[00:54:58] 10Scoring-platform-team, 10MediaWiki-extensions-ORES, 10editquality-modeling, 10User-Ladsgroup, 10artificial-intelligence: [Spec] Use `reverted` models in ORES review tool - https://phabricator.wikimedia.org/T146378 (10jmatazzoni) Hi @Ladsgroup. I'm not exactly sure what implementation means here. Are yo... [00:56:59] halfak: Do you have access? https://ieeexplore.ieee.org/abstract/document/4536957/ [00:57:29] also: WHAT. They’ve implemented “collaborative continuous auditing” using a platform called “JADE”. This is odd. [00:59:05] kk "Java Agent DEvelopment framework" [00:59:20] that’s a bad acronym! [01:14:44] (03CR) 10Awight: "Ladsgroup: Anything else I should change?" [extensions/JADE] - 10https://gerrit.wikimedia.org/r/447927 (owner: 10Awight) [01:15:23] 10Scoring-platform-team, 10Gerrit, 10ORES: Research Project Idea: Use AI to suggest improvements to patches uploaded to gerrit - https://phabricator.wikimedia.org/T195235 (10Paladox) You can use this api https://gerrit-review.googlesource.com/Documentation/rest-api-changes.html#apply-fix i think (I'm not sur... [01:16:32] (03CR) 10Awight: Service wrapper to prevent misspellings (031 comment) [extensions/JADE] - 10https://gerrit.wikimedia.org/r/448085 (owner: 10Awight) [01:48:33] halfak: Just noticed Sandvig’s writing about collaborative auditing, I should reach out... [07:16:56] 10Scoring-platform-team, 10ORES: Implement prioritization of request processing - https://phabricator.wikimedia.org/T148594 (10Ladsgroup) 05Open>03declined [10:37:41] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10Marostegui) >>! In T200297#4462189, @awight wrote: > @Marostegui, I'd like to explore how we might be able to use x... [10:40:59] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10jcrespo) > If we do that, is it possible to do queries that join between x1 and other production tables? You will... [11:58:17] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10daniel) Just to make sure we are all taking about the same thing: Do I understand correctly that the scalability c... [12:20:20] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10jcrespo) Data (external storage) storage is never a concern, because it is a key-value storage and is already shard... [14:09:38] that paper is not easy to get :| [14:21:21] hi halfak! what would be the easiest way to get topic predictions for a (somehow large) list of articles using your models? Thanks!! [14:21:46] How large is large? [14:22:13] (Order of magnitude) [14:22:31] let's say 1milion, ( I need to check) [14:24:57] halfak: having this info for all enwiki articles would be the ideal scenario [14:26:04] 1 million is certainly do-able by querying the service. For 10 million or more, I'd say we should figure out how to run an offline process on the stat machines. [14:26:14] Good news is the stat machines are the right environment to run our model :) [14:26:27] 10Scoring-platform-team, 10ORES: Enable wp10 and draftquality models for testwiki - https://phabricator.wikimedia.org/T198997 (10SBisson) Not sure this is the right task to bring it up but we would really like to have real scoring for 'wp10' and 'draftquality' on test.wikipedia.org. We plan on doing user testi... [14:26:31] * halfak wishes we could do this in spark. [14:33:00] ok, halfak, thanks! yes it would be greaet if we could run the process on stat/spark ;) [14:33:31] miriam, do you have experience with python virtual environments? [14:33:40] halfak, yes :) [14:35:51] OK can you set one up with python 3 and then follow the instructions at https://github.com/wiki-ai/revscoring of "Installation"? [14:36:19] Ad you might expect, you can skip any of the system commands (e.g. apt-get) because they are already done. [14:36:34] Mostly you need the nltk download of "python -m nltk.downloader omw sentiwordnet stopwords wordnet" [14:36:51] Then clone https://github.com/wiki-ai/drafttopic [14:37:37] * halfak waits lonely in the ORES docs meeting. [14:42:39] halfak, thanks - i'll follow your instructions! [14:43:54] Great. Let me know when you get that far. There will be some work to do depending on how you want to proceed :) [14:44:16] E.g. should we get text from XML dumps or maybe request 5 million texts from the MW API? [14:44:23] That's a future hurdle :D [15:02:42] 10Scoring-platform-team, 