Wikipedia:Bot Approvals Group/Draft consensus statement on date delinking

Background
For several years there has existed a disagreement between editors regarding when dates should be linked (e.g. "January 12, 2009") and when they should not (e.g. "January 12, 2009"). This stylistic dispute has become extremely personal to many editors, leading to countless reverts, blocks, incidents of incivility, RFCs, polls, and an arbitration case. Relevant links to background information are at Manual of Style (dates and numbers) and Linking.

In August on 2008, a straw poll at Wikipedia talk:Manual of Style (dates and numbers)/Archive D6 led datelinking to be considered deprecated by the community. In June, August, and September of 2008, three different BAG members approved three different bot tasks to automate date formatting changes: Bots/Requests for approval/Lightbot, Bots/Requests for approval/Lightbot 2, Bots/Requests for approval/Lightbot 3. In October, another related bot task was denied: Bots/Requests for approval/Cleanbot. Meanwhile, a semi-automated user script was designed to assist users manually delinking dates in a manner that would not require BAG approval: User:Lightmouse/monobook.js/script.js. Some editors used this script cautiously, and others did so carelessly. Tempers flared.

On January 12, 2009, an arbitration case was opened regarding the actions of users involved in this controversy: Requests for arbitration/Date delinking. A temporary injunction by the arbitration committee states: "Until this case is decided or otherwise directed by the Arbitration Committee, all editors are instructed not to engage in any program of mass linking or delinking of dates in existing articles, including but not limited to through the use of bots, scripts, tools, or otherwise." This has been in place for four months. A proposed decision mentions the BAG in a number of its findings, and it is likely that decisions about, and instructions to, the Bot Approval Group will be a part of the final decision.

The following is a draft statement of the Bot Approvals Group on date delinking. It is my hope that the BAG will be able to agree on a set of principles to help us deal with the issue as it arises, and to assist the Arbitration Committee in their formulation and implementation of their decisions. – Quadell (talk) 17:25, 14 May 2009 (UTC)

1. We recognize that decisions by the Arbcom are binding
The Bot Approvals Group recognizes that decisions made by the Arbitration Committee are binding, and we will endeavor to follow their directions to the best of our ability.


 * Comments by BAG members:
 * Proposed and support. – Quadell (talk) 17:25, 14 May 2009 (UTC)
 * Well, this is clearly true - does it need spelling out? Fritzpoll (talk) 17:41, 14 May 2009 (UTC)
 * I think it would be helpful. Sometimes, when creating a consensus statement, you have to state the obvious. – Quadell (talk) 19:35, 14 May 2009 (UTC)
 * Support Clearly true, and I've observed that ArbCom likes statements of the obvious. Anomie⚔ 20:19, 14 May 2009 (UTC)
 * Support. No more need be said. - Jarry1250 (t, c) 08:51, 17 May 2009 (UTC)
 * Abstain Per recusal.  MBisanz  talk 12:52, 18 May 2009 (UTC)
 * Support Stating the obvious. Q  T C 13:03, 18 May 2009 (UTC)
 * Oppose&mdash;In theory yes, but in this case, ARBCOM is assigning powers to BAG for which there is no community consensus to do so. We are an advisory panel, and I don't believe we have ever forced for users to have bots with a public source code.  Maxim (talk)  17:40, 18 May 2009 (UTC)
 * I don't see that Requests for arbitration/Date delinking/Proposed decision is giving BAG any extra power; they're actually taking a little bit away by instructing us that for that particular task we may not approve a bot without published source. Anomie⚔ 21:16, 18 May 2009 (UTC)


 * Comments by others:

2. We understand the relative importance of date formatting
Date formatting is not terribly important in the grand scheme of things. The creation and improvement of quality articles, adequate sourcing of statements, encouraging civil community, and removal of bias, are all incomparably more important than a stylistic issue. However, that does not imply that the issue is entirely unimportant; it is valuable to encourage consistent formatting in accordance with our Manual of Style and consensus.


