Wikipedia:Bots/Requests for approval/Snotbot 11

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Withdrawn by operator.

Snotbot 11

Operator: Scottywong (talk · contribs · SUL · edit count · logs · page moves · block log · rights log · ANI search)

Time filed: 14:19, Tuesday June 5, 2012 (UTC)

Automatic, Supervised, or Manual: Automatic

Programming language(s): Python

Source code available: Pywikipedia library with custom code

Function overview: Automatically revert and block a prolific IP-hopping vandal

Links to relevant discussions (where appropriate): Wikipedia talk:Abuse response, Wikipedia:Abuse response/2.30.196.109, User:Ebe123/F1 Vandal

Edit period(s): Continuous

Estimated number of pages affected: A couple hundred at most

Exclusion compliant (Yes/No): Yes

Already has a bot flag (Yes/No): Yes, but will need an admin bit

Function details: A prolific IP-hopping vandal has been causing havoc on various racing articles. Blocking his IP's for an extended period of time would make no difference as he can just hop to a new one. Rangeblocking his IP would cause far too much collateral damage. However, his edits follow a specific pattern, and the various IP ranges he uses have been deduced (see User:Ebe123/F1 Vandal).

The vandal's edits follow a specific pattern. Per WP:BEANS, I'd prefer not to divulge the exact pattern that I'm matching against, suffice it to say I have created a tool which pulls his edits out of the recent changes feed, and so far it has found zero false positives. There are even a few more checks that the bot could make (which the tool currently isn't checking) to make it even less likely for a false positive to emerge.

Basically, when the bot finds a matching edit in the recent changes feed, it will:

Automatically revert the edit
Block the IP address for a short time (1-2 hours, most likely)
Leave a message on the user talk page of the IP

It is our hope that the vandal will be discouraged after being immediately reverted and blocked after each vandalism edit. This task will likely only run for a short time, until the vandal either gives up or gets more creative by mixing up his pattern. -Scottywong | verbalize _ 14:19, 5 June 2012 (UTC)[reply]

Discussion

Brilliant response that resulted from a request at wp:Abuse response. Would, however, like to see {{signup}} posts on the IP talkpages prior to the outright blocks. Better the editor should engage in discussion than grow more creative at evading RBI. LeadSongDog come howl! 15:19, 5 June 2012 (UTC)[reply]

The message posted on the user talk page can certainly include encouragement and instructions on how to communicate with us, in the unlikely event that this user wants to make non-vandalism contributions to Wikipedia. However, from what I can tell, posting {{signup}} messages on the user's talk page (while he is actively vandalizing) has been tried in the past and did not result in anything positive. It is abundantly clear that this is not a well-meaning user who is simply confused or unsure of what he is doing. I think we need to strike a balance between encouraging communication and making it clear that we're not going to deal with the BS anymore (i.e. with immediate short blocks). -Scottywong | chatter _ 15:33, 5 June 2012 (UTC)[reply]

What if another person using one of the IP addresses of the range makes a good faith edit and that is reverted? Rcsprinter (tell me stuff) 19:51, 5 June 2012 (UTC)[reply]

That is a valid concern. However, the toolserver tool referenced above has proven to be 100% accurate (so far) in finding only this vandal's edits, in other words it has never mistakenly reported someone else's edit as a vandal edit. It goes well beyond simply looking at the IP address of the user. The same search this tool uses (with some added enhancements) would be used by the bot to find only vandal edits. The probability of the bot making a mistake is quite minimal. In addition, the IP will only be blocked for a short time (1-2 hours) so the disruption would be minimized, and they will receive a message on their user talk page with instructions on how to contact someone to request an unblock if there happened to be a mistake. Finally, I only expect this bot task to run for a few weeks at the most, before the vandal either gives up or figures out a way around it.

So, in summary, the chance of a mistake is quite low, the amount of disruption caused in the event of a mistake is relatively minimal, and the bot task will only be running for a relatively short period of time. -Scottywong | communicate _ 21:09, 5 June 2012 (UTC)[reply]

Has the possibility of using the WP:Edit filter to prevent these edits been considered? I don't see it mentioned at all here or in the linked discussion. Anomie ⚔ 12:05, 6 June 2012 (UTC)[reply]

Actually, that's not a bad idea. I'm not familiar with creating new edit filters, but either I could learn or we could find someone who is familiar with it and give it a try. -Scottywong | gossip _ 14:09, 6 June 2012 (UTC)[reply]

A couple of things:

This needs much wider advertising, please spam the usual notice boards
I would like to see the source code
You need to show that the vandalisim has reached such a level that this is nessary; block bots are an extreme last measure
- This includes the above mentioned by Anomie - Has an edit filter been tried? How will this task work better than an edit filter?
What happens when the vandal starts adapting their tactics to counter your bot?
What kind of emergency shutoff measures are in place?
- On that note, can someone please clarify, does media wiki still allow admins to block/unblock while they are blocked?

--Chris 14:45, 6 June 2012 (UTC)[reply]

I think I'd like to place this BRFA on hold until we can determine if the edit filter is capable of carrying out this task. -Scottywong | babble _ 15:22, 6 June 2012 (UTC)[reply]

Some replies:
A couple of things:

This needs much wider advertising, please spam the usual notice boards
I would like to see the source code
You need to show that the vandalisim has reached such a level that this is nessary; block bots are an extreme last measure
- It seems nessesary, see the page in my userspace that has the list of vandals. ~~Ebe 123~~ → report 00:18, 7 June 2012 (UTC)[reply]
- This includes the above mentioned by Anomie - Has an edit filter been tried? How will this task work better than an edit filter?
  - Scottywong stated that he has not considered that. ~~Ebe 123~~ → report 00:18, 7 June 2012 (UTC)[reply]
What happens when the vandal starts adapting their tactics to counter your bot?
- Then people will take over with some help with IP info and SPI possibly.
What kind of emergency shutoff measures are in place?
- On that note, can someone please clarify, does media wiki still allow admins to block/unblock while they are blocked?
  - Yes, admins may unblock themselves, although there was some discussion about that. ~~Ebe 123~~ → report 00:18, 7 June 2012 (UTC)[reply]

--Chris 14:45, 6 June 2012 (UTC)[reply]

Replied. ~~Ebe 123~~ → report 00:18, 7 June 2012 (UTC)[reply]

Withdrawn by operator. - The edit filter is a much better tool in this case. I've added and am currently testing filter 470. -Scottywong | spill the beans _ 23:39, 7 June 2012 (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.