10JADE: Bikeshed: Explore alternative names for "judgment" in JADE - https://phabricator.wikimedia.org/T200365 (10Milimetric) for what it's worth, I thought of Review when I was pondering on this over the weekend. I see it's already been proposed, and RADE doesn't seem that bad :) [15:06:18] heads up eqiad row B network maintenance, ores100{3,4} are going to have a network hiccup [15:09:16] Thanks for the notice! [15:31:01] (03PS1) 10Sbisson: Maintenance script to backfill scores in PageTriage queue [extensions/ORES] - 10https://gerrit.wikimedia.org/r/449475 (https://phabricator.wikimedia.org/T198982) [15:37:59] halfak sorry, I was in a meeting. Thanks. I will let you know how it goes! [15:43:08] (03CR) 10jerkins-bot: [V: 04-1] Maintenance script to backfill scores in PageTriage queue [extensions/ORES] - 10https://gerrit.wikimedia.org/r/449475 (https://phabricator.wikimedia.org/T198982) (owner: 10Sbisson) [15:53:56] I figured out how to adapt a general tool to local wiki configurations. [15:53:57] https://en.wikipedia.org/wiki/User:EpochFail/ArticleQuality-tester.js [16:21:19] (03PS2) 10Sbisson: Maintenance script to backfill scores in PageTriage queue [extensions/ORES] - 10https://gerrit.wikimedia.org/r/449475 (https://phabricator.wikimedia.org/T198982) [16:35:29] halfak: did you show off anything in particular at the hackathon showcase? [16:35:41] Yes. The three gadgets I worked on. [16:36:04] Related to the work I'm doing between meetings this morning. I want to generalize the gadgets and make them available everywhere. [16:43:53] ooh do show us [16:45:07] oho https://meta.wikimedia.org/w/index.php?title=User:EpochFail/ArticleQuality-system.js [16:45:19] :) [16:45:23] -> ORES/js modules [16:45:29] Right. [16:45:36] Like the "oresScore" function in there [16:46:29] nice. I caught some stuff in the T200297 net overnight [16:46:29] T200297: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 [16:46:45] I'm reading through it right now [16:47:02] I'm trying to work through contingency strategies for each pain point [16:47:10] In particular, the not being able to get data on Wiki Replicas [16:47:33] I've momentarily flipped into Cloud Mode and am wondering "are there other ways we could surface this data product" [16:47:53] harej: I’d love to have those contingency strategies on https://etherpad.wikimedia.org/p/JADE_scalability_FAQ [16:48:12] harej: Increasingly thinking that the “single wiki” alternative is total garbage, though. [16:48:45] It seems that it doesn’t even get our data onto “x1”, so doesn’t do much to temper the Ops concerns. [16:49:07] All it does is isolate our database a bit so that it can be killed en masse if needed, which I plan to never need. [16:50:10] When you guys are done with your current conversation would you mind telling me if the current Inspire campaign was at all a result of the stuff you do here an in the phabricator? [16:51:02] * awight rubs eyes after reading to the end of the task [16:51:18] (I have no context for how fragmented or close-knit the various aspects of the wiki community are.) [16:51:42] term|CONT: there are a lot of Inspire campaigns it seems, do you know which one you’re thinking about? [16:52:30] The one that got a notification on Wikipedia recently. The one about creating a metric for measuring the gestalt health of a wiki. [16:53:27] term|CONT: Can you find it here? https://meta.wikimedia.org/wiki/Grants:IdeaLab/Inspire/Ideas_by_category [16:53:50] Maybe https://meta.wikimedia.org/wiki/Grants:IdeaLab/Current_health_of_Wikipedia_editing_community ? [16:54:30] If you get the notices at the top of the page, it’s probably listed in https://meta.wikimedia.org/wiki/Special:CentralNotice [16:54:43] Well I mean that whole thing. [16:54:51] term|CONT: Which wiki are you on? en.wikipedia.org? [16:54:59] The current Inspire Campaign is on Community Health [16:55:27] That's what I came from but the campaign is apparently on meta.wikimedia and the accounts are the same. [16:55:27] ty [16:55:49] And to broadly answer the question, there is a Foundation-wide, "cross-departmental" program called Community Health which seeks to figure out some of these problems. The Research team formally collaborates on this program I think; I am not sure about the Scoring Platform (AI) team. [16:56:13] https://meta.wikimedia.org/wiki/Grants:IdeaLab/Inspire [16:56:31] I don't think we're "officially" part of that program, but it is a key point of our focus. [16:57:02] I think ORES models will be very useful in measuring community health and improving community health :) [16:57:09] We have some past research there. [16:57:19] https://commons.wikimedia.org/wiki/File:Demonstrating_the_Keilana_Effect_(OpenSym%2717).pdf [16:57:28] I think it's https://meta.wikimedia.org/w/index.php?title=Special:CentralNotice&subaction=noticeDetail¬ice=InspireHealth2018 but it doesn't show a preview of the notice so it's hard to tell. [17:00:20] term|CONT: Here’s the wiki page for banner content, https://meta.wikimedia.org/wiki/MediaWiki:Centralnotice-template-Inspire_Community_Health [17:00:36] So which IRC would most focus on the Community Health branch? [17:00:52] I don't think they really do IRC [17:01:24] Looks like you should look for Seddon on any channel, https://meta.wikimedia.org/w/index.php?title=MediaWiki:Centralnotice-template-Inspire_Community_Health&action=history [17:02:48] Yeah, for CentralNotice stuff Seddon is the best person to reach out to [17:13:00] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10awight) >>! In T200297#4464608, @Marostegui wrote: > What does: "our revision data on x1" means? Your own set of ta... [17:17:59] Does the Research team have an IRC? [17:19:06] #wikimedia-research [17:19:09] term|CONT, ^ [17:19:20] Thanks. [17:19:36] It's not the research team's channel exactly, but they occupy it along with lots of other research folk. [17:19:39] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10awight) >>! In T200297#4464943, @jcrespo wrote: > There is an issue because there is not way to prevent recursive s... [17:21:46] I guess most of the stuff will be happening after the campaign ends. [17:31:38] Things have grown so much that I'm honestly not sure what level of interface I'm supposed to use to figure out where to go. [17:34:43] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10Harej) >>! In T200297#4464630, @jcrespo wrote: > > You will not be able to join data on wiki storage and metadata... [17:36:08] harej: yeah x1 not sounding so hot now. [17:36:29] custom schema & no provision for analytical use? hell no. [17:37:25] What, exactly, is x1 anyway? Is it miscellaneous database hell? [17:37:29] Yeah. That's a bad suggestion. I think it is clear that the proposal comes from a deep misunderstanding of what JADE is. [17:37:37] harej, spot on. [17:37:48] It's for extensions to store things that are unrelated to revisions and pages. [17:37:48] lolol [17:38:43] The nice part about this… challenge, is that it gives me an excuse to perpetually refine our docs. [17:38:48] Is there a canonical article that gives an overview of how machine learning techniques are being used on these projects? [17:39:08] Feel free to jump in wherever we need more clarity! [17:39:24] Who? Me? [17:40:51] I have a decent but outdated grasp of this stuff, but I'm really not sure where to go to get caught up on what you guys are doing. [17:42:05] halfak: I finally pushed myself over the edge btw, and think we need to get in touch with CSandvig…. I’ll draft a letter & share [17:42:15] Cool! [17:42:17] term|CONT: hehe that was pointed at harej, sorry my IRC client is being flaky [17:42:55] halfak: yeah I found the part where he says, collaborative auditing is the most promising approach for tackling corporate opacity, and it made me want to strategize... [17:42:56] term|CONT, I'd love to give you an overview, but now is a bad time. Want to drop me an email and we can take it from there? [17:43:47] I don't really want to get involved I'm just sort of trying to figure out how far a complete outsider can get without having to talk too much to anyone. [17:44:11] awight: the current scalability FAQ assumes too much prior knowledge about JADE so I am going back to basics [17:44:30] Machine learning applications tend to be pretty intractable to people in general unless there's some decent effort put into making people understand it. [17:45:16] term|CONT, do you want to know about AI&Wikimedia generally or would it be OK if we showed you a specific AI service we maintain? [17:45:24] I mean if there's nothing up yet and the various projects aren't that far along that's fine too. [17:46:51] halfak: