IRC office hours/Office hours 2016-05-26

[17:58:11] <varnent> James_F: are you auditioning for the Phantom of the WMF?
[17:59:51] <James_F> varnent: Totally.
[18:00:09] <James_F> "Ding ding ding! Please remember to keep your hands and feet inside the vehicle at all times. If you're not in a vehicle, run!"
[18:00:19] <James_F> And we're live.
[18:00:29] * varnent for no apparent reason has been listening to London Phantom of the Opera soundtrack lately
[18:01:13] <foks> Yay for Comms!
[18:01:19] <James_F> Yay for foks.
[18:01:25] <varnent> yay for foks
[18:02:05] <foks> :D
[18:02:13] <James_F> Audio feedback?
[18:02:36] <Pine> James_F: it's suboptimal. Sounds like it has high compression.
[18:02:47] <milimetric> sounds really good to me on youtubew
[18:02:52] <James_F> brendan_campbell: There's audio feedback here?
[18:02:53] <matt_flaschen> Thanks to all these people for all their work over the years. :)
[18:03:09] <matt_flaschen> Also fine for me on YouTube.
[18:03:11] <foks> Sounds good to me on YT.
[18:03:12] <dedalus_> I hear María quite well
[18:03:28] <James_F> In the office we were hearing the video playback.
[18:03:32] <Pine> I'm getting audio artifacts.
[18:03:34] <James_F> Fixed now, thanks brendan_campbell.
[18:03:43] <brendan_campbell> sorry about that
[18:03:44] <i_jethrobot> Hey folks.
[18:05:36] <dedalus_> The video is out of focus
[18:05:50] <James_F> brendan_campbell: ^^
[18:05:54] <brendan_campbell> try refreshing youtube and selecting 720p
[18:06:10] <brendan_campbell> dedalus_: ^^
[18:06:14] <i_jethrobot> I readjusted my video quality just now, and it looks good.
[18:06:22] <dedalus_> brendan_campbell: thanks, that worker
[18:06:26] <dedalus_> r=d
[18:06:27] <brendan_campbell> awesome
[18:06:39] <Pine> I'm still having audio artifacts, but maybe that's just me.
[18:07:18] <i_jethrobot> Pine - Not just you.
[18:07:42] <Pine> It seems to be a challenge to get the audio to be good quality from month to month.
[18:08:30] <Pine> My video might be frozen. Maria's audio is way ahead of the video...
[18:08:38] <brendan_campbell> can you please describe the artifacts you are hearing?
[18:09:00] <foks> Yay Comms!
[18:09:01] <brendan_campbell> i'm also listening to the youtube stream right now
[18:09:30] <Amir1> \o/
[18:09:37] <Pine> I did a refresh of the page. Video is now synced. Audio sounds like it's got a warble and occasional quick bits of static.
[18:09:38] <i_jethrobot> brendan_campbell: Sounds like bits of static that pop up here and there. It's gotten a little better for me since Pine's message about it.
[18:09:54] <i_jethrobot> Oh yeah, the warble too, actually...
[18:11:28] <Maggie_Dennis> Weird. Sounds good for me.
[18:11:42] <foks> Yeah not bad for me either.
[18:12:14] <Ainali> I have good audio too
[18:12:15] <dedalus_> Jeff rocks
[18:12:16] <Pine> Maybe it depends on the browser.
[18:12:26] <milimetric> hm, instead of serving the pool counter notice page, couldn't we just serve old cached versions of it?
[18:12:37] <milimetric> sorry, I don't mean "just" as if it's easy :)
[18:13:18] <James_F> milimetric: We fixed this bug since then.
[18:13:26] <James_F> milimetric: (Yes.)
[18:13:41] <James_F> Also "we" but I didn't do any work. ;-)
[18:14:06] <milimetric>  :)
[18:14:13] <milimetric> nonsense though, you work on literally everything
[18:14:21] <foks> This is pretty funny, before this article happened I don't think I'd ever heard a full Prince song.
[18:14:23] <matt_flaschen> James_F, ori, could one of you explain the PoolCounter error and fix. I heard something about the cause, but I don't know the details.
[18:14:38] <Maggie_Dennis> foks, you shouldn't confess that kind of thing in public.
[18:14:43] <foks> :P
[18:15:11] <foks> A bit before my time. ;)
[18:15:28] <foks> "Interviews"
[18:15:42] <foks> It's ED!