 * Comments by BAG members:
 * Proposed and support. – Quadell (talk) 17:25, 14 May 2009 (UTC)
 * Support Clearly true, and I've observed that ArbCom likes statements of the obvious. Anomie⚔ 20:19, 14 May 2009 (UTC)
 * Support. No more need be said. - Jarry1250 (t, c) 08:51, 17 May 2009 (UTC)
 * Abstain Per recusal.  MBisanz  talk 12:52, 18 May 2009 (UTC)
 * Oppose&mdash;I don't see why we need this: we approve bots, not handle article directly. I, for one, do think that internal linking is a very useful—in fact, a useful part of articles—and not merely a "stylistic issue".  Maxim (talk)  17:40, 18 May 2009 (UTC
 * Given that this particular part of the MoS has no consensus right now, I'm not sure how relevant/accurate this is. Mr.Z-man 20:09, 19 May 2009 (UTC)
 * Comments by others:
 * I have to agree with Maxim. Date linking was the impetus for the backlash against autoformatting in the first place. I think that if had simply been about 19 May against May 19 (or even 2009-05-19), there would probably be no Arbitration over the issue. Simply put, "[l]inking is one of the most important features of Wikipedia", (taken from Linking) and the issue is more important than some make it out to be. By no means, however, am I considering it to be anywhere as important our five pillars. Dabomb87 (talk) 02:41, 19 May 2009 (UTC)

3. We recognize the need for good judgement
The approvals for automated tasks in relation to this issue were made in good faith, based on the information available at the time. However, subsequent events have helped inform the BAG about the issue, and we will strive to be very careful to exercise good judgement when considering bot tasks related to this issue.


 * Comments by BAG members:
 * Proposed and support. – Quadell (talk) 17:25, 14 May 2009 (UTC)
 * I guess the question is, to what extent do we check the community view of tasks being performed, or are we purely technical functionaries? Is the responsibility shared between bureaucrats and BAG in this respect that clear in practice?  Fritzpoll (talk) 17:44, 14 May 2009 (UTC)
 * These are valid questions. But I think that either way, we can strive to excise good judgment on this issue. – Quadell (talk) 19:35, 14 May 2009 (UTC)
 * Support I wasn't in BAG at that time, but I've taken the lesson to heart anyway. Whenever someone requests something that looks like it could be at all controverisal for a bot to do, I slap on a . OTOH, when a task being done by humans could be better done by bot and someone tries to claim it's controverisal, I tell them to take it up elsewhere (as with the recent R from other capitalisation complaint on WP:BON). Anomie⚔ 20:19, 14 May 2009 (UTC)
 * Support. Even more so than Anomie, I wasn't in the BAG at the time, but as a core fundamental of the BAG, there is no way this can be neglected. - Jarry1250 (t, c) 08:51, 17 May 2009 (UTC)
 * Abstain Per recusal.  MBisanz  talk 12:52, 18 May 2009 (UTC)
 * Support I'd have to check to see if I was active or not during that time, but we've always said if issues come up later with requests we like to know about it. Q  T C 13:06, 18 May 2009 (UTC)
 * Oppose&mdash;I don't think we should bear the brunt of the problem: the community can reach consensus that a task is not beneficial/operator is not trustworthy, and remove their privilege/ability to run a bot, as happened with the BetacommandBot sagas.  Maxim (talk)  17:40, 18 May 2009 (UTC)
 * Comments by others:

4. We understand what is expected of us
We agree with the following Arbcom statement: "Members of the Bot Approvals Group are trusted members of the community and are expected to follow Wikipedia policies. They are expected to pursue their duties to the best of their abilities. Occasional mistakes are entirely compatible with this; members are not expected to be perfect. However, consistently or egregiously poor judgment may result in the removal of BAG status."


