom he failed to prove his case.

A user who is a respondent in a current case, or who is currently the subject of a ban under these provisions, may not have a second case initiated against him using evidence older than the filing of the current case or the case that resulted in the current ban. Users who are currently topic banned may not initiate new cases.

In any case in which the principal complainant or respondent turns out to be an illegitimate sockpuppet, any sanction made against the other party or parties will be immediately overturned and the results of the case vacated.

Preserving neutrality

While the adjudicating administrators are empowered to decide which party is more in the right in any given dispute, their decisions will not be binding in any way upon the content of the article in question. Nor will administrators adjudicating similar cases in future be bound by precedent from past cases, although they may take note of such cases in supporting their decision. In this way, the Wikipedia commitment to NPOV will be maintained. Content can never be "set" in Wikipedia by past decisions, as consensus changes constantly.

Appeals

Any user upon whom a ban is imposed under these provisions has the right to appeal his ban to AE or Arbcom. However, if his appeal is rejected, his sanction will be increased by an additional three months. If arbcom upholds his appeal, the administrators who voted to sanction him may not participate in adjudicating of disputes under this process for a period three months. In the same way that adjudicating admins' judgements are not binding on content, judgements rendered by administrators or arbitrators in appeals lodged under this process will likewise not be binding on article content.

Overview

Contentious topic areas have long been a problem for Wikipedia. These areas tend to attract partisan supporters on one or both sides who, either subconsciously or deliberately, attempt to promote a particular point of view or stifle alternative viewpoints. Groups of likeminded editors can use their numbers to circumvent Wikipedia's content policies, or to wage edit wars with other groups, resulting in biased or substandard content. Disputes involving contentious topic areas have historically clogged up dispute resolution pages and wasted huge amounts of the community's time and energy.

Many different methods of reducing tensions or bringing order to contentious topic area pages have been attempted by the Wikipedia community, invariably with limited success. Part of the problem has been that disputes perceived as content-related (as opposed to conduct-related) have often been considered off-limits to existing dispute resolution processes, presumably (a) from fear of compromising Wikipedia's fundamental commitment to neutrality, and (b) from a reasonable concern that adjudication of content disputes would quickly overwhelm available resources. However, there is often a fine line between a content dispute and a conduct issue, as user misconduct is frequently manifested only (or mainly) through violation of content policies.

What is required, therefore, is an effective method of dealing with users who routinely misapply content policies as a means of—or at least with the result of—subverting those very policies. There are indeed plenty of examples of such methods being advocated or utilized under existing DR processes, but they have often gone unrecognized. The process described on this page seeks to formalize such tacitly recognized methods under a more comprehensive and more rigorous, but ultimately simpler and more accessible process.

Principles

With regard to concern (a) above, Wikipedia's neutrality is maintained under this process by adherence to the following principles:

An assumption that most administrators have enough experience to be able to make reasonably reliable decisions about basic matters, such as whether a source conforms to WP:V or an argument is based on original research. Such judgements are already made by the community on a daily basis, such as at WP:AFD (see also [9]); this process is essentially no more than a streamlined version of such processes, applied to chronically problematic areas;

Making all decisions made by administrators through this process non-binding on the project as a whole, to ensure that Wikipedia does not develop a "house position" on matters of content.

With regard to concern (b), availability of resources, it is assumed that the limited number of topic areas subject to ARBCOM sanctions, as well as the limited number of eligible editors active at any one time in these areas, together with the length and frequency of sanctions handed out under this process, will ensure that the process remains viable.

A set of defined and limited sanctions for breaches of content policy as adjudicated here, allowing for predictability and ensuring that users are not unduly penalized by rulings which may inevitably include a subjective element.

Goals

The first goal of this process is to prevent small disputes snowballing into larger ones by providing users involved in contentious topic areas a means of resolving common types of dispute quickly and easily, and without risking the kind of draconian sanctions that can sometimes result from resort to other dispute resolution processes. A secondary goal is to cut down on the number of cases brought to processes like WP:AE which can chew up a great deal of all participants' time.

The assumption behind this process is that it will provide a means of weeding out users with chronically problematic editing habits from contentious topic areas, allowing good faith users who generally edit according to policy to stabilize and improve articles. It is also anticipated that the process will result in reduced tensions and a more collaborative editing environment as the more responsible editors can work together without the usual level of disruption.

Self-policing

Because respondents to this process are required to be given an opportunity to self-revert their disputed edit or edits before a case here can proceed, it is anticipated that over time, users will learn to judge their edits more objectively and thus to self-correct when challenged—an outcome that should considerably reduce the need to resort to DR processes in the first place.

Accountability

A central principle of this process is that all participants will be held accountable for their actions by being subject to potential sanctions. Not only are the named parties to the case potentially subject to sanction, but so is anyone who chooses to add evidence to the page. Even the adjudicating admins are subject to potential sanctions in that they will be barred for a period from participating in the process as adjudicators if they are found on appeal to have made an unreliable decision. Moreoever, there are potential consequences for every action taken under the process, including appeals.

Note however that the risk of sanction under this process is voluntary for all parties concerned, since no-one is compelled to initiate a case, defend a case, add evidence to a case, or appeal a case. The only obligation involved in any part of the process is on the part of the user initially accused of violating a policy, who is given the choice of either reverting his contested edit if he thinks it dubious, or agreeing to have the case adjudicated if he believes it to be defensible.

