[00:14:22] DarTar, I'm being told you left a comment on the phab ticket but now I can't find the comment :/ [00:20:05] Ironholds: https://phabricator.wikimedia.org/T91443#1094262 ? [00:21:48] ta [06:21:22] halfak: heeeyyyyyyyyyyyyy I SENT YOU AN EMAIL [15:16:59] morning [15:51:45] hi everyone. [15:58:15] hi one [16:04:54] hi two [16:05:01] :D [16:31:38] * Ironholds yawns [16:31:42] Today is going to be a v. long day. [16:47:12] Ironholds: But it's Friday! [16:48:21] guillom, and I have to find out how many pageviews www.wikipedia.org gets and where from [16:49:43] Ironholds: Pew. You can do it in your sleep. [16:50:07] guillom, except www.wikipedia.org is not a wiki [16:50:12] we have a pageviews definition for MediaWiki [16:50:27] right now I'm limited to "it didn't 404 and it's text/html, fuckit" anywhere else. [16:50:30] I feel like a luddite! [16:53:00] Ironholds: Come play with search keywords! It's more fun! We have cookies! [16:56:04] guillom, heh [16:56:10] this is necessary and influential for search! [16:56:36] leila:yt? [18:27:14] Ironholds: you’re an admin, use your new powers responsibly [18:28:33] DarTar, TaTar [18:32:16] ping me when everything’s ready to take down the interwebs [18:35:27] halfak: btw, I also have the right to approve OAuth applications now. [18:58:11] leila: I just found out about https://meta.wikimedia.org/wiki/List_of_Wikipedias_by_sample_of_articles/Absent_Articles . Mentioning it in case you didn't know about it either. [18:58:44] Ironholds: heh! [19:00:53] hi guillom. We hadn't seen that. thanks for sharing. We were interested in language pairs though, basically, for each language pair (x,y), what are the articles that are available in x but not in y. [19:01:09] understood [19:03:25] guillom: the approach they've taken to generate the list is interesting. They've come up with a sample list of article that should exist in all Wikipedias, and they list the missing part of the sample for each Wikipedia. [19:04:08] It's not a trivial problem to say what needs to be there in every Wikipedia. I'm interested to learn how they made the call. [19:04:24] leila: Yes. The "list of articles every Wikipedia should have" in one of the things I'm considering for keyword monitoring. [19:04:38] leila: It's a manually curated list of ~1000 topics. [19:05:07] The calls have been made through painful discussion AFAIK :) [19:05:11] I see. that's interesting. It may be useful for our recommendation system, too, as one of the parameters to consider when ranking. [19:05:38] ooy! [19:33:17] Ironholds: if you’re interested in sharing the agents link more broadly, you should probably do it by noon / early afternoon today. Anything posted after that on a Friday afternoon PT will get little to no traction (and in that case you should probably wait until Monday) [19:45:17] DarTar, yep, will do [19:45:20] sorry, was in a meeting [19:46:05] also, jesus hell the datahub gets more ugly every year [19:51:47] DarTar, so when do you think you'll recover from the PTSD of this release and I can talk about a circadian patterns dataset? ;p [19:51:55] I want to call it "tempus edit" [21:30:32] hey Ironholds [21:30:39] yo! [21:30:46] just back at my keyboard [21:30:59] datahub: heh, yeah - but at least it does markdown [21:31:09] did you manage? [21:31:37] yup! see your inbox [21:31:47] let me check [21:32:00] * YuviPanda jumps at DarTar [21:32:05] have you seen jupyter.wmflabs.org [21:32:30] YuviPanda: I had a super-quick chat with Ori about it yesterday, I have’t yet [21:32:56] YuviPanda: did you send a note to analytics at all? [21:33:06] DarTar: not yet. Need to work out more kinks. maybe next week [21:33:11] splendid [21:33:40] you should at least ping a few other folks, including ellery and halfak, as early adopters [21:34:54] Ironholds: couple of things, [21:35:09] if you change the format type to text/tsv it should enable the preview [21:35:14] (I know, don’t ask) [21:36:31] okie [21:37:56] DarTar: yup, halfak is first point-of-ping [21:38:35] Ironholds: second, I would copy the docs from shiny [21:39:08] YuviPanda: good, definitely add ellery too (heavy NB user) [21:39:18] sure [21:39:19] DarTar: ooo, will do [21:40:27] Ironholds: and the last thing I would add is the period the data refers to [21:40:40] sure [21:40:48] with that, I think we’re ready to roll [21:42:21] DarTar: have emailed ellery [21:42:28] done and done [21:42:36] YuviPanda: great [21:43:31] Ironholds: awesome [21:46:38] cool! [21:49:45] Ironholds: tweet it (with some plain English description consumable by ordinary human beings ;) [21:50:49] the app or the datahub or the figshare? ;p [21:51:17] I’d say the app (and figshare via DOI if you have room for it) [21:52:11] DarTar, https://twitter.com/quominus/status/573964292929679362 [21:52:15] the DOI is in the app [21:52:28] great [22:01:55] DarTar: did we cancel the backlog grooming? [22:08:45] leila: grace did: she had a conflict and today half of the team is off, we’ll continue on Tuesday [22:08:58] you get an hour back! [22:09:02] woo-hoo [22:09:18] okay, we have quite a few items in backlog, DarTar, but yeah, I won't say no to an hour back. ;-) [22:10:10] leila: that’s right but nothing particularly urgent or that we haven’t already discussed [22:11:01] yeah, you're right, DarTar. [22:11:12] unique client feedback time then [22:11:54] yes, I’ll look into that too in a moment [22:36:12] hmn. [22:36:16] how would one cite a preprint? [22:50:23] Ironholds: you usually don't, go with the Working Paper if there is a such a thing [22:50:40] leila, wassat? [22:51:10] sometimes researchers keep track of their research as a working paper, it's an incremental publication basically [22:51:18] they usually have to have a link to it on their websites. [22:51:51] and if the authors are in Universities, then they sometimes have file numbers associated with those working papers [22:52:22] so it goes like: name_of_the_project, authors, Working Paper, name_of_the_University, Working Paper number [22:52:23] aha [22:52:34] okay. How would I cite /my/ preprint? ;) [22:52:36] not everyone does it though. [22:52:40] hahaha! [22:52:41] :D [22:52:43] you don't! [22:52:58] I don't know. it's different in every field. [22:53:14] is it already accepted? [22:56:03] Ironholds: you can totally cite an arxiv preprint, don’t listen to leila [22:56:49] ow of course, is it already on arxiv, Ironholds? [22:57:17] I thought it's not published anywhere, DarTar. [22:57:42] erm, right, a preprint that lives on someone’s hard drive would not be citeable :p [22:57:53] :P [22:58:18] DarTar, cool! [22:58:21] leila, yes! [22:58:28] I want to cite my paper with Halfak in my paper with Scott [22:58:34] so I can paper while I paper and paper in my paper. [22:58:36] * Ironholds nods [22:58:38] mmn. Paper. [22:59:19] ow okay. so then just cite it, Ironholds. :D [22:59:28] buffalo’ing papers is a severe violation of the academic code of conduct [23:00:32] Ironholds: seriously, are you using bibtex? [23:00:36] DarTar, yep [23:00:38] leila, okie! [23:01:00] See https://github.com/Ironholds/MobileUserInterfaces [23:01:09] (https://github.com/Ironholds/MobileUserInterfaces/blob/master/Paper/Figures/new_contributor_revert_rate.svg makes me sad) [23:01:28] http://tex.stackexchange.com/questions/3833/how-to-cite-an-article-from-arxiv-using-bibtex [23:02:35] Ironholds: hmm are these error bars? I don’t quite grok the plot [23:03:08] DarTar, they are! They're just tiny because it was from permutation testing and resampling [23:03:20] and, well, turns out the rate really doesn't change much between 1000-edit chunks [23:04:18] I lack some context, I should read the whole thing [23:07:32] DarTar, or I'll just discuss it at a research meeting when Scott and I have gone back and forth on the methodology! :D [23:08:08] but unless my sciencing is way off: mobile is not substantially more geographically diverse than desktop, and editing is not really linked to probability of internet access, and mobile users are reverted more, have shorter sessions and make fewer edits. Which, duh, from our POV, but interesting for an outside audience. [23:08:24] And I think the lack of increasing diversity is something it's worthwhile having visuals to hammer in with, because we need to work on that. [23:08:48] If we get the next billion users and they can't edit, we have a problem. Hopefully wikigrok shows a more diverse userbase and is more accessible. [23:09:11] Ironholds: that svg doesn't work in GitHub nor gwenview for me; but imagemagick gave me a PNG :) [23:09:19] I have pretty conflicting results showing that early editing activity of newbies is significantly higher than on desktop [23:09:28] 9 % revert rate is remarkably low ;) [23:09:50] DarTar, huh! Let's compare methodologies? [23:09:52] when I'm less tired [23:10:05] ot's 6pm and I'm still dealing with legal's question [23:10:43] DarTar, wait, by higher activity do you mean more edits? [23:11:00] then yeah, that's not contradicting. [more edits in {time}] versus [more edits in {session}] [23:11:28] Ironholds: I’m referring to the activation rates (and I have more work in the making about new active editor rates, as a spinoff of a request from Erik) [23:11:38] I guess so [23:11:50] DarTar, oh cool, let's chat next week then? [23:11:55] sure [23:12:02] amusingly that'd make mobile editors match mobile readers [23:12:09] shorter sessions, fewer pages per session, but more sessions [23:13:45] the interesting finding from the early analysis I did of active editor data is that last year for the first time ever (i.e. since 2001) we didn’t see the typical seasonal drop in December, and after removing the seasonal component we see an indication of a trend inflection in 2014 [23:14:03] now “no seasonal drops in December or weekends”, does that ring a bell? [23:14:51] (I don’t want to give too much importance to active editors, which we all know is a pretty coarse metric, but still) [23:17:58] DarTar, it does! [23:18:00] :D