Wikipedia:Bots/Requests for approval/DannyS712 bot 21

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

Denied.

DannyS712 bot 21[edit]

Operator: DannyS712 (talk · contribs · SUL · edit count · logs · page moves · block log · rights log · ANI search)

Time filed: 08:38, Tuesday, March 19, 2019 (UTC)

Automatic, Supervised, or Manual: automatic

Programming language(s): AWB

Source code available: AWB

Function overview: Auto-classify stubs as stub-class

Links to relevant discussions (where appropriate): Wikipedia:Village pump (idea lab)/Archive 28#Automatically mark all stubs as stub-class

Edit period(s): As needed, large run at first

Estimated number of pages affected: Lots (631744 total unassessed pages, so I'd guess a ballpark around ~100,000?)

Exclusion compliant (Yes/No): Yes

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

Function details: Automatically assess articles as stub-class if they are tagged as stubs.

Discussion[edit]

@DannyS712: I'm confused about the task. Does the bot task attempt to determine which articles are stubs itself, or does it rely on a stub template being placed in the article. {{3x|p}}ery (talk) 00:35, 20 March 2019 (UTC)[reply]
@Pppery: it only clasifies pages that are already tagged as stubs with stub templates in the article itself. The bot would make to determinations itself --DannyS712 (talk) 01:16, 20 March 2019 (UTC)[reply]
In that case, (1) AnomieBOT is already approved for a superset of this task (ping: Anomie), and (2) I agree with Czar in the linked discussion that this is better suited for coding added to the individual assessment templates (or Template:WPBannerMeta) than a bot task. {{3x|p}}ery (talk) 01:23, 20 March 2019 (UTC)[reply]
@Pppery: the superset AnomieBOT is approved for is for tagging at the request of a wiki project, and only edits that wikiproject’s template (Anomie please correct me if I’m wrong) - this would tag ‘’all’’ unaddressed stubs as stub-class. --DannyS712 (talk) 01:41, 20 March 2019 (UTC)[reply]
You're correct as to what AnomieBOT is approved for there. Also I haven't had much time/motivation to run that task in a long time now. Anomie ⚔ 12:10, 20 March 2019 (UTC)[reply]

@DannyS712 and Czar: I've coded a (somewhat crude) template that detects whether a page is a stub at User:Pppery/is page a stub. {{3x|p}}ery (talk) 01:49, 20 March 2019 (UTC)[reply]
@Pppery: such an approach would mean that wikiprojects can't opt out of the auto-categorization, and editors can't see that there is an assessment when looking in edit mode. --DannyS712 (talk) 02:52, 21 March 2019 (UTC)[reply]
Proper coding of the template would remedy the first stated downside (it would be fairly easy to add a |AUTODETECT_STUB=, for instance). To me, the ability to see a stub assessment in edit mode is not a virtue worth expending tens of thousands of bot edits to produce. {{3x|p}}ery (talk) 03:28, 21 March 2019 (UTC)[reply]
@Pppery: toollabs:pageviews displays an article's class. It would be trivial to display an icon representing the article's rating in edit mode using a module – no bot edits required. Qzekrom 💬 ^they_them 00:35, 22 March 2019 (UTC)[reply]
@Qzekrom and Pppery: okay. Is there consensus to do both of these though? {{WPBannerMeta}} is rarely edited. If not, then a bot would be a less contentious route... --DannyS712 (talk) 00:40, 22 March 2019 (UTC)[reply]
@DannyS712: One edit, even to a highly used template, seems less contentious than ~100,000 bot edits. I think that, given that at least two users (me and Czar on the idea lab page) have said that it is preferable to code this in the template than use a bot, this bot request needs broader discussion. {{3x|p}}ery (talk) 00:53, 22 March 2019 (UTC)[reply]
@Pppery: then where should the discussion be held? WikiProject Council? --DannyS712 (talk) 00:53, 22 March 2019 (UTC)[reply]
@DannyS712 and Pppery: WT:COUNCIL is a good place. I'll also suggest VPPRO. Would it be redundant to also make it an RfC? Qzekrom 💬 ^they_them 02:36, 22 March 2019 (UTC)[reply]
@Qzekrom: I don't know - I filed this per a bot request, and don't mind if it takes a while to go through a bunch of discussions. Feel free to do whatever discussion you feel is necessary for ~100,000 edits --DannyS712 (talk) 02:38, 22 March 2019 (UTC)[reply]
@DannyS712: I think an RfC is the best way to attract as many editors as possible. I spent a lot of effort tagging all those WikiProject talk pages and only a few people commented. By contrast, an RfC would be noticed almost immediately. With an RfC, it also doesn't matter as much where it is posted. Qzekrom 💬 ^they_them 03:08, 22 March 2019 (UTC)[reply]
@Qzekrom: Then go ahead whenever you want, but note that I won't be able to spend much time in the next week on-wiki, so I may be delayed in answering any questions. --DannyS712 (talk) 03:34, 22 March 2019 (UTC)[reply]