It's more like, if you showed me the AI service, would I even understand how it's being used without first understanding more about Wikimedia and how it works? [17:47:23] https://www.mediawiki.org/wiki/ORES is a good start for that question [17:47:33] https://www.mediawiki.org/wiki/ORES/Applications for some examples of how people are using it. [17:47:39] harej: Awesome. Yeah I think this project is actually worth any obsessive documentation energy we all have to put into it. We’re pioneering some important stuff, IMO. [17:47:52] Wikipedia and the associated *medias seem kind of gargantuan and I'm not sure how anyone navigates any of it without just picking something and pursuing that specifically. [17:47:53] Starting from scratch is solid. [17:48:34] term|CONT, indeed. Explaining Wikimedia is a bit out of scope for us. [17:50:39] term|CONT, FWIW, I do explain Wikipedia in the talks I give about ORES. [17:51:08] If you want to understand the system from a more conceptual level, there's papers like https://commons.wikimedia.org/wiki/File:ORES_-_Facilitating_re-mediation_of_Wikipedia%27s_socio-technical_problems.pdf [17:51:15] term|CONT: picking something random and pursuing it is what a lot of people do. It's not organized all that well. [17:51:40] Also, I think there's a good argument that fluency only comes from in-context interactions. [17:51:51] Our users gain fluency with ORES by playing with it. [17:52:14] halfak: Alright, thanks. That /Applications link was helpful. [17:52:23] It's interesting how deep of an understanding they develop of the systems potential and limitations without formal knowledge of jargon like precision and recall. [17:52:27] great! [17:53:25] I'm running off to lunch. I'll back in ~ an hour. [17:53:47] o/ [17:54:06] halfak|Lunch: That just means you're doing a good job creating a product. [17:56:18] harej: Yeah there's nothing wrong with that at all. [17:56:31] I just tend to think in terms of gestalts. [17:56:44] Wikimedia is more of an ecosystem [17:56:57] I can see that. [17:57:07] There are parts. They interact multi-directionally. Some parts exert bigger control over the ecosystem than others. [17:57:21] For more, read Morgan et. al. [17:58:52] The problem there is that when you have a system of humans without clear accountability rules, accountability slowly erodes to the point where it can even become irrecoverable. [17:59:32] Which conflicts pretty heavily with the concept of volunteers (editors). [17:59:51] For anything other than an encyclopedia it'd probably fail. [18:00:38] term|CONT: That’s one tendency for sure, but there are others such as self-organization which naturally counteract this human nature. [18:01:11] Well the thing is, the true measure of accountability for an encyclopedia is the quality of its sources. [18:01:50] The admins and dispute resolutions and so on are only necessary because the editing happens in an unprofessional context. [18:02:00] That’s one good measure sure, but we can’t disregard interpersonal accountability [18:02:02] But even that's slowly been changing as Wikipedia expands. [18:02:49] ah yeah like you just implied. There’s also quality of writing and so on, not sure why we would reduce accountability to this one dimension. [18:03:38] Well, quality of writing means nothing if your source is fabricated. [18:04:10] So there's the intellectual accountability beyond the bureaucratic/administrative accountability. [18:04:53] (I have a terrible habit of trying to engineer solutions to things when the problem is hard to state.) [18:21:33] Amir1: ping https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/JADE/+/447927/ [18:26:31] halfak|Lunch: Thanks, that link was nice. [18:35:21] This is excellent. I see that the in the Wikilabels schema, in the "task" table, the "data" column is a json blob. That means making a new "view" that labels multiple revisions (a "session") will not be as hard as I thought from a database perspective. [18:38:39] awight: Hey! I assume you're aware of this, but just in case, I just got an email from github that says there's a security vulnerability in JADE [18:38:51] notconfusing: I <3 postgres’s JSON functions [18:40:21] ewhit_: Oh, thanks for that! I gave up on reading because I get spammed from every Wikimedia repo. [18:40:32] Lol fair [18:40:55] I