 * Comments by BAG members:
 * Proposed and support. – Quadell (talk) 17:25, 14 May 2009 (UTC)
 * Support Clearly true, and I've observed that ArbCom likes statements of the obvious. Anomie⚔ 20:19, 14 May 2009 (UTC)
 * Support. No more need be said. - Jarry1250 (t, c) 08:51, 17 May 2009 (UTC)
 * Abstain Per recusal.  MBisanz  talk 12:52, 18 May 2009 (UTC)
 * Comment Without opening the can of worms that is the BAG appointing process, I'd agree with the idea of this statement. Along with the previous finding, hindsight is 20/20 and consensus does change. Also agree with Maxim that clarification of removal needs to be made. Q  T C 19:46, 18 May 2009 (UTC)
 * Oppose&mdash;I don't agree with the last sentence: it's vague as to whether ARBCOM is implying it will remove someone from BAG, or the community will? No precedent has been set: the only forcible removal from BAG other than the end of the trail period was Betacommand, but someone who can't run a bot can't approve them, so that's wouldn't be poor judgment during the course of BAG duties that led to the removal.  Maxim (talk)  17:40, 18 May 2009 (UTC)
 * Comments by others:

5. We recognize style guidelines
Automated tasks that change stylistic issues in articles must conform to our style guidelines. Bot tasks which would change articles in contravention of our guidelines should not be approved.


 * Comments by BAG members:
 * Proposed and support. – Quadell (talk) 17:25, 14 May 2009 (UTC)
 * Support Clearly true. We even tend to avoid the ones where the style guideline is not absolute in its statement. Anomie⚔ 20:19, 14 May 2009 (UTC)
 * Support. No more need be said. - Jarry1250 (t, c) 08:51, 17 May 2009 (UTC)
 * Abstain Per recusal.  MBisanz  talk 12:52, 18 May 2009 (UTC)
 * Support, although I question the need for this, as this goes without saying that bot tasks must adhere to community consensus, which is partially conveyed through guidelines, policies, etc.  Maxim (talk)  17:40, 18 May 2009 (UTC)
 * Support, All bots should adhere to the current guidelines, so it should go without saying that approval requires following current consensus. Q  T C 19:47, 18 May 2009 (UTC)
 * Comments by others:

6. We recognize the sensitivity of the issue
Due to the history of this issue and the feelings involved, any bot that unlinks dates must be held to a very high standard. It must be carefully constructed so as not to unlink relevant dates that are linked in compliance with our guidelines. The sourcecode must be available, it must not editwar, and it must respect the nobots template. The bot operator must be civil and courteous, and must be willing to pause the bot and discuss problems if they arise. There are some people who will object to the bot and oppose its operation no matter how scrupulously the operator behaves, and that is unavoidable, but the operator must do his or her best to remain helpful and civil regardless.


