Wikipedia:Bots/Requests for approval/DannyS712 bot 40


 * 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

DannyS712 bot 40
Operator:

Time filed: 22:09, Saturday, May 18, 2019 (UTC)

Automatic, Supervised, or Manual: manually triggered on bot contributions, automatic rollback

Programming language(s): Javascript

Source code available: User:DannyS712 test/massRollback.js

Function overview: Rollback the edits of malfunctioning bots, either mine or others upon request

Links to relevant discussions (where appropriate): Administrators%27 noticeboard

Edit period(s): As needed

Estimated number of pages affected: Thousands

Exclusion compliant (Yes/No): n/a

Already has a bot flag (Yes/No): No

Function details: Bots are not perfect, and sometimes during trials or after approval it is useful to revert a certain series of edits. I'll make another account for this task, since it'll require rollback rights. The benefits of using a bot to revert such changes are fairly straightforward. A bot wouldn't show up in recent changes and flood the queue, unlike normal rollbackers; since the original changes were bot edits that didn't show up, it stands to reason that reverting them also doesn't need to be seen. Furthermore, it would remove the ratelimit that I face when using rollback; it is far faster to rollback edits than undo them, but rollback only works until another editor edits the page, so being able to revert the edits all at once (or in batches of 1000, rather than 90 per minute) would reduce the likelyhood of needing to manually undo such edits. Previous examples of me rolling back mass edits:, (both my bot),  (a different bot, per request by operator)

Discussion

 * BRFA filed --DannyS712 (talk) 22:09, 18 May 2019 (UTC)
 * BotSpeedy Trusted bot op. &#32; Headbomb {t · c · p · b} 22:10, 18 May 2019 (UTC)
 * are you sure? Thousands of pages being rolled back? --DannyS712 (talk) 22:11, 18 May 2019 (UTC)
 * Unless you're planning on being an idiot with this, I don't see the issue. &#32; Headbomb {t · c · p · b} 22:13, 18 May 2019 (UTC)
 * Nope, no plans on being an idiot. Can you please get flagged with both   and   rights? Thanks so much, --DannyS712 (talk) 22:17, 18 May 2019 (UTC)
 * , this is one for you. &#32; Headbomb {t · c · p · b} 22:19, 18 May 2019 (UTC)
 * - so I disagree with your speedy approval here. There has been no real discussion, the I need a bot to rollback all of my own bad future edits use case is shaky at best. Not sure why this would need a separate account at all - the same operator is in control and already has a bot account that the rollback flag could just be added to. Also, no testing, not even any mention of what software will be used. Also, how will use cases such as: BotATask1 makes a good edit, then BotATask2 makes a "bad edit", then this wants to be used - if it is using 'rollback' it will void the good edit at the same time. —  xaosflux  Talk 22:36, 18 May 2019 (UTC)
 * And was this even suggesting an edit rate of thousands per minute?? — xaosflux  Talk 22:38, 18 May 2019 (UTC)
 * I made another account because most of my tasks don't need rollback rights, and out of an abundance of caution I'd rather avoid having my primary bot account be able to rollback 1000 edits in one click (the software is a modified version of User:Writ Keeper/Scripts/massRollback.js, located at User:DannyS712 test/massRollback.js) --DannyS712 (talk) 22:40, 18 May 2019 (UTC)
 * Additionally, and I am not a BAG member, but it would seem inappropriate to speedy-approve something that was ostensibly tied to a request of your own, . ~ Amory  (u • t • c) 00:26, 19 May 2019 (UTC)
 * I'm sure I'm out of the loop, but why did you need to rollback over 1000 edits made by your own bot less than two months ago? ~ Amory  (u • t • c) 00:50, 19 May 2019 (UTC)
 * you're not out of the loop - this is the first time I've been asked. In that case, I was dealing with pages that had a category twice by removing one. However, for ~1000 pages the bot removed the category from a redirect to the category. See Wikipedia talk:Categories for discussion for my attempt to understand them, but the pages were not actually double categorized, because 1 instance of the category was a redirect. I hadn't encountered this in trial, so I didn't account for it in my regex until I noticed it, at which point I went back and fixed the edits. --DannyS712 (talk) 00:53, 19 May 2019 (UTC)
 * Here are some stats on rolling back bot edits: query/35255 --DannyS712 (talk) 16:34, 20 May 2019 (UTC)
 * I am not sure that this is needed. What is the benefit of doing it this way as opposed to having a sysop rollback the edits with a script? I worry that this task would lead to reduced scrutiny over bot malfunctions. WJBscribe (talk) 11:06, 21 May 2019 (UTC)
 * It would just be easier for me, since it would mean that I wouldn't need to get an admin's help. I can assure you that this would not lead to reduced scrutiny - it would be used to supplement such scrutiny by making fixing mistakes caught by the scrutiny easier to fix --DannyS712 (talk) 16:22, 21 May 2019 (UTC)
 * But at the moment, the fact that a bot has made mistakes requiring mass rollback would come to the attention of one or more admins. They could then consider whether to block the bot and/or request that BAG review the approval of the task (or indeed, in an extreme case, the suitability of the bot operator to run a bot on enwiki at all). If this task were approved, the bot operator of the faulty bot would come to you instead. Your bot would then quickly and quietly clean up the mess, and the underlying problem might never get any wider attention. WJBscribe (talk) 16:28, 21 May 2019 (UTC)
 * That's rather a misconception of how the bot policy, bot approval process, or blocking policy works. Bot blocks are issued to prevent damage from malfunctioning bots. Here the damage was already done and stopped. No block needed. Bot approval is not in need to be revoked, because the bot hasn't loss consensus to do the edits it should be doing. What happened was a configuration mistake (i.e. a bug), which has since been fixed. I could have reverted it all myself (the onus is after all, on the bot operator to fix bot's edit), but that's not required by policy, or even desired. Simply that the damage is undone. These situation happen every now and then, and when they happen, it's really irrelevant if things get undone by an admin, a twinkle user, a script user, or a bot. Doing it by bot, is, however, the most efficient and least obstrusive method of mass undoing edits, especially if undoing bot edits. &#32; Headbomb {t · c · p · b} 16:43, 21 May 2019 (UTC)
 * Actually, no. You are correct that "Bot blocks are issued to prevent damage from malfunctioning bots" and it might well be that no block or further action is needed in most cases. But that still needs to be evaluated. If the conclusion was reached that the problem was not an isolated one but that there was a real risk that the bot would continue to make problematic edits, blocking (or at least disabling the faulty task) pending review by BAG would be appropriate. A pattern of such incidents would suggest a competence issue by the operator which, again, would need further review. I am therefore against creating a mechanism that makes it less likely for these incidents to come to wider attention. Your post to WP:AN (which appears to have been the catalyst for this request) seems to me to have been a satisfactory way to approach the issue with your bot and no doubt means that it came to wide enough attention. WJBscribe (talk) 16:52, 21 May 2019 (UTC)
 * "A pattern of such incidents would suggest a competence" maybe so, but that's also unrelated to whether or not problematic edits should be undone. The answer is yes in both cases. WP:ABF is misplaced here. &#32; Headbomb {t · c · p · b} 18:02, 21 May 2019 (UTC)
 * mass rollback just refers to anything more than 90 edits. Or, the edits could have been made as part of a trial, and then realized to be faulty and reverted, like --DannyS712 (talk) 16:33, 21 May 2019 (UTC)

