[13:53:13] * halfak groans [13:57:22] GODO MORNING [14:02:51] o/ ottomata [14:02:52] :) [14:03:36] GOOD MORNING HALFAK [14:03:38] HOW ARE YOU TODAY [14:03:48] halfak: this is super duper WIP POC, but comments will be helpful [14:03:48] https://github.com/wikimedia/restevent/pull/5 [14:03:56] Better than yesterday. Still clearing out the sick stuff. [14:04:11] ottomata, anything in specific you want me to look at? [14:05:47] Hmm... Never had hexchat die like that before. [14:05:49] o/ ottomata [14:05:51] sorry, back now [14:06:10] halfak: uh, well, i mean, they are schema designs for things you really care about [14:06:16] so you know, whatever you think is relevant [14:06:17] "article edit" dear god. [14:06:20] good-o :) [14:06:22] exactly :) [14:07:49] halfak, so guess what I built this morning? :) [14:10:33] You finished off the talk page parser! [14:11:30] halfak, no, I built a dashboard that automatically generates and displays the proportion of pageviews that are directly search engine sourced and which SE they come from [14:11:34] which is not as cool, I'll admit :( [14:12:00] Still pretty cool. Link? [14:12:30] not live yet; needs deployment! But I can send you the repo in ~30s [14:16:49] https://gerrit.wikimedia.org/r/#/c/249126/ [14:19:26] Boo. No code still does not run on meat CPU in my head. [14:19:42] ottomata, thanks for directing me to this. What a mess! [14:22:01] ha, it is PoC, that's why you are here! [14:22:37] halfak: you should follow this discussion too [14:22:37] https://phabricator.wikimedia.org/T116247 [14:22:47] we are mostly debating event metadata fields there, not actual content ones [14:23:02] It is hard to follow all of this. [14:23:08] just those two! [14:23:11] don't worry about the rest [14:23:16] you just keep your eyes on schema design, ok!? :) [14:23:19] It would be nice if there were clear interaction points. [14:23:31] E.g. we want you to review the schema by [14:23:52] well, ok. i mean, its PoC at the moment, so you don't have to do it now if you don't want to [14:23:57] but, the sooner the better [14:24:07] RFC meeting for EventBus is tomorrow [14:24:18] i think things will directionalize (!)a little more after that [14:24:20] Well, "following" takes like 10x as much effort as having someone let me know when they'd like me to look at a thing. [14:24:22] at least, that's my plan :) [14:24:42] halfak: just follow this 1 phab ticket :) [14:24:48] if there is anything else, i will poke you [15:22:28] Good morning, researchers. [15:22:34] o/ guillom [15:22:56] hey halfak [17:13:00] etherpads are impossible to work with today :-/ anyone managing to edit them? [17:28:18] Bah! DarTar was here for 2 minutes and I missed it. [17:30:16] you must be swift to catch the wily DarTar [17:32:15] beware the JabberTar, my son, the cites that catch, t-tests that stun [17:40:12] jabbertar was vanquished by carlos dioxide [17:50:57] OMG it's DarTar. Get 'em, folks. [17:51:05] ;) [17:51:19] End up deciding on some IRC office hours? [17:51:42] I’ve been around more often than you may have realized [17:51:49] nicknames FTW [17:51:52] ;) [17:51:57] LAME [17:52:13] * halfak needs to learn DarTar's nicknames [17:52:35] BTW, I'm not having trouble with etherpad. [17:53:11] halfak: *shrug* I gave up after 30 minutes of trying from multiple browsers [17:53:20] Ha! Got one [17:53:24] Oh wait [17:53:26] lol [17:53:32] I blame WMF wifi [17:53:58] I got some weird server-side errors, doesn’t look like a wifi issue [17:54:04] Weird. [18:06:32] halfak, guillom: success! the UW network came through. [18:06:42] WooOT [18:07:05] :P [18:07:09] :D [18:07:30] * DoorTar digs into celery to figure out how we'd check queue length [18:08:00] * RoadTar checks research candidate's take homes [18:18:55] Amir1: I need to meet with J-Mo this afternoon before writing the email. I'll initiate the email after that meeting, so it will be morning your time before there is a conclusion. [18:37:58] halfak: DarTar I'm trying to go through and finish up grading all the take homes now [18:38:01] 2 down! [18:39:17] sorry about it having taken this long [18:45:35] YuviPanda: thanks, man [18:46:21] give priority to those with no assessment yet, feedback from either you or halfak is enough [18:51:37] goddamit he's gone [18:54:14] halfak: ok I think all of them have a review from you or me. I couldn't find the ones you said you marked as 'need review from me' - I couldn't find such a thing on greenhouse [18:54:47] YuviPanda, already interviewed those poeple :P [18:54:54] whelp [18:55:09] well I'm in a better shape now and will try to not fall behind so much on tasks [18:55:12] sorry again [19:24:59] halfak, J-Mo1, are you familiar with Letters to a Pre-Scientist? [19:25:54] No worries YuviPanda. :) [19:25:58] Ironholds, negative [19:26:19] halfak, so it's a programme whereby professional scientists have middle schoolers who are into science as pen-pals [19:26:38] and said middle-schoolers can ask questions and generally exchange knowledge and advice [19:26:43] that's awesome and also adorable [19:26:47] it's pretty cool - do recommend as a way of spending an hour a week [19:26:58] http://www.prescientist.org/ [19:27:09] +1. Seems like this could be some good volunteer work. :) [19:27:47] It's targeted particularly at students who are from disadvantaged backgrounds. It's frickin' GREAT. [19:27:53] ~~sign up~~ [19:35:08] hey halfak: do you have time to go through Juliet’s questions? I have a batch of candidates to look into in the next hour or so [19:35:35] Sure. I'll have a look before the 401k stuff starts [19:36:44] cool, thx (I can do personalized 401k advice in return :D ) [19:54:29] DarTar, looks like you guys want to rename the whole project :P [19:54:46] If you didn't like the name ORES, you should have said so during our initial bike-shedding. [19:59:11] halfak: no, not really. I’m around if you want to chat for a moment [19:59:25] ORES is the name of the web service. [19:59:31] Not sure how we're going to get around that. [19:59:40] Our team/project is called Revision scoring. [20:00:59] Also, the current tools in Wikipedia *do* fall under Zeynep's critique. [20:01:12] ORES is no less a subjective algorithm than Huggle/ClueBot/Facebookfeed [20:01:15] There's also WikiProject X. WikiProject X is a thing that uses ORES. [20:01:19] using two names in the blog post which are jargon-heavy is not going to help, this is why I suggested we focus on either instead. I’ve always seen the project advertised as revision scoring and that’s where my preference goes [20:01:32] :D We're featuring WikiProject X, harej :) [20:01:51] DarTar, but the service is called ORES. Is it less confusing to invent a new name? [20:01:55] Also, it's in the URL [20:02:03] WikiProject X is a brand. Our products are called wikiproject_scripts and WPX UI. [20:02:13] halfak: featuring it in what way? [20:02:23] current tools falling under Zeynep’s critique: it’s totally not obvious from a single paragraph where we reference work that is not accessible (a talk) [20:03:01] 4:02 PM WikiProject X is a brand. Our products are called wikiproject_scripts and WPX UI. << more confusingly, the bot that does most of the legwork is Reports bot, but SuggestBot (technically a third party service altogether) also pulls some weight [20:03:31] ORES is no less a subjective algorithm than Huggle/ClueBot/Facebookfeed <— agree, that’s why I find that section too compressed to do its job [20:04:04] DarTar, options: #1 extend it, #2 drop it [20:04:25] DarTar, from the abstract: "lgorithms, often aided by big data, now make decisions in subjective realms where there is no right decision, and no anchor with which to judge outcomes. What is good? What is relevant? What is important? Who is right? What is desirable? What is valuable?" [20:04:37] revscoring is not a new name, right? In any case, if you feel strongly about ORES I’d suggest: 1) expanding the acronym, 2) removing as much as possible the “also known as” references [20:05:05] subjective problem spaces != subjective algorithms [20:05:18] DarTar, that's literally her definition [20:05:32] I am just pointing out that that sentence is completely obscure, no matter if it’s someone verbatim definition [20:05:59] What's obscure? I've never heard someone raise that complaint of this term before. [20:06:41] this sentence: “While artificial intelligence may prove essential for solving problems at Wikipedia’s scale, subjective algorithms can also dehumanize and subjugate people and obfuscate inherent biases” [20:07:58] How about "algorithms that replicate subjective judgements" [20:08:24] yeah better, replicate/amplify [20:09:00] I don't think that feedback loops are something we can discuss in this space. [20:09:05] I'm realizing now that you' [20:09:17] ve probably never seen me make this argument before. [20:09:28] * halfak curses the scheduling gods that keep DarTar out of showcases. [20:11:39] I'd also like to give an introduction to standpoint epistemology in this section. [20:11:40] yeah it still requires a lot of context, my point is that in a very compressed format this comes across more as “oh Zeynep, we love your work and will give you credit for these ideas” but at the cost of garbling the narrative. I literally read that sentence as suggesting that pre-ORES algos *on Wikipedia* are “dehumanizing and subjugating people”, which requires a lot more space/effort to articulate [20:12:03] DarTar, that's the conclusion of the Rise and Decline paper/. [20:12:29] I haven’t seen evidence that ORES fixes that problem (yet) [20:12:46] DarTar, no, but the way it goes about *trying* to fix the problem is interesting. [20:12:58] Or so my collaborators and audience keep telling me. :P [20:13:01] yeah but that’s not what that sentence conveys [20:13:01] Did I reactivate this user after 5 years? ;) https://en.wikipedia.org/wiki/Special:Contributions/Terry0051 [20:13:24] and the fair answer is that we don’t know how it’s going to turn out [20:13:47] DarTar, we don't need to know to talk about it. [20:13:47] well the problem is right into its name, objective? uh [20:13:50] it’s an interesting problem, as you noted, to use open algorithms, we may end up exacerbating the problem [20:14:12] DarTar, indeed. If you take Zeynep at face value, you might assume we would. [20:14:20] "By building this service and making it open we’re purposefully making space for new voices to engage in the conversation about how wiki tools should work." [20:14:40] now, *that* I like [20:14:50] That's in the draft :P [20:14:58] I know and approved it [20:15:17] gotta prep for upcoming meeting now, sorry dude [20:15:34] pushing candidates and blog posts to later in the day [20:15:43] let’s catch up later [20:16:03] kk [20:22:41] halfak, in what sense are you featuring WikiProject X? [20:23:05] harej, screen shot of Assessment tasks with predicted class and a link to the WikiProject X project. [20:23:10] Who is Zeynep? [20:23:35] Zeynep Tufekci is a public scholar who is notable for studying the conversation patterns around the Arab Spring. [20:23:50] thanks! [20:23:55] She's recently gotten press for discussing the social problems causes by Facebook's feed. [20:24:15] Specifically, she calls for great caution around "Subject Algs." [20:35:12] Nemo_bis, missed your comment re. "objective". In a way, that name isn't exactly seriously intended. The service makes subjective judgements, but does so based on objective fact and probability. [20:35:30] Is this good? Well, 95% of good things look like that, so yes. [22:31:25] J-Mo1 I'm few min late [22:31:26] sorry [22:31:42] no prob lzia [22:41:33] lzia: OK :) [22:41:34] thanks