[00:01:12] Hi, yes, although it's getting quite late in Sweden =P. I will be back tomorrow! [00:05:30] Icebear_ OK. I'm UTC -5 and an early bird. [00:05:33] Bah too late! [02:54:52] halfak, if you're still around [02:54:58] I have had a Thought [04:35:25] Thought with a capital T. [05:58:14] Ironholds: think of huggle for android [06:00:55] Guerillero: and not huggle for google glass??? ;P [06:01:49] I try to stay in the relm of sanity [13:34:02] o/ Ironholds [14:38:56] morning nerds! [14:39:09] halfak, you know, I've now completely forgotten what the thought was. It'll come back to me, I'm sure. [14:39:45] Boo. OK. [14:39:49] * halfak regrets missing it. [14:40:22] Say, I suggested that we hack together on a specific problem at a meeting a while back -- but I forgot which thing that was. [14:40:24] Do you remember? [14:40:51] oh! [14:40:52] yes! [14:41:13] it was investigating the difference HHVM made to mobile edits in respect to whether there was a significant variation in connection type between the groups. [14:41:23] Does the change help uniformly, for people trying to edit on cellular connections... [14:41:31] Oh yeah! That would be awesome. [14:41:41] What's the best way to turn an IP into a probable connection speed? [14:42:15] https://github.com/Ironholds/WMUtils/blob/master/inst/geo_netspeed.py boom [14:42:23] (obviously you won't need argparse etc) [14:42:33] there's also an R wrapper depending on when in the process you want to do it [14:43:16] Ironholds, unrelated, but I thought you would be interested in this: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2511947 [14:43:29] I plan to pick it up later. [14:43:54] neat! [14:44:06] Those /usr/share/GeoIP dirs -- are they on stat2? [14:44:19] and also stat3 [14:44:24] * Ironholds jazz hands [14:44:27] Woot. [14:44:40] assuming otto puppetised them like I asked him to, anyway [14:44:42] They're deff on stat2 [14:45:03] ooh, and I may have worked out how to solve the thing blocking me on circadian rhythms! [15:16:06] dammit, where's J-Mo, I have a UDub question [16:03:05] halfak, do you remember who we were meant to forward recommended candidates to? [16:03:11] or, people the recruiters should poke, rather? [16:03:34] Ironholds, good Q. I have just been sending people to the job listing. [16:03:54] yeah, this is one of the "you have no idea who I am but I fricking love that paper" crowd [16:03:56] I suppose we should be letting someone know that they are a known applicant. [16:04:09] Ahh. probably just the job listing then. [16:05:15] But it's still a good question how to handle referrals. [16:05:56] I'll ask Emily [16:41:03] I like today! [16:52:48] * quiddity guesses no meetings? [16:53:53] quiddity: only 5h of meetings, Id' think [18:03:11] Ironholds, are you joining? [18:03:12] :-) [18:05:00] leila, the meeting with no hangout call? [18:05:03] oh! it has a hangout now! [18:05:27] :-) [18:16:46] halfak, are you there? [18:19:18] leila: have I missed a meeting that you wanted with me again? [18:19:20] i am bad with calednars [18:20:02] noo, you're good ottomata. I haven't requested any new ones. [18:20:05] thanks for checking. [20:02:37] huh [20:02:47] hey halfak, you heard of a guy at UMN called Robert Cooley? [20:03:16] Or Jaideep Srivastava? [20:17:47] Ironholds, I have been introduced to Jaideep Srivastava, but I wouldn't say I know him. [20:17:48] I [20:17:58] 'm trying to figure out if I have met Robert Cooley [20:20:59] gotcha. Nothing major! I've almost got a present for ya, though. [20:22:47] oooh [20:23:00] halfak, https://meta.wikimedia.org/wiki/Research:Session_analysis#HCI_prior_art [20:23:02] * Ironholds jazz hands [20:23:18] wait, what? [20:23:19] "prior art"... [20:23:21] there's a paragraph missing [20:23:41] hangon, don't read [20:23:43] I screwed up an edit. [20:24:17] halfak, okay now read ;p [20:24:29] and, sorry: lawyer. https://en.wikipedia.org/wiki/Prior_art [20:26:14] Will scope it out tonight. [20:32:07] cool! [20:32:15] in exchange I'd ask for some stuff added to the unique clients page ;p [20:42:51] halfak: I could use some time to get more mobile stuff done today, do we have urgent stuff to talk about or can we push the 1:1 to another time? [20:43:14] I'd like to talk. We can try to keep it brief. [20:44:02] alright, deal [20:47:53] halfak, is it okay to contact users (by email) that have been editing for some time to do surveys? [20:49:54] leila, that's a complex question. sec. [20:50:11] yes! [21:01:52] OK. back. [21:02:12] Leila, re. sending messages randomly to Wikipedia users. I need to start with history. [21:02:32] Way back in 2006, I watched