still don’t see the notice, but can wait for it to roll in. That code isn’t deployed anywhere, and should probably be deleted shortly [18:41:21] ok, cool. I can forward it to you, if you need me to [18:41:29] How are interviews? Must aalmost be in the summarization phase? [18:41:35] sure if you have it open, ty [18:41:42] Good! I have about 2 left! :D [18:42:03] Just sent it [18:42:18] Whew, that’s some perseverance! [18:42:39] I'm just glad I'm almost done interviewing! Still working on transcribing, but that should be done soon too [18:42:54] ooh yeah that would be a scary vuln if it were currently deployed [18:43:10] ooh. Well, glad it's not [18:43:26] The Keilana paper is interesting too. [18:44:37] ewhit_: oh dang, exploit databases now come with an estimated price to get a haxor to use the vuln in an attack! [18:44:41] so practical. [18:47:56] awight: How close would you say you might be to a method for measuring the effectiveness or timeliness of a grant? [18:48:30] A good person to ask on that would be Sati Houston, shouston@wikimedia.org [18:48:41] awight: Tusedays are Wikidata day :) [18:49:49] awight: Oh, fantastic [18:50:33] Amir1: they can’t have you! :p [18:50:55] :))) Then you talk to angry Lydia :D [18:54:41] of a given grant* [18:55:25] halfak|Lunch: Remind me who we took wiki “entity” language from? [18:56:24] Ah EventBus [18:56:30] https://www.mediawiki.org/wiki/Topic:Tzygtjcyrclilp8u [18:58:20] Er, oh. Got the names confused. halfak|Lunch ^ above question was for you. [18:59:44] o/ [18:59:47] awight: some of the FAQ language I am writing uses rather simplified language. This is not to be condescending to the site reliability engineers but so the writing can be used in other contexts as well. [19:00:11] Here comes the siren [19:00:12] excellent, ty! [19:00:16] haha [19:00:17] 10Scoring-platform-team, 10DBA, 10JADE, 10Operations, 10TechCom-RFC: Introduce a new namespace for collaborative judgments about wiki entities - https://phabricator.wikimedia.org/T200297 (10awight) @jcrespo Another angle of this that I'd like your input on is the association between judgment page title a... [19:00:24] term|CONT, re. modeling grant success, I don't think we have anything like that right now. [19:00:28] (San Francisco tests its civil defense sirens every Tuesday at noon) [19:00:38] I don't know if there's any analysis we might draw from in the literature either. [19:00:42] oh I thought that was a lunch siren! [19:01:09] It could very well double as a lunch siren. [19:03:55] harej: Great, FAQ changes are looking good so far…. I didn’t realize you were rewriting in place, that’s even better. [19:04:25] halfak: Logically speaking, grants are allocated based on various PR motives, suggesting that there's always a preexisting cause of the movement necessary to improve article quality en masse. [19:09:38] s/PR/democratic/, we would hope [19:10:03] PR is inherently democratic. [19:16:02] Is anyone working on a full web of all inter-page links on Wikipedia? [19:17:06] PR as in public relations? [19:17:17] Yes. [19:17:58] Saying that PR is inherently democratic is like saying that technology is inherently peaceful or bellicose, IMO. [19:18:26] Public relations? [19:18:58] Even democracies vote for war. [19:19:24] Listening to people is good. Status quo PR, attempting to control their thoughts based on this study is hardly democratic, but a completely new and unusual form of PR in which public opinion was respected and somehow invited to help forge their own fate would indeed be democratic. [19:19:55] Or maybe u meant that WMF PR is democratic, which would be closer to the truth in my view. [19:20:25] It does vary per organization. [19:20:51] Hehe or I’m just poking at details. term|CONT I like your idea above of discovering corporate motives by studying their actions. [19:21:22] I wouldn't say "discovering." [19:21:38] Corporate motives tend to be fairly transparent to a consultant's eye. [19:24:27] (I would separately argue that technology is the sole determining factor in quality of life and thus is indeed peaceful.) [19:26:21] I think I can see how the *medias and *pedias are organized now. [19:27:39] ^ this is getting pretty off topic [19:28:12] notconfusing: Are you working on backend for search? [19:39:09] awight: how does the page naming for diffs