Wikipedia:Bots/Requests for approval/VWBot 9

The following discussion is an archived debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA. The result of the discussion was

Withdrawn by operator.

VWBot 9

Operator: VernoWhitney (talk · contribs)

Automatic or Manually assisted: Automatic

Programming language(s): Python

Source code available: Not written yet

Function overview: Mass rollback of all articles a user's contributed to

Links to relevant discussions (where appropriate): Wikipedia:Administrators' noticeboard/Incidents/CCI (intermittently, I found comments regarding it in at least the "Implementing bot?", "Questions", and "A running count of progress please" sections) and User talk:Moonriddengirl/Archive 26#CCI tools.

Edit period(s): Occasional, as needed

Estimated number of pages affected: Many

Exclusion compliant (Y/N): N

Already has a bot flag (Y/N): Y

Function details: This will identify the earliest edits which meet some threshold (such as those generally used at WP:CCI of increasing an article's size by 100 bytes and excluding those edits which are likely reversions) made by a particular contributor to all articles. It will then roll back the articles to the version immediately prior to the contributor's first substantial edit and leave an appropriate message on the article's talk page (probably based upon {{CCId}}).

Since rolling back all of a known copyright violators touched articles has been mentioned (even in the Signpost), I figured it would be a good idea to have a bot ready in case there is support for such an action. This would be for use in the same situations where a sufficient amount of an editor's contributions have been determined to be copyvios that Special:Nuke is used for their created articles. VernoWhitney (talk) 20:20, 22 September 2010 (UTC)[reply]

Discussion

Proposed bot is a follow-on to Uncle G's article blanking operation approved a few days ago and which has done a preliminary run (I think about 10% of the task). I believe we're waiting for user experience and feedback from Uncle G's preliminary run before going on with the other 90%. This follow-on has been discussed at CCI and seems generally supported by the people engaged with such details, barring possible surprises from Uncle G's operation. This is a good time to be developing and testing VW's bot, but IMO deployment shouldn't begin until we've gotten some more experience (at least a week's worth, say) from the results of the first operation. I'm guessing it will take that long to get all details of VW's bot ironed out anyway. The total # of articles to be rolled back is presumed to be around 13,000. General overview of the surrounding issue is at:

Wikipedia:Contributor_copyright_investigations/Darius_Dhlomo#Quick_overview - overview

Wikipedia:Contributor copyright investigations/Darius Dhlomo/Task explanation - bot task description.

71.141.90.138 (talk) 00:32, 23 September 2010 (UTC)[reply]

Just to be clear: the first execution of this task may be for the Darius Dhlomo CCI if there is solid support for it, but I also wish it to be a possible tool for other CCIs should the need arise. VernoWhitney (talk) 00:37, 23 September 2010 (UTC)[reply]

{{BAGAssistanceNeeded}} Is anyone out there? Judging from the progress so far it won't be needed for this particular CCI, but I still think it would be handy to have this tool available, so any feedback whatsoever would be appreciated. VernoWhitney (talk) 19:05, 8 October 2010 (UTC)[reply]

Assuming any runs are prior agreed upon from CCI, there should not be any major problems with. But it should be agreed what's the first date, change threshold, what talk page template to use, etc. Also, what constitutes a reversion? Edit summary with script tag or phrases like rv/revert/undo? Finally, I think it would be best if the bot could make a list of all proposed reversions, and outline borderline cases for manual review. DD case is huge, and regular cases aren't that big as to taking too much time to properly review. — HELLKNOWZ ▎TALK 23:16, 14 October 2010 (UTC)[reply]

Yes, it would only be run if there was consensus for it at CCI. I imagine the "default" settings would be for edits from any date that added more than 100 bytes of content, since those are the standards for listing edits for human CCI review, but those of course could be set differently for any given run. I erred when mentioning the talk page template earlier: the talk page template would be based on {{CCI}}, but additionally include at least the fact that it was done automatically, and a link to the particular version immediately prior to the contributor's first edit which meets whatever threshold has been set up for the run. By reversion I mean simply replacing the current content of the article with that of an earlier version.

The whole point of the bot is to avoid going through proposed reversions and borderline cases, because it is only to be used when so many of a contributor's edits that the collateral damage is acceptable (again, akin to Special:Nuke). A likely case for the use of this is the nigh-inevitable return of Siddiqui (talk · contribs)—the sockmaster behind both Wikipedia:Contributor copyright investigations/Paknur and Wikipedia:Contributor copyright investigations/AlphaGamma1991. While DD's CCI is huge, it is (I'm fairly certain) not our largest and it's only one of the 40 cases open right now. VernoWhitney (talk) 00:21, 15 October 2010 (UTC)[reply]

Approved for trial. Please provide a link to the relevant contributions and/or diffs when the trial is complete. Maybe 3 - 5 users worth of rollback for a trial. MBisanz ^talk 22:31, 19 October 2010 (UTC)[reply]