[18:15:59] <Pine> BMaggie_Dennis: BTW please check your Meta talk page sometime in the next several days. :) I have a question for you there.
[18:16:05] <Pine> Ugh
[18:16:10] <Pine> Maggie_Dennis: ^
[18:16:13] <Maggie_Dennis> Pine, you will or you do already?
[18:16:20] <Pine> Did yesterday.
[18:16:27] <Maggie_Dennis> Ah, okay. Will look when I can. :)
[18:18:00] <ori> matt_flaschen: PoolCounter was being rendered largely ineffective due to a bug -- ParserCache entries were being rejected even when ParserCache::get() was being passed $useOutdated = true
[18:18:30] <ori> that was due to https://gerrit.wikimedia.org/r/#/c/122847/ . The task is https://phabricator.wikimedia.org/T133659
[18:18:52] <James_F> Thanks, ori.
[18:19:07] <RoanKattouw> I think that ping was for milimetric
[18:19:24] <milimetric> which what?
[18:19:53] <milimetric> (matt asked about the bug, but I saw the explanation too, interesting)
[18:20:18] <foks> nice job, Ed17
[18:20:19] <ori> well, not rendered ineffective -- its queue should not have overfilled, even given the spike
[18:20:20] <foks>  :)
[18:20:29] <matt_flaschen> Thanks, ori. Makes sense.
[18:20:52] <Ed17> foks: thank you :-) Hope we gave you enough credit, it really was your idea that started NoW
[18:21:10] <Maggie_Dennis> Pine, responded. :)
[18:21:40] <foks> Ed17, I'm furious.
[18:21:44] <foks> But not really. :)
[18:22:06] * Ed17 high-fives foks
[18:22:24] <foks> If people want to read that article on the Wikimedia Blog, the link is: https://blog.wikimedia.org/2016/04/22/prince-death-wikipedia/
[18:22:53] <Ed17> also you should be happy, I was thinking about ad-libbing a lament about our sadness at losing you to the tyrannical poachers in CE. ;-)
[18:22:55] <Pine> Maggie_Dennis: I'm particularly interested in encouraging staff who ordinarily never mix with the community to do so at that conference. Admin, legal, support folks, etc. I think the experience would give them a broader sense of the mission and their role in it. :)
[18:23:09] <ori> (q for tilman) is it "holding steady" if the population of internet users has grown as much as it has?
[18:23:18] <James_F> ori: In my list, thanks.
[18:23:21] <foks> Ed17, :D
[18:23:59] <milimetric> ori: good point but for that we probably have to look at unique devices, pageviews seem like such a fuzzy metric to me
[18:24:53] <ori> sure, but the question is relevant to all traffic indicators. http://www.statista.com/statistics/273018/number-of-internet-users-worldwide/ says "the number of internet users worldwide was 3.17 billion, up from 2.94 billion in the previous year."
[18:24:54] <milimetric> (and we don't really have good unique data to compare to three years ago)
[18:24:59] <i_jethrobot> I'd be interested in seeing the standard deviations for those montly pageviews/device...
[18:25:05] <milimetric> agreed, question is def. relevant
[18:25:13] <i_jethrobot> or standard error
[18:25:38] <WereSplchqrs> @ori if the growth of the internet is skewed towards China then we should expect to lose market share simply as a consequence of not having the biggest online encyclopaedia in that language
[18:25:52] <James_F> milimetric: "Page views per Internet-connected human" is still reasonable, I think.
[18:26:11] <milimetric> I donno, James_F, it
[18:26:21] <milimetric> *it's so fragile to how people use the site
[18:26:35] <Maggie_Dennis> Yay, Vahid! Welcome. :)
[18:26:37] <milimetric> like, someone sitting there reading in depth for two hours the same article vs. someone clicking through three pages
[18:26:39] <James_F> Sure, most data is.
[18:26:49] <i_jethrobot> Welcome Vahid!  : )
[18:27:03] <James_F> C.f. how page views are down even as uniques are (?~) up, because of the desktop -> mobile switch.
[18:27:04] <milimetric> well, unique devices per internet-connected devices might be a better measure of reach
[18:27:30] <James_F> I think that given the known double-counting of uniques (and the fact that we don't have any historic data) it might be good to have both.
