[03:17:34] o/ [03:17:42] don't ask why I'm up so early [04:45:21] greetings! [04:45:53] i just got done reorganizing literally all of my files across all my devices, clouds, etc. [04:45:56] going back to 2006 and earlier [04:52:44] 10Jade, 10Scoring-platform-team (Current), 10MediaWiki-Configuration, 10Patch-For-Review, 10User-Ladsgroup: Rename JADE->Jade in beta cluster configuration - https://phabricator.wikimedia.org/T212182 (10Ladsgroup) The only way I see forward here is to branch both extensions for a while, deploy the gerrit... [08:52:05] 10Scoring-platform-team (Current), 10Wikilabels, 10articlequality-modeling, 10artificial-intelligence: Build article quality model for Galician Wikipedia - https://phabricator.wikimedia.org/T201146 (10Theklan) We have a similar "problem" in Basque Wikipedia, and is because some articles are not rated consi... [13:33:28] I think we can explore some of groceryheist's questions with mechanical turker labels. [13:33:41] It might even give us a *better* window into bias. [13:36:47] 10Scoring-platform-team, 10Serbian-Sites: Investigate srwiki goodfaith model, why is it so bad? - https://phabricator.wikimedia.org/T199355 (10Halfak) An etherpad is directly editable. You should be able to just type into it. [13:37:51] 10Scoring-platform-team, 10Serbian-Sites: Investigate srwiki goodfaith model, why is it so bad? - https://phabricator.wikimedia.org/T199355 (10Halfak) So, as it stands, more than half of the items labeled badfaith are actually goodfaith upon review. I'll look into these labels to see if I can see some sort of... [13:48:08] I just found a "null" goodfaith value that was converted to "false". That's bad. [13:48:13] * halfak digs. [13:52:43] halfak: i mean i think thats a fair default [13:52:58] Na. Should be "true" as the default. [13:53:02] if it cant determine the value imho it should default false [13:53:02] If you don't know, AGF. [13:53:11] :P [13:53:17] halfak: i think AGF that is unknown should be determined by a human then [13:53:29] therefore marked false [13:53:34] Well we have passed that stage and the human left it blank. [13:53:51] then replace the human :P [13:54:03] Also, all of the data I'm looking at suggests a second reviewer would most likely decide it was goodfaith. [13:54:15] We're working on that! Shhh. Don't tell anyone. [13:54:19] halfak: thats usually so [13:54:25] halfak: you just kinda told everyone [13:54:32] plus its logged now [13:56:07] halfak: maybe getting community consensus would be a good idea? [13:58:14] Uh. This is a pretty low level detail. [13:58:20] Also it was documented as going the other way. [13:58:35] I think doing a consensus process for this would be a little heavyweight. [14:00:35] I dont think a null value in this context should go true, should go false [14:06:52] I'm done for the day, see you later (by the way, May 1st is public holiday in Germany and after that I have overtime leave) [14:08:47] 10Scoring-platform-team, 10Serbian-Sites: Investigate srwiki goodfaith model, why is it so bad? - https://phabricator.wikimedia.org/T199355 (10Halfak) It looks like a lot of the edits that were labeled "badfaith" but that we have no re-labeled "goodfaith" were saved by @Zoranzoki21. That might be simply becau... [15:03:48] 10Scoring-platform-team (Current), 10Wikilabels, 10editquality-modeling, 10Spanish-Sites, and 2 others: Create editquality campaign for Spanish Wikiversity - https://phabricator.wikimedia.org/T209670 (10Halfak) We can definitely play with the "trusted edits" set. @Lsanabria, are there any user-rights on S... [15:32:34] Amir1, hare: Standup! [16:04:27] 10Scoring-platform-team (Current), 10Wikilabels, 10editquality-modeling, 10Spanish-Sites, and 2 others: Create editquality campaign for Spanish Wikiversity - https://phabricator.wikimedia.org/T209670 (10Halfak) a:05Ladsgroup→03Halfak [16:07:38] 10MediaWiki-extensions-ORES, 10Scoring-platform-team (Current), 10Patch-For-Review, 10User-Jdlrobson, 10User-Ladsgroup: Remove SpecialContributions::getForm::filters hook call - https://phabricator.wikimedia.org/T219238 (10Halfak) [16:09:20] 10Scoring-platform-team (Current), 10Wikilabels, 10articlequality-modeling, 10User-Sebastian_Berlin-WMSE, and 2 others: Build article quality model for svwiki - https://phabricator.wikimedia.org/T202202 (10Halfak) @Gilles, are you still working on this task? [16:58:31] halfak: what happens if I can’t join in today’s panel? I realized I have a meeting right before which is time I would’ve used to prep [17:40:41] hare, that would be bad. Do you already have your questions drafted? I guess I could ask them for you and we'd just need to do without your notes. [17:41:20] I’ll prepare quickly. I think I have my questions ready anyway :) [17:50:55] OK great :) Looks like the rest of us have questions in the doc. Feel free to just get in order and add the questions afterwards. [17:51:06] I'm in the meeting now if you want to join up and chat. [17:51:08] hare, ^ [18:54:51] query halfak [18:54:54] doh [18:55:48] RESPONSE: query not found. [18:56:06] query is a verb, not a noun [18:58:37] RESPONSE: ERR42 no question is present. [18:58:57] :) [18:59:16] Lunch! [18:59:32] Woops. [18:59:36] Nevermind. I don't get to eat yet. [19:59:36] finally! [21:24:50] ...That reminds me, I forgot to eat lunch [21:25:02] I have become one of those people who forgets to eat. [22:33:59] hi wikimedia ai folks -- feature injection question if anyone has a moment. Is there any reason it might not be working today? I would almost swear I'm using the same syntax as I've used in the past, but it's not working for me [22:34:11] goodfaith/damaging in particular [22:39:13] here's an example: https://ores.wikimedia.org/v3/scores/enwiki/213315011/damaging?feature.revision.user.is_anon=true and https://ores.wikimedia.org/v3/scores/enwiki/213315011/damaging?feature.revision.user.is_anon=false [22:39:52] once upon a time, I ran these two, and got different answers for anon true vs anon false [22:40:03] but today, I am getting the same answer for each [22:40:17] Woah. [22:40:19] Looking. [22:41:05] This is very strange. You're sure it was working yesterday? [22:41:08] kaylea, ^ [22:41:46] It wasn't yesterday that I looked at it, it was a while ago [22:41:55] Oh I misread. [22:42:05] I think I might know what's up. [22:46:56] kaylea, I'll see to getting a patch together tonight, but it won't be deployed until tomorrow at the soonest. [22:47:03] I can ping you when it is available. [22:47:25] ok, that's great [22:47:41] some kind of caching issue? [23:03:35] maybe is the same problem, but another oddity is when I'm seeing different scores than I got in my previous run -- anon true and anon false are not identical, but they don't match the old run either. example: https://ores.wikimedia.org/v3/scores/enwiki/202821899/damaging?feature.revision.user.is_anon=false and https://ores.wikimedia.org/v3/scores/enwiki/202821899/damaging?feature.revision.user.is_anon=true [23:03:55] Ho old is the odl run? [23:04:13] Oh! I see. It's probably related to our cache. [23:04:33] Which means we'll likely need to clear it :| [23:05:56] the last time I touched this stuff was april 10th, so sometime between now and then, something changed maybe [23:06:58] Yeah. We had a deployment in there. [23:07:32] Essentially, our test needed to work with a dependency tree that is more than 2 levels deep. Because it didn't, we didn't catch this bug. [23:07:49] ^ The dependency tree we use to solve features for a model. [23:12:13] hm, interesting. glad it was so quick for you to spot [23:31:20] 10ORES, 10Scoring-platform-team (Current): Non-root features no longer being injected. - https://phabricator.wikimedia.org/T222121 (10Halfak) [23:31:28] 10ORES, 10Scoring-platform-team (Current): Non-root features no longer being injected. - https://phabricator.wikimedia.org/T222121 (10Halfak) p:05Triage→03High [23:37:24] https://github.com/wikimedia/ores/pull/327 [23:37:28] There we go. [23:37:38] I'm off. Have a good one, folks. [23:39:26] wikimedia/ores#1342 (root_features_fix - bae07fc : halfak): The build passed. https://travis-ci.org/wikimedia/ores/builds/526210318 [23:39:39] Yer damn tootin'