 * Comments by BAG members:
 * Proposed and support. – Quadell (talk) 17:25, 14 May 2009 (UTC)
 * Support Clearly true. And it'll probably take forever to get approved anyway, as I'm sure the various Date Warriors will have to fight out exactly which dates the bot would touch and which it wouldn't. Anomie⚔ 20:19, 14 May 2009 (UTC)
 * Conditional Support. Yes. As Anomie said, ArbCom is likely to lead the way when specifying what needs doing, but the above remains true. Clearly, we do things by consensus not by dictat. - Jarry1250 (t, c) 20:19, 19 May 2009 (UTC)
 * Earlier statement created at 08:51, 17 May 2009 (UTC); above is a lightly modified version.
 * Abstain Per recusal.  MBisanz  talk 12:52, 18 May 2009 (UTC)
 * Conditional support&mdash;I assume this is a common sense thing that a bot editing an issue as controversial as this must be up to a high standard, but sourcecode and nobots thing should probably be discussed and agreed internally and with the community to outside of this somewhat quick statement.  Maxim (talk)  17:40, 18 May 2009 (UTC)
 * Oppose - not bot should be delinking dates until MOSNUM is stable on a community-approved version, then we can talk terms for a bot. Mr.Z-man 20:13, 19 May 2009 (UTC)
 * Comments by others:
 * Perhaps a middle ground is possible when it comes to disclosing source code. I understand that for Lightbot, just one portion of the program runs to 15 pages of A4 paper. It may not be always wise to throw this open to the public, but selective disclosure – perhaps to a subcommittee of BAG and arbitrators – could be a workable solution. --Goodmorningworld (talk) 18:14, 18 May 2009 (UTC)
 * More: Some time ago, I recommended to Lightmouse that he lay open his source code for Lightbot. My thinking then was, if the community could see the extreme lengths to which Lightmouse goes to incorporate criticism and suggestions by programming exceptions into the code, it would stop short some of the unfounded attacks on him. (I have made no secret of my opinion that the heavy attacks on Lightmouse and especially the lying equation of his conduct with Betacommand's have been simply disgraceful.) Lightmouse did not act on my suggestion, and he may have been right. To some of Wikipedia's nattering nabobs of negativism, every defense offered by one of their targets is immediately snapped up and turned directly into fodder for yet more attacks.
 * For some functions (e.g. anti-vandalism), it may be productive not to release source code. In the majority of cases, it won't matter.  For some, however, the function's sensitivity (in terms of controversy among the community) makes it sensible to publish source code, so it is abundantly clear what the bot will do.  Bots doing controversial tasks should be exceptionally well behaved, the task should be exceptionally well explained and the documentation should be exceptionally thorough.  The bot operator should preferably be an experienced bot operator and, ideally, one without an axe to grind in the dispute in question.   [[Sam Korn ]] (smoddy) 15:12, 19 May 2009 (UTC)

ARBCOM proposal is unworkable
An issue such as date delinking will not only need BAG approval, but direct general community consensus. We feel that ARBCOM is forcing BAG's hand somewhat, and setting numerous precedents that may not be necessarily healthy in the long run, such an open source code requirement and vague methods for removal of BAG membership. In conclusion, the question of a date delinking bot should be remanded to general community, as well as the entire question of the MOS guideline, and not merely to BAG under conditions.


 * Comments by BAG members:
 * Proposed and support.  Maxim (talk)  17:40, 18 May 2009 (UTC)
 * Oppose. I believe the ArbCom proposal is workable, regardless of whether I agree with it. – Quadell (talk) 18:06, 18 May 2009 (UTC)
 * Abstain Per recusal.  MBisanz  talk 12:52, 18 May 2009 (UTC)
 * Comment I agree with a few points above. While I agree that the source would be nice, from a community point of view, implementation is less important then the logic behind it.  While open source does not induce an undue burden to operators, I feel knowing Why is more important then How.  The next point is your comment about remanding to the community.  BAG has always worked under the auspices of the community, we're already supposed to take consensus into account.  Q  T C 19:56, 18 May 2009 (UTC)
 * Oppose The bot approvals process is supposed to include both the technical and community aspects: BAG approval should not be given without both. Some in the community seem to have the impression that BAG considers only the technical aspects and ignores the question of community consensus; I don't know if that was actually true in the past, but for my part I'm trying to make sure it's not true now. As for the "vague methods for removal of BAG membership", it sounds to me like the same vague methods we have for desysoping an admin: ArbCom, or overwhelming community consensus. I suspect the intent of the source requirement is so that there is no ambiguity over just what the date-delinking bot is being approved to do. Anomie⚔ 21:41, 18 May 2009 (UTC)


 * Comments by others:
 * Support. Moreover, "direct general community consensus" already exists and has existed since at least December 2008, when the community said, dates should be linked rarely if ever (92.8%), we don't care for date autoformatting (83.6%), and we're fine with having a duly approved bot implement the will of the community (80%). --Goodmorningworld (talk) 18:02, 18 May 2009 (UTC)