[18:23:42] good morning [18:33:13] Hi MissGayle! All set for metrics very early I see [18:46:00] * James_F waves ahead of the Metrics meeting. [18:48:03] James_F: Are you IRC czar again? [18:48:47] (I'll be the IRC liaison today.) [18:51:59] OK, let's hope I don't fall off the network again. [18:52:05] Trying to be!:) [18:52:12] Thanks James_F [18:52:26] MissGayle: Always a pleasure. [18:56:08] Due to start in 4 minutes' time. [18:57:12] Today's theme is going to be the Global South. [18:58:16] Presentations from Siko, Amir and Asaf, I believe. [18:58:29] As normal, we'll have questions at the end. [18:58:43] But please ask them as they occur to you, and I'll proxy them (or try to answer them then and there). [18:59:07] And of course, other members of the channel may know, too. :-) [18:59:39] So what can we troll James_F on today? [18:59:52] Starting now. [19:00:13] Initial welcome is from Gayle. [19:00:21] Stream lag as usual [19:00:31] got the stream [19:00:50] Good to hear. [19:00:51] Welcome, everyone. :) [19:01:13] would it be possible to clone this channel to #mediawiki-office? I keep going there :( [19:01:39] Lot of great milestones, including Android app and ContentTranslation. [19:01:44] marcoil: We could set it up with a +F forward. [19:02:02] Damon talking about the upcoming weekly VisualEditor triages [19:02:13] James_F: :) [19:02:19] Now Asaf leading the main presentation sequence. [19:03:12] Yay, technology. [19:03:16] The presentation not loading gives us a sense of how crappy internet connectivity can be in the Global South :) [19:03:19] We're currently trying to get Google Sheets to load. [19:03:23] guillom: * Internet. [19:03:24] But yes. [19:03:50] James_F: WMF Comms docs say we don't capitalize Internet at the WMF :P [19:04:01] guillom: However, spelling says otherwise. [19:04:05] :D [19:04:10] Are you still holding out for e-mail too? :) [19:04:25] superm401: Of course. [19:04:44] guillom: Siko's slides say "Internet", for instance. [19:05:05] 20% is actually more than I expected. Good to see progress there. [19:05:12] Because it's *correct* [19:05:42] Rising terminal? [19:10:40] .. [19:11:33] James_F: spelling says no such thing :P [19:11:54] Internet is a proper noun, you capitalize proper nouns [19:11:57] jamesofur: I meant real spelling, not your derivative, "spilung". [19:11:59] ;-) [19:12:07] I stand by my statement :) [19:12:11] marktraceur: Indeed. I use several internets to access the Internet. [19:12:11] how does audio sound on the remote side? [19:12:20] levels? [19:12:26] "spelling" is not a reliable source [19:12:27] cajoel, fine. [19:12:31] marktraceur: I would agree you capitalize proper nouns, I would disagree that it is a proper noun [19:12:35] How about we talk about something else? [19:12:42] superm401: Like, the meeting. [19:12:45] cajoel: Pretty good [19:12:48] New goal: 100+ Wikipedias with 100,000+ articles? [19:13:28] can you hear the new speaker? [19:13:36] Yes [19:13:37] Mmhmm. [19:13:38] thx [19:13:50] when I've been remote speakers like this have tended to be even better [19:13:50] "Not reaching the majority, not reaching every single human being" [19:13:57] Q: Are we reaching the majority of "internet users"? [19:14:13] she sounds a little digitized -- she might be on a slow connection. [19:15:05] halfak: Assuming http://www.emarketer.com/Article/Internet-Hit-3-Billion-Users-2015/1011602 is even half correct, presumably not? [19:15:19] Gotcha [19:15:39] This may or may not be the map we're supposed to be seeing: https://wikimediafoundation.org/wiki/Wikipedia_Zero#mediaviewer/File:Map_of_Wikipedia_Zero_countries_Jan_2015.png [19:15:42] halfak: But I imagine they have real data. [19:15:45] Internet users != all users that matter, but they do represent who we could reach with our current methods of distribution. [19:16:29] halfak, we have experimented with distributing to non-Internet users in various ways (DVDs, Kiwix, etc.). [19:16:36] Some of these have worked from what I've heard, but nothing large-scale. [19:16:50] halfak: I like the characterise "0.5bn MAU" as us failing to reach 2.5bn users each month. [19:16:56] Plus books that republish Wikipedia content; not sure how widely used these are. [19:17:58] graphics are boring, anyway. :-p [19:18:09] "Many people are not aware that it's an editable encyclopedia" [19:18:09] :) [19:18:14] Once Google brings Internet to everywhere using balloons, we won't have to worry about non-Internet users! [19:18:15] True in Global North too. [19:18:29] abartov: True Wikipedians believe in the power of words, I suppose. :-) [19:18:32] superm401: yup [19:18:44] James_F: pwecisely. [19:18:51] guillom, and Eon Musk in prep for Mars people. [19:19:09] http://www.telegraph.co.uk/news/worldnews/northamerica/usa/11353782/Elon-Musk-announces-space-Internet-plan.html [19:19:14] Latency may be an issue for Mars [19:19:37] guillom, luckily we don't have collaborative real-time editing, so latency is not a problem. ;) [19:19:44] lol [19:19:48] heh [19:20:09] superm401: Yet. :-) [19:20:35] Yeah. It's my top desired feature for 5-year plan. [19:20:46] I guess we'll need to transition to an actually distributed infrastructure when we start disseminating througout the Solar system/ [19:21:32] What's the name of our transliteration input technology again? [19:21:32] guillom: It's on my roadmap, ish. [19:21:40] superm401: jQuery.IME? [19:22:06] we need lagrange point caching centers. [19:22:17] Some research suggests that synchronous editing is less efficient than async collaboration. [19:22:18] That's probably the new one. I remember an older one that didn't have jQuery in the name. [19:22:27] andrewbogott: Heat venting will be an issue. [19:22:39] * halfak digs for cite [19:22:43] superm401: That's the one that the Language Engineering team built a couple of years ago. [19:23:01] yeah, the fans won’t do much in a vacuum [19:23:05] halfak: Interesting. It sounds mildly unsurprising. [19:23:15] andrewbogott: And direct Solar glare won't be ideal. [19:23:44] James_F & superm401: citation: http://paul.is/pubs/andre-coordination-chi2014.pdf [19:23:51] halfak: Thanks. [19:24:39] TL;DR: Territoriality (or feelings of it) are strong when you can see someone else typing. [19:24:56] My sense is that a better goal might be to improve merge-saving pattern. [19:25:00] * James_F nods. [19:25:07] We have plans in that area too, thankfully. [19:25:14] I'm stoked for that. [19:25:15] :) [19:25:23] James_F: Through sentence-level versioning? [19:25:31] Hopefully technically easier too. [19:25:41] Narayam (https://www.mediawiki.org/w/index.php?title=Extension:Narayam&oldid=739989). It's apparently discontinued, though. [19:25:48] i have some interest in that (sentence-level versioning) but that’s probably a way off :D [19:25:49] guillom: Yeah, and DOM-level diffing for changes. [19:25:55] Great [19:25:56] brion2: Yup. :-) [19:26:15] * James_F shuts up before he promises something TrevorP|Away isn't OK with delivering. :-) [19:26:37] * brion2 tosses James_F some brief notes on the insane: https://www.mediawiki.org/wiki/User:Brion_VIBBER/2015_scratchpad [19:26:42] do we attempt to synchronize edits across all *pedias? [19:26:59] eg. Global event, gets updated one place... does that feed out to others? [19:27:04] brion2: Nice. [19:27:10] cajoel: manually [19:27:17] cajoel: No, though edits to Wikidata are pushed out. [19:27:21] There is no canonical Wikipedia. [19:27:29] on purpose [19:27:39] Any article can be started anywhere (assuming it complies with their policies), and optionally translated from anything to anything. [19:27:43] "truth" isn't a real thing in many instances :) [19:27:46] And there is no requirement to translate. [19:28:01] A lot of articles about the same topic are not in fact translations, which is fine; a lot of others are. [19:28:17] cajoel: i think over time we’ll migrate to the “data” living in machine-readable wikidata attached to the text in the articles [19:28:18] But encouraging translations is a great way to help share knowledge at a lower level of effort than starting from scratch. [19:28:20] does a citation need to be in the same language? [19:28:32] cajoel, no. [19:28:40] But of course, people have a tendency to prefer it. [19:28:54] It's not just Low-bandwidth friendly UX" [19:29:09] Facebook also probably has a lot closer data centers even for many places in the developing world. [19:29:21] superm401: yes! [19:29:34] superm401: Yeah; before we do Lagrangian DCs, we should do non-Western ones. [19:29:35] true! we have really bad round-trip times to central asia [19:29:38] James_F, yep, I think translations are great. Just pointing out that's not the only approach. [19:29:46] drop a cache pop [19:29:51] brion2, and probably even places with more internet users than that. [19:30:06] * James_F nods. [19:30:15] Anyone who wants to dig more into these citation issues, check out Heather Ford's work. [19:30:17] long ago we had a small cache center in south korea (thanks yahoo for the hosting!) but we outgrew it [19:30:36] brion2: Do we have a list of good places to add new caches? [19:30:50] Here's a good entry point for Heather's work: http://www.opensym.org/wsos2013/proceedings/p0203-ford.pdf [19:31:05] James_F: i think the ops guys probably have some recommendations, but i don’t know if legal’s vetted anything else outside US/NL [19:31:33] Wow, that's a great first revision. [19:31:33] brion2: I'd imagine HK, ZA, QT and a couple of others. [19:31:55] CSD (Criteria Speedy Deletion) A7 has always been heavily misused. [19:31:56] Note that the average time between edits is 7 minutes [19:32:12] So when you CSD that fast, you probably edit-conflicted someone. [19:32:24] It was originally for articles that did not even (implicitly) claim the subject was important, which is very rare. [19:32:34] It's expanded over time, both in interpretation and in text. [19:32:51] "Feisty." [19:33:45] superm401: Because it makes it "easy" to delete. :-( [19:33:45] quick delay check [19:33:55] he just changed to talk about communities [19:34:01] 11:33:50 [19:34:09] on stream, please mark when he changes. [19:34:10] Talking about communities now. [19:34:14] https://blog.wikimedia.org/2014/12/12/tapping-into-the-knowledge-of-indigenous-communities/ is a great example of how oral citations can be verifiable. [19:34:19] Hopefully what happens then is that the admin who reviews the tag realizes it's inappropriate and declines. :/ [19:34:22] 10-15s [19:34:29] There's a lot of room to facilitate that. Oral Citation wikipedia app that records things to Commons? [19:34:35] cajoel, FYI: I had that typed and ready to hit enter when he switched slides [19:35:27] halfak, thanks for Heather Ford’s link! [19:36:17] halfak: ok, closet to 20s [19:36:19] MissGayle, np. She works with Mark Graham and she's pretty awesome. It would be nice to pull her to the WMF for a talk. :) [19:36:29] that’s a lovely idea - I’d like to do that [19:37:32] superm401 is also reminding me that I need to read our blog more often too - bookmarking to read later [19:37:46] I need to read it more too. Kind of got lucky with that one a while ago. [19:38:10] I think there have been earlier experiments with oral citation as well [19:38:16] a few years ago [19:38:26] User talk is not necessarily a great way of measuring this. [19:38:36] link! [19:38:37] superm401: Once Flow is everywhere we can have proper numbers. [19:38:41] People may use user talk less when article talk is working well. [19:38:46] I'd love to add it to our scren on 3 [19:38:47] That's true. [19:38:49] But user talk is good for some things, like mentorship. [19:38:54] * James_F nods. [19:39:30] http://haithams.github.io/community_visualization/ [19:39:52] It probably correlates to community health.{{citation needed}} [19:40:21] http://haithams.github.io/community_visualization/ is the link in question if you're following along and don't have it. :-) [19:40:24] Do anyone have a link to that tool for discussion interactions? [19:40:37] (hint: check out dutch wiki) [19:40:41] superm401: Plausible. [19:40:47] is haitham in here? [19:40:57] Philippe: … wow. [19:41:03] * Ainali feels stupid [19:41:04] cajoel: Don't think so. [19:41:41] here :) happy to answer any questions here or off-list at hshammaa@wikimedia.org [19:41:44] That tool is fascinating [19:41:56] Hi, I have some questions about Visual Editor and the citation tool [19:42:11] Oh, he HaithamS, sorry. [19:42:16] * Philippe has been worshipping Haitham for a year or so for this tool. [19:42:24] Profoss: Hey, we're having a meeting right now. #mediawiki-visualeditor is a good place to ask questions. [19:42:31] HaithamS: I'd like to add those interesting diagrams to the 3rd floor [19:42:37] Philippe, wow, that is quite interesting. [19:42:42] cajoel, can we talk seperately? I have an idea for that. [19:42:45] and wanted to ask if you could make a 'show me a random one' link [19:42:45] yep [19:42:49] anyone can contribute [19:42:54] https://haithams.github.io/community_visualization/Commons/network/index.html Fascinating indeed. [19:42:55] but yes, sure [19:42:55] James_F: cheers [19:43:50] Does anyone have a question? I've got halfak's about the size of the prize, but nothing else, I think? [19:43:52] cajoel, what ever Philippe said, but that would be a great idea in general :) [19:44:09] - ever :/ [19:44:36] I'm glad to see more emphasis on local language content (though it's good to have people contributing to the big wikis too). [19:44:45] HaithamS: I think a little bit of code for a 'choose a random' one would help me a lot. [19:45:02] A couple All Hands ago I think there was a discussion of encouraging people to edit in one of the Global South countries, but mostly in English IIRC. [19:45:13] cajoel, let's talk about