[00:26:43] 10Scoring-platform-team, 10Core-Platform-Team, 10MediaWiki-Special-pages, 10Wikimedia-log-errors: SpecialRecentChangesLinked::doMainQuery blocking database infrastructure - https://phabricator.wikimedia.org/T134976 (10CCicalese_WMF) [09:19:20] PROBLEM - puppet on ORES-web02.Experimental is CRITICAL: CRITICAL: Puppet has 13 failures. Last run 3 minutes ago with 13 failures. Failed resources (up to 3 shown): Package[ldap-utils],Package[libnss-ldap],Service[nscd],Service[nslcd] [09:47:20] RECOVERY - puppet on ORES-web02.Experimental is OK: OK: Puppet is currently enabled, last run 1 minute ago with 0 failures [13:11:48] (03PS1) 10Sbisson: Store class 0 of models with more than 2 classes [extensions/ORES] - 10https://gerrit.wikimedia.org/r/445402 (https://phabricator.wikimedia.org/T199358) [13:30:25] 10Scoring-platform-team: Investigate srwiki goodfaith model, why is it so bad? - https://phabricator.wikimedia.org/T199355 (10SBisson) a:05Catrope>03None [13:59:04] 10Scoring-platform-team, 10Fundraising-Backlog: Machine Learning for Fraud Detection - https://phabricator.wikimedia.org/T190523 (10saurabhbatra96) Updates - * ML code snippets are being tracked here - https://github.com/saurabhbatra96/wmf-samplecodes * PR plots for various classifiers with the dummy dataset... [14:36:08] o/ [14:36:47] I've been working some extra hours the last couple days do I'm coming in a bit late today. Thought I'd hop on irc in case anyone needed something [15:02:51] (03PS10) 10Sbisson: Introducing QueryHelper [extensions/ORES] - 10https://gerrit.wikimedia.org/r/444252 (https://phabricator.wikimedia.org/T198748) [16:30:12] halfak: Good news about the scalability discussion, it looks like we have a confirmed meeting time next Tuesday AM. [16:30:40] I heard that. Was just talking to mark. I'll be in flight so I'll leave it to you. [16:31:52] halfak: I'll probably bug you later today to help refine my points. I already forget your killer argument about why we don't want JADE page-per-page... [16:32:29] "JADE page-per-page"? [16:32:38] I thought it was something sophisticated, but maybe it was just to point out that the page length will grow linearly with the number of revisions on the page [16:33:35] "page per page" being a barbaric shorthand for the schema where each JADE: page is a collection of judgments on every revision of that page, rather than a JADE page per entity being judged [16:34:27] awight: I will be joining the discussion if that's ok (sked jynus) [16:34:40] apergos: That's great, thanks for making the time! [16:34:46] happy to do so [17:00:28] halfak: hey, do you know if anyone is coming to the research lab today? There's a request to cancel conference rooms at 10 Pacific, if you think there aren't office people attending. [18:01:41] awight: meeting! [18:02:26] O_O [18:02:28] running [18:02:30] to meeting [18:03:04] :q [18:11:17] ^ that's how I feel too, sometimes [18:32:35] ty for the ping [18:35:49] hi awight! :-) [18:38:52] AFK to change locations. Back in an hour or so! [19:31:42] o/ [19:31:44] Looks like documentation time is all about Wikimania for me today [19:32:44] * awight grinds teeth trying to use scipy to plot an integral [19:33:00] Wachoo working on awight ? [19:33:48] that b.s. estimate of storage requirements [19:35:22] typical econometrics garbage: pulling parameters out of my butt like "4 years until full adoption", then making a meticulous plot of the results as if the initial assumptions are realistic [19:41:26] Gotcha. Making a logistic curve to get up to full capacity? [19:41:33] exactly [19:41:50] Dazzle them with graphs :) [19:42:26] I was going to graph * logistic curve of optimistic, organic adoption * integral showing total impact on pages and revisions over time, * alternative graphs showing how we can limit or even reverse adoption [19:47:30] blech [19:47:31] https://github.com/adamwight/jade-workflows/blob/master/adoption.ipynb [19:54:51] awight, I think you might be overcomplicating this. I like your graph :) [19:54:55] brb [19:58:33] Just thinking about math... that red line should hit (4.0, 100000) I believe [19:58:58] oooh [19:59:01] I'm an idiot [19:59:07] The graph is correct. [19:59:31] cool, now I can move on. [20:05:09] food [20:05:12] back in 45 [20:08:47] o/ [21:12:38] * awight groans [21:12:54] something's wrong. https://en.wikipedia.org/wiki/Wikipedia:Size_of_Wikipedia says that enwiki is only growing by 20k articles/month [21:13:29] ooh [21:13:32] argh [21:13:45] revisions vs pages. the whole reason this is a big deal. [21:13:50] * awight goes back to corner [21:15:57] (03CR) 10jenkins-bot: Localisation updates from https://translatewiki.net. [extensions/JADE] - 10https://gerrit.wikimedia.org/r/445502 (owner: 10L10n-bot) [21:18:01] screw these graphs [21:52:11] halfak: https://etherpad.wikimedia.org/p/JADE_scalability_FAQ [21:52:22] I'll come back to that tomorrow. [21:52:27] Woah. nice docs. [21:52:30] * halfak reads. [21:53:09] I think I'm missing a lot, please do jump in and either add or just point out which arguments need reinforcement. [21:55:10] ty! [21:57:40] I ditched the graphs fwiw, cos it was so speculative and ridiculous. [21:57:57] But I do think I should circle back and include some estimates in this FAQ [22:01:24] +1 [22:01:31] I think it looks good though. [22:05:12] I've added numbers [22:05:17] "How much storage do we need?" [22:16:46] I'm heading out. awight you're around tomorrow, right? [22:16:49] yep! [22:16:57] Cool! See you then :) [22:23:27] 10Scoring-platform-team, 10ORES: Document differences between features in Python and API - https://phabricator.wikimedia.org/T199485 (10awight) [22:23:52] 10Scoring-platform-team, 10ORES: Document differences between features in Python and API - https://phabricator.wikimedia.org/T199485 (10awight) p:05Triage>03Low [22:27:32] 10Scoring-platform-team (Current), 10ORES: Experiment with LIME integration for ORES, providing explanations for its predictions - https://phabricator.wikimedia.org/T196475 (10awight) [22:36:42] 10Scoring-platform-team, 10JADE: Updates to JADE diagrams - https://phabricator.wikimedia.org/T199486 (10awight)