[18:27:41] <milimetric> cool - those kids in red shirts look like pomegranate seeds :D
[18:28:05] <brendan_campbell> (dog barks, chuckles ensue)
[18:28:20] <Pine> Is Ollie attending the meeting?
[18:28:37] <WereSplchqrs> It would be good to get figures on number of devices that people use Wikipedia on. There will be some people who in the courese of a month come here by home PC, smartphone and work PC.
[18:29:27] <matt_flaschen> James_F, question: Do the 150K articles created include articles that were deleted?
[18:29:35] <James_F> matt_flaschen: Will aks.
[18:29:37] <James_F> Err. Ask.
[18:29:56] <matt_flaschen> Thanks
[18:30:16] <James_F> WereSplchqrs: We have them. Did you miss that bit of the presentation?
[18:30:40] <milimetric> WereSplchqrs: we've got some data since the beginning of 2016: https://analytics.wikimedia.org/dashboards/readers/#projects=enwiki/metrics=UniqueDevices
[18:30:53] <James_F> Aha, yeah, thanks milimetric, was trying to find the link.
[18:30:55] <milimetric> (not announced yet)
[18:30:56] <Pine> James_F: please ask, does "partnering" mean providing WMF engineering staff time, does it mean providing a grant, or a mix?
[18:31:07] <James_F> Pine: Kk.
[18:31:15] <ragesoss> https://outreachdashboard.wmflabs.org
[18:31:17] <matt_flaschen> Pine, what about "providing WikiEd engineering time"? They wrote the dashboard to begin with.
[18:31:46] <James_F> Yeah, I imagine it's WEF not WMF time.
[18:31:49] <James_F> But I can ask.
[18:31:51] <Pine> matt_flaschen: Tighe said that WMF is "partnering" with WEF about the education dashboard. I'd like more detail about what that means.
[18:32:14] <HaeB> ori: i think i may have said "holding steady"more than once ;) which metric/slide was this question about?
[18:32:17] <ragesoss> In February, several WMF engineers worked on the dashboard.
[18:32:49] <matt_flaschen> Good to know, thanks. Sounds like a collaboration, rather than resources flowing only one way.
[18:32:59] <ori> I guess it's a general question about site traffic metrics
[18:33:32] <WereSplchqrs> Thanks James, but unique devices doesn't tell you unique humans, unless Ive missed something,
[18:33:55] <ragesoss> matt_flaschen: yeah, that was the case in February. WikiEd also working in May-June to continue the i18n work, with a contract from WMF.
[18:34:16] <i_jethrobot> James_F - Page views per device sounds like an average, so there should be a standard deviation associated with it
[18:34:23] <ragesoss> matt_flaschen: and I understand that the tentative plan is that Community Tech may work on in more around July.
[18:34:41] <i_jethrobot> James_F - But not a big deal, just mostly curiosity.
[18:35:12] <James_F> i_jethrobot: I think it might just be ∑A/∑B
[18:35:12] <matt_flaschen> James_F, I still have the WikiEd deletion question.
[18:35:16] <James_F> Yeah.
[18:35:29] <i_jethrobot> James_F - Ah, I see. Thanks.
[18:35:36] <James_F> (I think.)
[18:35:48] <awight> kaldari: interesting question!
[18:36:22] <awight> kaldari: I know from Wiki Edu Foundation that they struggle with editor personality conflicts, so staying out of existing articles might actually make sense
[18:36:33] <awight> hard to get integrated...
[18:36:54] <kaldari> all things worth doing are hard :)
[18:37:01] <milimetric> matt_flaschen and others interested: those kinds of questions and more are what we're trying to answer with wikistats 2.0
[18:37:09] <Pine> James_F: question for Maria. Did I hear correctly that there is a community consultation coming about Wikimetrics? If so, can you elaborate on what that will involve?
[18:37:26] <matt_flaschen> Thanks, James_F
[18:37:29] <James_F> Pine: I think it's the on-going one on meta?
[18:37:41] <Pine> Ok, not a new one than. Thanks.
[18:37:42] <ragesoss> that's in a relatively narrow range of topic areas, though, and we don't actively steer people away from those areas, we just put more resources into making sure course projects can succeed.