it, and see what do you have in mind. [19:47:04] James_F: How does VE behave on poor connections? [19:47:06] And we need data centers in the Global South. [19:47:51] guillom: Poorly; lots of VE uses network requests (e.g. look ahead search for links, templates, …). [19:48:25] guillom: It'll soon be about the same amount of data transfer to open the editor, but thereafter it may be more. [19:48:32] Screaming a documentary? :) [19:48:54] guillom: If you just open, write 'fish' and press save, it'll be the same as the wikitext editor, roughly. [19:49:01] Ah, thanks Asaf for Achal's name, that's what I was thinking of earlier: https://meta.wikimedia.org/wiki/Grants:PEG/Achal_Prabhala/Oral_citations [19:49:11] And https://meta.wikimedia.org/wiki/Research:Oral_Citations [19:49:31] * James_F nods. [19:49:38] James_F: thanks [19:51:39] Any more questions, please do ask. [19:51:52] Give or take half a billion. So we might only be 2 billion short. [19:52:00] HaithamS: https://meta.wikimedia.org/wiki/WikimediaKiosk [19:52:04] Give or take /our entire reader base/! [19:52:15] halfak: Indeed! [19:52:24] James_F, up to you if this is good to ask now, but: [19:52:36] How do we figure out where to put a new data center? [19:53:14] superm401: Will try to ask if there's time. [19:53:21] Figure out where latency is the worst? [19:53:24] superm401: http://blog.wikimedia.org/2014/07/09/how-ripe-atlas-helped-wikipedia-users/ [19:53:47] guillom: I'd guess we'd also weight it on impact. [19:53:53] Probably [19:54:14] slashdot-ish meta moderation, rating the moderators? [19:54:15] The summit of Himalaya might have high latency but low impact :) [19:54:19] guillom: Addressing latency in Southern India is probably more valuable than Eastern Siberia (or whatever). [19:54:27] guillom: Or that. :-) [19:54:29] and also on ability to deploy --- not every country in the globalsouth has good datacenters/power/cooling/staffing [19:54:49] PageCuration is not only about deleting. It's also about constructive feedback. [19:54:59] +governmental support +isps [19:55:03] ebernhardson: i agree, meta-patrolling seems like a good fit [19:55:12] Revscoring! [19:55:16] :) [19:55:38] halfak: Oy. [19:56:09] I hope to use revscoring to upset the status of quality control/socialization. [19:56:16] Without ruining how awesome it is. [19:56:43] e.g. getting an echo notification when a page on your watchlist was likely damaged. [19:57:04] Or routing good-faith newcomers to WikiProjects (we're already doing that a little bit) [19:57:53] I don't think there are any bots that automatically nominate articles for deletion. [19:58:20] superm401, most automation that I'm concerned about isn't fully-automated [19:58:27] It's human-in-the-loop [19:58:31] halfak: I may have shared this old discussion with you before, but just in case: https://www.mediawiki.org/wiki/Quality_assessment_tools_for_Wikipedia_readers [19:58:38] i wonder if you could tie meta moderation into games, perhaps your farm grows corn twice as fast for 8 hours if you meta-moderate 10 wiki moderation actions :P [19:58:39] You don't need to be an admin to mark something for deletion, but as Moonriddengirl girl said, there is always a check before it actually gets deleted (maybe a long check in the case of AFD). [19:59:12] I think there was a bot on Wikidata that used to nominate things for deletion [19:59:30] guillom, looks like wikitrust. Is that right? [20:00:18] Hmm, didn't know that [20:00:34] halfak: no, it was a discussion about having a tool that aggregates indicators (like wikitrust, or wikiproject assessment, or revscoring) so that readers can choose the metrics they want to evaluate quality [20:00:36] What Jimmy suggests is something that I've been pitching for at least two years. [20:00:57] I don't think we ask Russian Wikipedia admins whether we're allowed to run fundraising banners. [20:00:59] guillom, you think readers would actually use such metrics? [20:01:02] aharoni: We sort-of do this for Wikipedia Zero. [20:01:08] Edits are just as important a kind of contribution, if not more so. [20:01:20] * Ainali is amused by the understatement of the day [20:01:32] Ainali, content is different? [20:01:55] halfak: I think that basically the premise was that there isn't one master tool that can give a True evaluation of quality, so an array of them might be more useful [20:01:59] It seems there are some places where the encyclopedia model is not the best model for representing knowledge. Are we exploring other models? [20:02:08] superm401: "It might be a little tricky" [20:02:09] IT was 5 years ago though :) [20:02:11] it [20:02:33] jsahleen, we've explored a lot: Wikibooks, Wikinews, Wikiversity, Wikipedia, Wiktionary, Wikimedia Commons, Wikispecies, Wikidata. [20:02:48] guillom, indeed, but are indicators of quality necessary? We seem to be gaining ground without the indicators. [20:02:58] Some have been more successful than others, partly because of the community, significantly also because of how much support they get from WMF. [20:03:05] I suppose the bits about linking to talk threads is independently cool though. [20:03:09] halfak: I don't know [20:03:11] The technology to do the functionality of target geolocated language suggestions is very easy. Pretty much done, actually. It may have some performance issues, but we are supposed to not worry about performance, aren't we? :) [20:03:13] [ https://en.wikipedia.org/wiki/Wikipedia:Don%27t_worry_about_performance ] [20:03:30] ebernhardson: you start there and end up with in-app payments: "Buy 100 WikiGems™and get 20 un-reversible edits!" [20:03:32] Reasonator should be the stub article for every small wiki!J [20:03:39] aharoni: No no no. /Editors/ are not meant to worry. We are. :-) [20:03:44] Where would we translate it from? [20:03:44] marcoil: lol :) [20:04:23] And is that necessary given people can already Google Translate it for reading purposes? Chrome even has this built-in. [20:04:47] Google Translate *sucks* for most languages. [20:04:58] aharoni: +1 [20:05:05] (Sorry about the candor.) [20:05:12] aharoni, yeah, but the proposal was to seed the smaller Wikipedias with essentially Google Translate articles. [20:05:27] I'm arguing that's not really better (and probably worse) than machine-translating it at read-time. [20:05:30] superm401: Yes, that proposal would be pointless. [20:06:02] superm401: no, in any small language, use Reasonator for the other 3million articles [20:06:02] Bye everyone! [20:06:12] spagewmf: Why not just all languages? [20:06:51] aw, amir got cut off :) [20:06:52] "hello? can anyone hear me?" then. suddenly. nothing. [20:06:53] superm401: some people started doing such "seeding" in Tamil in 2009/2010, and they hated it so much that they added a new permission designed to prevent it. [20:07:06] thanks! [20:07:11] aharoni, yeah, I'm not in favor of it. [20:07:12] aharoni: that makes sense though :) [20:07:14] I'm here if anybody has questions about translation!!! [20:07:22] the kinds of people who are active on there are going to hate the translated content [20:07:22] ContentTranslation is a better approach. [20:07:30] James_F: maybe so. That way you'd find BLP articles "on" enwiki for people that enwiki deemed non-notable [20:07:32] but that doesn't mean the readers will hate having the content available [20:07:50] If the machine translation is "pretty good" for a given language pair, then people can quickly finish it up with manual corrections and have a high thoroughput. [20:07:52] maybe the problem is creating good UX that differentiates the "bot content" from the "human supervised version" [20:08:11] If it's not good enough (so lots of translation corrections are needed), then doing bot translations then saving is obviously not a good idea. [20:08:13] so like, if you're looking at a purely automatically translated page, it would be light blue in the background or something with a little note [20:08:14] milimetric: there's no point in saving 100% machine-translated pages. google translate can generate them in a second. [20:08:30] right - they don't have to be saved, they can be "live translated" [20:08:38] until someone touches it and makes it "human" [20:08:43] in ContentTranslation we show a big orange warning to users who try to save an article with more than 90% machine-translated text. [20:09:01] milimetric, if people don't see the edit button (which a lot of people don't), they're probably not going to notice light blue accents. [20:09:03] yeah, i like that, i've been meaning to use it more to contribute to rowiki [20:09:09] :) [20:09:15] we gotta get everyone glasses [20:09:21] WikiSight Zero [20:09:22] If it's on Wikipedia, it's On Wikipedia. [20:10:50] Also, there are other ways to seed articles besides translating from a dominant language. [20:11:02] E.g. English WIkipedia used to auto-create articles on census place. [20:11:11] Some smaller wiki (I forget which) was doing the same for species. [20:11:46] So if some wiki decides they want to increase the number of articles quickly, that is probably a better approach than straight machine translation without correction. [20:27:47] superm401: Many wikipedias did this, and not much else.