@Qzekrom: I believe that data that toollabs:pageviews shows comes from assessment tags on the talk page, so is irrelevant here. {{3x|p}}ery (talk) 00:53, 22 March 2019 (UTC)[reply]
@DannyS712, Pppery, and Czar: Please review the draft RfC at User:Qzekrom/sandbox/rfc, and feel free to edit it if you think the technical details need to be clarified or corrected. I'll subst it to WP:VPPRO when we all think it's ready. Qzekrom 💬 ^they_them 05:42, 22 March 2019 (UTC)[reply]
@Qzekrom: Note that the 100,000 is a complete guess - we should actually run the numbers first. I'll try to find a way, but we should actually have a ballpark number before opening the rfc. --DannyS712 (talk) 05:44, 22 March 2019 (UTC)[reply]
@Pppery, is it possible to autodetect stub based on prose length (and not just having a stub template) too? If so, would want to add that to the RfC. (not watching, please {{ping}}) czar 11:20, 23 March 2019 (UTC)[reply]
@Czar: Yes, that would be technically possible, but quite a bit more work (what counts as "prose" exactly from the point of view of a computer), and I'm not sure whether it is a good idea anyway, because "having < x characters of prose" does not necessarily equal "being a stub". {{3x|p}}ery (talk) 14:41, 23 March 2019 (UTC)[reply]
@Pppery, from my experience, any article with less than 1,500 characters in User:Shubinator/DYKcheck.js is universally considered a stub, fwiw czar 15:55, 23 March 2019 (UTC)[reply]
As I indicated in the earlier discussion: Anything that Danny can do to get stubs out of Category:Unassessed medicine articles is a positive contribution, and I very much hope that he will be able to do that. On a related point, I've done assessment work for more than a decade, and I suspect that the overall lack of responses is because nobody actually objects to this. If you wanted to rate B-class, it'd be much more fraught, but everyone uses the same basic definitions of stubs. WhatamIdoing (talk) 17:05, 22 March 2019 (UTC)[reply]
So as far as assess as stub-class if there is a stub-template goes - should these be getting assessed to see if they are actually stubs anymore? The article certainly could have out grown the stub template. — xaosflux ^Talk 20:43, 22 March 2019 (UTC)[reply]
@Xaosflux: Adding the stub-class assessment would, among other things, alert the wikiprojects that they have stubs, which could encourage them to reassess and improve the pages. I'll note, however, that tagging a talk page as stub-class is far less noticeable to the reader than the stub template on the page itself, so such tags would have very little viewer-facing downsides, while providing the potential benefit of getting the pages reassessed and the stub tag removed. --DannyS712 (talk) 21:15, 22 March 2019 (UTC)[reply]
@DannyS712: aren't these already alerting the project they have unaddressed pages that need to be reviewed? — xaosflux ^Talk 21:22, 22 March 2019 (UTC)[reply]
@Xaosflux: yes, but if this task were run and wikiprojects were notified of it, it would provide a clear source of pages that specifically should be reassessed, since the bot is tagging pages without doing its own assessments (by design). I could specifically mark them as bot-assessed (|auto=stub, see Template:WPBannerMeta/doc), which would further help wikiprojects by telling them what pages had been unassessed but were tagged as stubs, since it would / could prompt the wikiproject to review the assessment. Does this explanation answer your questions? --DannyS712 (talk) 21:32, 22 March 2019 (UTC)[reply]