Policies covered by this process

Alleged breaches of the following policies or guidelines can initially be adjudicated at this page:

WP:NPOV itself (apart from UNDUE) will not initially be covered by this process as NPOV is broader in scope than the other policies and may prove more difficult to adjudicate. Concerns about chronic POV-pushing should instead be addressed at other venues such as WP:AE. Should this process prove effective, NPOV could be included in the above list in a trial of its own.

Typical issues handled by this process involving WP:V might include, for example, misrepresentations of source, insertion of demonstrable falsehoods into articles, or of claims unsupported by a source or only supported by dubious sources; and so on.

The process

Opening a case

Any user with more than 500 mainspace edits can bring a content dispute related to a contentious topic area to this noticeboard. The user, known as the complainant, will explain how he believes his opponent, known as the respondent, has violated a content policy with a specific edit or group of edits.

Before opening a case over a given issue, the complainant must warn the potential respondent on his talk page that he intends to do so, thus giving the latter an opportunity to self-revert. If the latter refuses to do so, or if he fails to respond to the warning after 24 hours and has since resumed editing, then and only then can a case be opened.

Once a case has been opened, the respondent must be notified of the case on his talk page by the complainant, and the respondent given at least 48 hours after the notification to make a statement of his own. The respondent may request extra time to prepare a case if he feels it necessary, but if he fails to respond altogether within 48 hours, the case can proceed without him.

The complainant must supply one or more diffs, stating precisely how he believes these diffs violate policy, and which policies they violate.

The respondent may not introduce material not directly related to the issue or issues brought forward by the complainant. This is to ensure that cases remain focussed and do not become about the entire history of each user's contributions. For example, the respondent may not bring up edits by the complainant in an attempt to demonstrate the complainant's editing is "just as bad" or worse. He may not bring up his protagonist's past history of sanctions, if any (and neither may the complainant). If the respondent wants to present evidence against the complainant, he must start his own case, subject to all the other conditions outlined on this page. The respondent's evidence, in short, will be limited to defending only the diffs presented as evidence by the complainant. In cases of doubt, an adjudicating admin may decide what evidence is and is not relevant to the current case.

Adjudicating a case

Once both parties have put their side of the dispute, uninvolved administrators will decide upon the merits of the complainant's case. Any number of admins may comment on the case and any number may vote on whether or not to sanction the respondent. A minimum of two admins must vote in order to close a case; a minimum of three if the first two admins vote differently. A 60% majority is necessary to ensure that a case is upheld. A case may not be closed until 24 hours after the latest vote has been cast, to allow other admins a chance to respond to the case before it is closed.

The adjudicating administrators may not decide to sanction both parties. The purpose of this restriction is to remove the temptation on administrators to apply collective punishment.

The burden of proof in this process is solidly upon the shoulders of the party bringing the case. If the adjudicators feel that neither party is at fault, that both parties are equally at fault, or that fault cannot be determined, then the complainant and not the respondent will be sanctioned. This is to discourage frivolous cases and thus ease the burden on adjudicating admins.

At any time before the admins start voting, the case may be closed if both the complainant and respondent come to an agreement about the content under dispute.

Adjudicating administrators may only adjudicate every other dispute involving either the same issue or the same respondent or complainant. That is, if they adjudicate over a case involving a given issue, they should stand aside for the next case involving that issue, and similarly for the next case involving one or more of the named parties to the case.

Sanctions

If the complainant's case is upheld, the respondent will be topic banned for a period of twelve weeks. If the complainant's case is rejected, the complainant will be topic banned for a period of eight weeks.

Users who violate a ban will have their ban reset and be blocked for the original duration of the ban.

In additional to the original complainant and respondent, other users may add statements to the case in support of either the complainant or respondent. However, every user who adds a supporting statement will be subject to a six week topic ban should the party supported lose the case. This is to ensure that cases do not turn into lengthy exercises in mud-slinging - a common problem on many dispute resolution pages.

The complainant may name more than one party to the dispute if he chooses. However, for every respondent he names, he must prove the case against them or be subject to an additional sanction himself, of eight additional weeks of ban for every additional user against whom he failed to prove his case.

A user who is a respondent in a current case, or who is currently the subject of a ban under these provisions, may not have a second case initiated against him using evidence older than the filing of the current case or the case that resulted in the current ban. Users who are currently topic banned may not initiate new cases.

Preserving neutrality

While the adjudicating administrators are empowered to decide which party is more in the right in any given dispute, their decisions will not be binding in any way upon the content of the article in question. Nor will administrators adjudicating similar cases in future be bound by precedent from past cases, although they may take note of such cases in supporting their decision. In this way, the Wikipedia commitment to NPOV will be maintained. Content can never be "set" in Wikipedia by past decisions, as consensus changes constantly.

Appeals

Any user upon whom a ban is imposed under these provisions has the right to appeal his ban to AE or Arbcom. However, if arbcom rejects his appeal, his sanction will be increased by an additional twelve weeks. If arbcom upholds his appeal, the administrators who voted to sanction him may not participate in adjudicating of disputes under these provisions for a period of 26 weeks. In the same way that adjudicating admins' judgements are not binding on content, judgements rendered by arbitrators in appeals lodged under this process will likewise not be binding on article content.