[18:37:50] <James_F> I.e. https://www.mediawiki.org/wiki/Analytics/Wikistats/DumpReports/Future_per_report
[18:37:51] <awight> kaldari: It would be fun to A/B test different introductory portals in the Wiki Ed context
[18:38:03] <ragesoss> awight: I'm very interested in that.
[18:38:04] <Maggie_Dennis> Cheerful wave at ragesoss. Nice to see you. :)
[18:38:06] <James_F> Not meta, MW.org
[18:38:29] * awight smiles at ragesoss
[18:38:31] <dedalus_> and LiAnna was in Berlin at WMCON
[18:38:40] <K4-713> James_F: Have you ever had people try to *answer* questions from IRC?
[18:38:42] <ragesoss> awight: I'm thinking about putting together a non-Wiki Ed version of the training modules on the dashboard, and doing an A/B test with hostbot.
[18:38:49] <HaeB> i_jethrobot: happy to talk more about this kind of thing (modeling the time series of pageviews, how noisy it is etc. - see also for example https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2015-09-30/Recent_research#Predicting_Wikimedia_pageviews_with_2.25_accuracy ) ...
[18:38:54] <ragesoss> maybe I'll get a chance this summer to put that together.
[18:39:06] <varnent> Reminder that the Interim ED Office Hours will be in about 22 minutes in this channel - more info: https://meta.wikimedia.org/wiki/Wikimedia_Foundation_Executive_Director/May_2016_office_hours
[18:39:15] <dedalus_> brb
[18:39:31] <awight> ragesoss: wow, IMO that would be well worth the WMF contract money...
[18:39:37] <HaeB> ...i just didn't quite understand the question as asked (pageviews are not a static number that we try to estimate by measuring repeately - in that case asking for the standard deviation would make sense for example)
[18:39:47] <James_F> K4-713: Lots of times.
[18:40:11] <i_jethrobot> HaeB - Yeah, I'd be interested in hearing more about that-- what is the noise from month to month?
[18:40:23] <James_F> K4-713: Gayle and Erik M. used to do it quite a bit to reduce the length of Q&A.
[18:40:27] <HaeB> and ori, totally agree that static total PVs are not a reason for complacency at all
[18:40:31] <K4-713> James_F: Do you read off their answers on the mic? Why have I never noticed this before?
[18:40:48] <dedalus_> the meta page links to the youtube recording of 12 may https://meta.wikimedia.org/wiki/Wikimedia_Foundation_Executive_Director/May_2016_office_hours what will be the url for the youtube stream?
[18:40:48] <James_F> K4-713: Mostly I just leave it to IRC.
[18:40:49] <HaeB> ...i did want to correct the misconception though that they have been steadily falling in recent years
[18:40:57] <James_F> HaeB: Thank you.
[18:41:08] <James_F> dedalus_: I don't know if it's set up yet.
[18:42:14] <dedalus_> ok, I'll wait
[18:43:10] <Pine> brendan_campbell: I notice that the audio seemed better later in the presentation.
[18:43:12] <HaeB> i_jethrobot: defined how? so what one could do (and the researchers in that paper did) is to separate the time series - PVs per day, say - into predictable part and an unpredictable part (to express it simplistically)
[18:43:35] <HaeB> the predictable part would include weekly and yearly seassonal compontents
[18:43:36] <awight> HaeB: The number I understood was steadily falling is the change in number of active editors, right? or is that roughly at a steady zero along with pageviews?
[18:44:24] <HaeB> ...and the other term, the unpredictable part, would be what you could call noise
[18:44:47] <HaeB> the researchers in that paper had this noise part down to 2% within a week (they were more interested in short-term prediction)
[18:45:56] <HaeB> awight: they were falling in earlier years but have been holding pretty much steady in recent years https://reportcard.wmflabs.org/graphs/active_editors
[18:46:38] <HaeB> ...and see also https://blog.wikimedia.org/2015/09/25/wikipedia-editor-numbers/
[18:46:39] <awight> Got it, thanks!
[18:46:58] <HaeB> ... neilpquinn is the expert there
[18:48:14] <i_jethrobot> HaeB: Right, that makes sense to me. So regarding the pageviews/device numbers presented for metrics, can you clarify how those figures were calculated?
[18:49:28] <i_jethrobot> James_F suggested it might have been something like ∑A/∑B?
[18:52:04] <James_F> I was just guessing based on how they've been discussed, I don't have any special knowledge