@DannyS712 and Xaosflux: A future project might involve training an ML algorithm to determine whether an article is likely to have improved beyond stub quality (using removal of the stub tag as proxy for improvement), and then alerting WikiProjects when an article is ripe for reassessment. Do we have anything that does that? Qzekrom 💬 ^they_them 01:13, 23 March 2019 (UTC)[reply]
@Qzekrom: a bot isn't needed (I think) - can you give me a wikiproject, and I can give you a list of "stub-class" articles without stub templates (see Category:Stub-Class articles) Thanks, --DannyS712 (talk) 02:25, 23 March 2019 (UTC)[reply]
@DannyS712: WP:COMP ...wait, was that rhetorical? Qzekrom 💬 ^they_them 02:40, 23 March 2019 (UTC)[reply]
@Qzekrom: Not at all. There were some false positives, but see User:DannyS712/sandbox5 for the list of 5081 pages that the wikiproject tagged as stubs but that currently don't have stub templates. --DannyS712 (talk) 04:56, 23 March 2019 (UTC)[reply]
@DannyS712: Should we add stub templates, though? We ought to be careful to add the most specific ones. For example, the ones about TLDs, such as .gr, should be tagged with {{Compu-domain-stub}} whereas A+ (programming language) should be tagged with {{Prog-lang-stub}}. Qzekrom 💬 ^they_them 05:07, 23 March 2019 (UTC)[reply]
@Qzekrom: Um, what? You posted that A future project might involve training an ML algorithm to determine whether an article is likely to have improved beyond stub quality (using removal of the stub tag as proxy for improvement), and then alerting WikiProjects when an article is ripe for reassessment. - here is a list of articles that have been improved beyond stub quality and should be reassessed, not a list of untagged stubs. We should not ad stub templates, since these are generally no longer stubs. --DannyS712 (talk) 05:13, 23 March 2019 (UTC)[reply]
@DannyS712: Oh, when you said "'stub-class' articles without stub templates" I assumed that you weren't specifically discussing ones with their tags removed. Qzekrom 💬 ^they_them 05:28, 23 March 2019 (UTC)[reply]
@Qzekrom: I'm not specifically discussing ones with their tags removed. These are pages where the wikiproject says they are stubs, but they are not tagged as stubs. Either one could be correct, but it would take a WP:CONTEXTBOT to figure that out automatically --DannyS712 (talk) 05:33, 23 March 2019 (UTC)[reply]
@DannyS712: I see. Thank you for clarifying. Qzekrom 💬 ^they_them 05:40, 23 March 2019 (UTC)[reply]
I alluded to this in the linked conversation but I think the point of tooling here is moot when we should be looking instead to develop consensus on Phase 2 of meta:Community Tech/PageAssessments, which would pull banner assessments offsite and thus be easier to manipulate and less obtrusive to editors. (not watching, please {{ping}}) czar 11:20, 23 March 2019 (UTC)[reply]
@Czar: Phase 2 is labelled Not done. Dropped from roadmap., so that may not be an option. Furthermore, that would require a lot bigger consensus, because it would fundamentally change the way wikiprojects assess quality and importance, while this task works within the current framework. --DannyS712 (talk) 23:56, 23 March 2019 (UTC)[reply]
The talk page has more info on that, but yes, it would change how we log assessment data if we as a community were to have that discussion. By detaching assessments from on-wiki edits, it should become much easier to auto-assess articles, as proposed, and manipulate assessment data. It's a different project but one that positions us better for the long run. (not watching, please {{ping}}) czar 23:06, 24 March 2019 (UTC)[reply]
@Czar: I see. However, as I noted above, that would require much greater consensus. For this, it works within the system that already has consensus, and if specific wikiprojects want to opt out that can probably be arranged. --DannyS712 (talk) 18:20, 25 March 2019 (UTC)[reply]
{{BAGAssistanceNeeded}} --DannyS712 (talk) 19:35, 2 April 2019 (UTC)[reply]

Denied. I'm going to deny this bot task. Its scope, as proposed, is simply too massive, with logic that's too ill-defined to to evaluate. Zero prejudice against submitting a similar task with a more limited scope (e.g. adding |class=stub to specific Wikiproject banners, upon consultation with a Wikiproject, if prose size < x words / bytes, or if prose size y < words/bytes AND tagged with a stub tag) or clearer logic. Autodetection of stub status through LUA magic being much much preferable to bot runs on hundred of thousands of talk pages. Headbomb {t · c · p · b} 04:21, 9 April 2019 (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.