work? if revision Y replaces revision X, is it Judgment:Diff/Y ? [19:40:39] I wasn’t aware of revisions being replaceable [19:40:45] Can you give an illustration? [19:41:09] I assume you’re talking about something other than normal editing... [19:42:04] You have a page. You are looking at the current revision of the page. You make an edit. That edit you make is the new current revision. [19:42:19] The new edit is displayed as "the page" instead of the old one that was there before you hit edit. [19:43:12] So, Judgment:Diff/1 is judgments about ?diff=1 and Diff/2 is about ?diff=2 [19:44:20] but 1 relative to what? relative to the rev_id of the edit that came immediately before? [19:46:15] parent, yeah. [19:46:28] When we are talking about an edit, it's always in comparison to the parent. [19:46:35] We could do "Edit" rather than "Diff"? [19:46:37] Would be OK [19:46:53] I think "diff" makes sense -- I just wanted to confirm that my assumption matched reality. [19:46:58] +1 :) [19:47:35] And it's just Judgment:Page and Judgment:Diff, anything else? [19:47:59] Do we have judgments for individual revisions accounted for? [19:49:01] harej: Ah, no the number is the revision ID [19:49:17] Diff/, Revision/, Page/ [19:49:32] As I figured. Thank you. [19:50:37] Is this correct: [19:50:42] How are these annotations stored? [19:50:42] In the Judgment namespace, annotations for a given page, revision, or diff are stored on a single wiki page. For example, the annotations concerning the page with an ID of 123 would be at Judgment:Page/123. The annotations concerning revision 456 would be stored at Judgment:Revision/456. The annotations concerning the difference between revision 456 and its parent revision would be stored at Judgment:Revision/456. [19:51:53] harej: Have you seen mw:Extension:JADE recently? Should be there. [19:52:02] * awight almost misses AsimovBot [19:52:08] [[mw:Extension:JADE]] [19:52:08] 10[1] 10https://www.mediawiki.org/wiki/Extension:JADE [19:52:12] oh, there you are. [20:11:34] harej, The annotations concerning the difference between revision 456 and its parent revision would be stored at *Judgment:Diff/456* [20:28:54] ^ +1 [20:42:15] oops, typo [20:42:20] thank you! [20:47:05] I've got to do some work tonight so I'm going to take the opportunity to run some errands now. I'll be back online in 2 hours or so but I'll be mostly in a meeting. [22:13:17] harej: Have you thought about generalizations of ORES, by any chance? [22:13:31] I know it’s a bit outside the scope of what we’ve been working on, but I’m picking it up as a side project. [22:13:54] Basically, ORES has some rad machinery that should have been included in scikit-learn, so I’m exploring upstreaming it somehow. [22:14:31] Eventually, I’d like to have the reusable part upstreamed and the MediaWiki-specific machinery in its own package. [22:16:29] I haven't given any thought to how upstreaming of ORES might work. [22:16:50] OK cool, well holler if you ever feel inspired about that [22:17:34] It turns out, training models, serializing and setting up a web service is in high demand. [22:17:59] There’s an obvious strategic angle as well, we can get into that later… [23:16:06] 10Scoring-platform-team, 10Core-Platform-Team, 10MediaWiki-Special-pages, 10Wikimedia-log-errors: SpecialRecentChangesLinked::doMainQuery blocking database infrastructure - https://phabricator.wikimedia.org/T134976 (10Krinkle) Still seen at 10Scoring-platform-team, 10Core-Platform-Team, 10MediaWiki-Special-pages, 10Wikimedia-log-errors: SpecialRecentChangesLinked::doMainQuery blocking database infrastructure - https://phabricator.wikimedia.org/T134976 (10Krinkle) @Halfak @CCicalese_WMF Sorry to ask, but it seems both your teams moved the task... [23:18:21] term|CONT: no, I'm working on creating a new label-type in Wikilabels, a "session" label of multiple revisions [23:32:49] 10Scoring-platform-team, 10MediaWiki-extensions-ORES, 10editquality-modeling, 10User-Ladsgroup, 10artificial-intelligence: [Spec] Use `reverted` models in ORES review tool - https://phabricator.wikimedia.org/T146378 (10MMiller_WMF) Hey @Ladsgroup -- I'm also interested in the answers to Joe's questions,... [23:38:34] harej: halAFK: https://github.com/wiki-ai/ores-diagrams/blob/master/ores_dependencies.pdf [23:38:55] Is anything obviously missing?