De-approved. While I don't believe there was anything wrong with a speedy approval, people obviously want to chip in here, so let's keep this open for a bit longer. I'll have more thoughts later. &#32; Headbomb {t · c · p · b} 16:38, 20 May 2019 (UTC)
 * I don't really like the idea of the non-regulation of this request. "Whoops, I just made a big mistake and need to fix it, let's ask Danny to roll everything back and no one will know." Also, I wonder how much we really need this bot - sure, Headbomb requested it once and Danny was able to help, but not everything requires a bot task to handle it. As mentioned previously, an admin can do the rolling back just as easily. Is this sort of mass-rollback-on-demand really necessary? Primefac (talk) 23:08, 23 May 2019 (UTC)
 * It may not be necessary, but it would be useful. For example, when I first trialed Bots/Requests for approval/DannyS712 bot 15 with 50 edits, I had to roll them all back; it would have been less of a flood to recent changes if the rollbacks had been done by a bot --DannyS712 (talk) 23:29, 23 May 2019 (UTC)
 * I don't really like this request - if an editor is making lots of bad edits that need to reverted with we have plenty of admins that can handle it, who can additionally evaluate if there is a bigger issue occurring and escalate/block/etc as needed,  has been available for admins for a long time. —  xaosflux  Talk 21:50, 26 May 2019 (UTC)
 * This isn't intended for reverting back the actions of an editor, just those of bots that are malfunctioning, etc --DannyS712 (talk) 03:07, 9 June 2019 (UTC)
 * There are too many concerns about necessity and oversight to approve this request. Primefac (talk) 18:26, 9 June 2019 (UTC)
 * While I don't agree with these concerns, I agree that the bot lacks consensus for this task. I was about to close this as denied myself, but beat me to it. &#32; Headbomb {t · c · p · b} 19:41, 9 June 2019 (UTC)
 * 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.