a bunch of researchers attempt to recruit on Wikipedia and fail. [21:02:41] halfak: there in sec [21:02:51] No worries Dartar [21:03:10] leila, I worked to develop policies and guidelines for researchers in Wikipedia. We actually passed them by RFC, but that brought substantial push-back [21:03:28] Woops. Going to have to continue later. [21:03:32] Need to 1:1 with DarTar [21:07:01] leila, DarTar's battery died. More history! [21:07:41] OK. So, there was a policy on enwiki and then it died -- mostly due to concerns about "policy creep" and the cost in disruption that research activities would bring to Wikipedia. [21:08:13] So, a couple years later, DarTar set up the "Research Committee" -- a group of people to improve wiki research. [21:08:23] ... DarTar is back [21:17:00] thanks, halfak. more after your meeting then. :-) [21:37:41] leila, back for a few minutes. OK So, with the introduction of RComm, it turns out that there were other people interested in the problem of research recruitment. [21:38:44] So I worked with them to draft a minimal process that only specified that the researcher had to (1) define their project, (2) hold a discussion with Rcomm members and wikipedians and (3) obtain consensus that the recruitment would be non-disruptive. [21:39:04] This process worked for about 3 years and lots of projects came through. [21:39:33] Recently, there were some concerns raised about whether this process is really necessary any more (I can get links to this is you are interested). [21:39:59] However, there seems to be clear consensus that the Right(TM) way to recruit subjects is to document your study and have a conversation first. [21:41:21] If it were me, I'd write up a study description on Meta including the stuff an IRB would want. Then I would invite prominent Wikipedians who are concerned with Research Ethics (e.g. Bluerasberry) and see how the discussion goes. [21:41:48] Given that you might be doing this a lot, I'd consider writing that fact into the proposal so that new recruitments don't need a whole new proposal and discussion. [21:42:35] I suspect that this is something we can lean on Rachel's team for help with. [21:43:13] I'm also very happy to help get things together and work toward consensus in the conversation. [21:44:51] Now, this is all very heavyweight stuff and I think there is an argument for why the design team wouldn't need to do this *every time* that they have a new need to recruit research subject. I think that should start with an on-wiki conversation. I'd support a standing approval of Design's recruitment assuming that all projects are well documented at the time of recruitment. [21:45:51] halfak, thanks! this is very helpful [21:46:09] there is a survey which is probably going to go out on Friday [21:46:19] it will be sent on mailing lists [21:46:46] Oh! I was imagining emails to individual users. [21:46:55] I don't see why the WMF emailing a survey link would be a problem. [21:47:00] I was suggesting that we start to have an opt-in pool of participants who choose to participate in future surveys [21:47:04] *to a mailing list [21:47:12] +1 [21:47:26] that's what I first had in mind, individual emails. but I thought that's too disruptive if every team decides to do that [21:47:32] (already I know a team who does that) [21:51:55] ack! Well... I guess they aren't really being disruptive if they have been getting away with it so far. [21:52:39] yeah, but I don't want to support it. I think it can cause trouble. [21:53:49] Indeed. I think it's good to be proactive here. Ethically and practically. [21:57:45] alright, checking out for the evening. [21:58:09] halfak, good luck with the connectivity tests. Warning that if you find anything interesting, Scott and I are shanghaiing you for our full paper on mobile connectivity issues and editing patterns. [21:58:19] I know how much you hate publishing ;) [21:59:41] also, before I go: comments on the cache test. I bumped the thread this morning so nobody has any excuses for not finding it! :P People who /aren't/ halfak or I, looking at you. [21:59:54] anyway. Dinner date and ramen. [22:06:48] Ironholds, woot and *shakes fist* respectively. [22:11:40] halfak, i guess its actually good that we both made that pmcid data set - verifiablity [22:11:52] +1. I didn't know about pmc's [22:12:04] Are those also "pmids" or are they some other identifier? [22:12:18] there is a subtle difference that i don't quite understand [22:12:27] i think its because one system got folded into another [22:14:37] Gotcha! Cool. I'll go set up a new run. [22:14:43] How many pmids did you find?