Wikipedia:Bots/Requests for approval/StraussBot 2


 * 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

StraussBot 2
Operator:

Time filed: 17:09, Tuesday, June 27, 2017 (UTC)

Automatic, Supervised, or Manual: Automatic.

Programming language(s): Perl.

Source code available: Link.

Function overview: StraussBot declines completely unreferenced Articles for Creation submissions and notifies the page creator.

Links to relevant discussions (where appropriate): Consensus being sought

Edit period(s): One-off run on the whole AFC backlog if approved and then continuous run at least twice daily on Category:AfC pending submissions by age/0 days ago in order to capture submissions from all time zones. Task Scheduler on my PC will enable the bot to keep to a regular timetable.

Estimated number of pages affected: Approximately 30 per week (15 submissions and 15 user talk pages – figures based on tests I have done whilst programming the bot).

Exclusion compliant (Yes/No): No.

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

Function details: Problem: rising backlog at Articles for Creation with approximately 15 completely blank submissions being created per week. Such drafts can be declined straight away per WP:V, enabling reviewers to focus on sourced drafts. Solution: StraussBot declines completely unreferenced Articles for Creation submissions using the same template syntax a human reviewer would use with AFCH. It does not add the declined template to all drafts in the draft space: it only looks at pending submissions. “Completely unreferenced” is defined as a submission not containing “http”, “www”, “ref”, “bibliography”, “source” and “cite”. Therefore, some unreferenced submissions may be skipped if they contain such words in a different context but further coding to tackle this can be done in due course (see test data set 2 below). It also notifies the creator on their talk page that their submission has been declined. A count of how many pages it has declined is also programmed in and this count will be displayed on its user page. In my opinion, the bot should not be exclusion-compliant merely because AFC creators are unlikely to know about the template exclusion system and even if they did there is the potential for abuse by creators indiscriminately putting the template at the top of their submissions. An emergency block button will be provided or if bugs do arise I can merely remove the bot from my task scheduler. I see no reason why draft articles should be considered risky for bot operations and StraussBot would pose no risk to other pages as it would not have the need nor the capability to edit them.

Set 1: John Smith (1954-1996) was a maths teacher who killed his wife Penelope.

The bot should decline this submission. The submission is unreferenced and contains none of the skip-words.

Set 2: ''John Smith (1954-1996) was a football referee who killed his wife Penelope. Whilst in prison, he invented the Hypertext Transfer Protocol (HtTp).'' This is a poor submission because it contains no references but it does contain the string “ref” and “HtTp”. |||

The bot should skip this submission as it is currently unable to discern between the contextual usage of words such as “ref” and “http”. Improvements in the coding in future should allow the bot to decline this. Note that the bot is case insensitive.

Set 3: ''John Smith (1954-1996) was a Catholic priest who killed his wife Penelope. ''

The bot should skip this submission as it is unable to evaluate the reliability of sources. It merely detects whether any sources are present.

Set 4: ''John Smith (1954-1996) was an American lawyer who killed his wife Penelope. ''

The bot should skip this submission as it is unable to evaluate the reliability of sources. It merely detects whether any sources are present.

Discussion

 * I can't find the discussion seeking consensus at the link that you gave above; is it this discussion? TheMagikCow (T) (C) 12:07, 28 June 2017 (UTC)
 * A couple of things: it looks like in the discussion that TheMagikCow (thanks, by the way) linked to, editors prefer that the bot decline only completely blank pages. That task sounds much less controversial to me, given my past experience with the AfC project. Second, instead of Task Scheduler, you may want to look into running your bot on Tool Labs, since it's much more reliable that way. Enterprisey (talk!) 23:08, 29 June 2017 (UTC)
 * Also, I took the liberty of collapsing the test data sets you put under "Function details"; hope you don't mind. Enterprisey (talk!) 23:10, 29 June 2017 (UTC)
 * This strikes me as a WP:CONTEXTBOT issue. It may be possible to get around that (e.g. have a bot that initially declines with advice but leaves a message saying it's automated and you can resubmit if the bot made an error). That would need consensus. I'm most concerned about parenthetical citations to offline sources, which are perfectly valid but would not contain the strings you identified. ~ Rob 13 Talk 17:23, 5 July 2017 (UTC)
 * I do wonder how many pages this would affect, given that the draft creation wizard adds in a references section and the template when creating a draft. This would result in a match for the   keyword that you have suggested and is also a non blank page. I am just not sure how much time this bot would save compared the the risks of false declines. For me to be convinced that the bot is necessary, I would have to see a generated list of pages that would be declined and check for false positives and a large-ish number of pages. I also say that draft articles are risky (against this from the details: I see no reason why draft articles should be considered risky for bot operations ). It is typically newcomers that create drafts, so there must be a special effort to avoid false positives, as false positives can seem a bit WP:BITEY. Having a good draft auto declined will be frustrating. With editor retention a big issue at the moment this namespace should have special consideration, not less. TheMagikCow (T) (C) 16:02, 11 July 2017 (UTC)
 * From an AFC perspective, there are anywhere between 150 and 200 pages submitted per day. A review of a no-reference page, especially for experienced editors, takes maybe 30 seconds. Even if 60 drafts are submitted with zero references, this only saves about 30 total minutes (spread out over all reviewers) of editing time per day. I'm not really sure it's necessary, and I'm pretty sure there's an AFC conversation somewhere that kind of agreed with that. Too tired to find a diff at the moment. Primefac (talk) 05:26, 16 July 2017 (UTC)
 * Due to the lack of consensus and my departure from the AfC project I'd like to withdraw the bot request. If I return to AFC, the BRFA can be re-opened.    Dr Strauss   talk   10:48, 16 July 2017 (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.