{{OperatorAssistanceNeeded}} what is the current status of this request? ΔT _{^{The only constant}} 01:02, 11 November 2010 (UTC)[reply]

I asked about an appropriate editor/target for this trial at WT:CCI and didn't get a response - and then promptly forgot about it with working up to my RFA. I'll ask for some more attention and see if any of the current CCIs are good candidates. VernoWhitney (talk) 01:11, 11 November 2010 (UTC)[reply]

Trial complete. First user trial completed. Feedback continuing at Wikipedia talk:Contributor copyright investigations/Archive 1#Rollback bot. VernoWhitney (talk) 14:09, 12 November 2010 (UTC)[reply]

{{OperatorAssistanceNeeded}} I could wade through all the various threads of the discussion, or I could just ask you whether there were any problems :) How's it looking? Still needed? - Jarry1250 ^{[Who? Discuss.]} 18:08, 3 December 2010 (UTC)[reply]

The only problem that showed up was one unnecessary edit to an article's talk page when the article had already been reverted by another editor, and I've since added a check for that. Other than that the conversation included talk about tweaking the edit summaries and the message the bot uses, but there hasn't been another clear occasion to do another test run yet (since MBisanz said 3-5 users). There are thankfully few cases where all of an editor's contribs are not worth checking, but I think it is still needed at least for the same situation as I ran the first test on: a contributor indef-blocked for copyvios who keeps returning as a new sock adding more (and mostly) copyvios. VernoWhitney (talk) 19:43, 3 December 2010 (UTC)[reply]

Judging by the comments at User talk:VernoWhitney, the bot is creating an over abundance of unproductive edits that may or may not be related to CopyVio problems. This is expecting a number of other users to keep tabs on a bots edits and a very large number are being reverted. This is not the purpose of bots. Bots should be making uncontroversial edits that, except in rare cases, don't need oversight by real users. I am very appalled by the approach this bot is taking to editing Wikipedia and would like an immediate halt of it's use and a very big rethinking of the purpose and process by which the bot makes edits. A bot that creates more work for users is not a prodoctive bot, but instead a vandal! Sadads (talk) 18:35, 10 December 2010 (UTC)[reply]

This is not a task which can produce edits that do not require any user attention. There is practically no chance that all articles edited have been just copyvio and the editor remained only contributor. That said, there does need to be consensus to use a tool to make such edits. — HELLKNOWZ ▎TALK 13:26, 19 December 2010 (UTC)[reply]

Needs wider discussion.. Seeing how this has/will get stalled, I see BAG's reluctance to take action for a task that can be applied for more than one case. There is support for using such a tool to revert copyvios; but there seems to be little open, direct support for the results produced by this particular implementation. Although, as I pointed above, it would be near impossible to create a perfect tool that would not require user attention. The question is whether the community supports the current implementation. I suggest you start a broader discussion referring to the actual trial edits and make a straight point: "Does the community want this kind of output from this kind of task?". Of course, it's all up to you, but at least then the BAG can refer to this a "consensus for the task", because at present this will probably not get blanket approved for copyvio cases. — HELLKNOWZ ▎TALK 13:26, 19 December 2010 (UTC)[reply]

Obviously there's not consensus for the current implementation (I can provide links to the discussions if you'd like), but what about continuing trial once I've gone through the code to incorporate the feedback that I got from the aborted second trial and reduce the false positive rate? VernoWhitney (talk) 14:43, 19 December 2010 (UTC)[reply]

Approved for extended trial (30–50 edits and/or 1–2 users). Please provide a link to the relevant contributions and/or diffs when the trial is complete. OK, let's do a run with the feedback incorporated. However, following that, the "wider discussion" and community response will be necessary if the task is to be further trialed/approved. Do you have any links to Uncle G bot's post-run feedback? — HELLKNOWZ ▎TALK 16:06, 19 December 2010 (UTC)[reply]

Has this trial been done? Mr.Z-man 04:16, 2 January 2011 (UTC)[reply]

Not yet, the holiday has delayed the coding needed. VernoWhitney (talk) 12:42, 3 January 2011 (UTC)[reply]

{{OperatorAssistanceNeeded}} Any progress? Anomie ⚔ 03:27, 24 February 2011 (UTC)[reply]

Some progress, but not enough to address all of the issues which people had with the first run. Since it looks like it will be a while before I can finish coding you can consider this withdrawn for now and I'll just reopen it after I have the time I need to put into it. VernoWhitney (talk) 12:59, 24 February 2011 (UTC)[reply]

Ok,

Withdrawn by operator.. Just undo this edit, add any necessary comment, and relist it when you're ready. Anomie ⚔ 00:08, 25 February 2011 (UTC)[reply]

The above discussion is preserved as an archive of the debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA.