[05:15:45] I so cannot wait for work tomorrow [05:25:17] YuviPanda, yt? [08:47:20] Ironholds: yup [15:28:43] <^demon|away> Ironholds: Special:Blankpage is a Domas invention :) [15:28:58] <^demon|away> He used to use Special:Version as a benchmark, but then we added too many features to it so it was slow. [15:29:10] <^demon|away> So he wrote a page that does nothing on purpose :) [16:39:57] Good morning leila! [16:40:01] Happy new year! [16:40:14] good morning, halfak! happy new year to you as well. Hope you have a great 2015. :-) [16:40:18] hey Ironholds. [16:56:26] o/ J-Mo [16:56:29] happy 2015 [16:56:54] to you as well, halfak! Enjoy your vacations/holidays? [16:57:28] Indeed. Had a good one. Got bored and I'm ready to be back :) [16:58:01] yeah, having that much time off was kind of a shock to the system [17:11:25] :) [17:11:35] * YuviPanda prevented that by not taking any time off other than the weekends [17:26:49] Ironholds, we're not meeting at 12:30 your time it seems [17:27:07] there is no business given that everyone has just got back. [18:00:02] leila, have you talked to Ironholds this morning? [18:00:21] no, halfak, and I was wondering where he is. [18:00:30] kk I'll drop him a call. [18:00:39] k, hope all is well [18:14:08] argh [18:14:43] that argh had a distincg Ironholds quality to it [18:15:23] my hangouts keep crashing [18:15:32] even after restart [18:15:34] I hate you, google [18:15:41] halfak, shall we IRC it? Sorry about this :( [18:15:49] No worries Guest19441 [18:15:51] :P [18:26:24] DarTar-lunch, ping [18:27:06] howdy [18:27:16] wat, am I really DarTar-lunch? [18:34:15] DarTar, so, Shilad wants to know if you'd like to be thanked in the paper, or if mentioning you were involved in dataset prep is going to lead to you being...I dunno, fired by the Association of Sociophysicists [18:36:02] Ironholds: my membership is expiring anyway so by all means you guys should feel free to thank me [18:36:31] cool! [18:36:39] which reminds me we have a ton of conversations to resume with the Oxonienses [18:44:21] ^ not sure if they’re just making up the names or not [18:47:25] YuviPanda, all names are made-up. [18:47:36] * halfak make a mindsplosion face [18:47:41] woaaaah [18:56:07] YuviPanda, people from Oxford university [19:06:34] halfak, are people from Minneapolis Minneapolitans? [19:06:48] related: I would like ice cream please. [19:08:41] Visit Italy then ;) [19:09:00] I believe that Minneapolitan is the most appropriate term that we would never use to describe ourselves ;) [19:11:34] halfak, I'm gonna make it a thing [19:11:59] Bostonian. [19:12:23] lol. A lot of stuff in boston has "bostonian" in the name. [19:13:01] On the contrary, only a blog and a bunch of dictionaries reference minneapolitan [19:22:59] http://www.theguardian.com/australia-news/2015/jan/05/south-australia-26-homes-and-41-sheds-wiped-out-in-raging-bushfires australians take their sheds really seriously [19:29:37] YuviPanda, anything ever come of this? http://www.gossamer-threads.com/lists/wiki/wikitech/348122 [19:30:19] halfak: nope. mostly hit the ‘why not just do it on the wiki page’, got bored, and let the domains expire [19:30:36] Gotcha. Seems like a good idea. [19:30:47] Also, the domains can point to a wiki page. [19:43:36] halfak: true. [19:43:50] halfak: however, the initial idea was much simpler and ‘snazzier’ :) [19:43:53] than a wiki page [19:43:59] and I was interested in it because of the snazziness [19:44:13] and then I guess some other shiny object popped into my vision... [19:44:57] Gotcha. Snazzy and wiki page is hard. I don't see a good reason to make something like this not a wiki page, but we all seem to agree that mediawiki's border stuff is dumb looking. [19:45:41] Maybe we should have stand-alone services that will read a wiki-page and present it outside of the mediawiki stuffs. [19:46:45] halfak: indeed. I was going to do something like that for toollabs help [19:47:02] That would be great. [19:47:12] I find tool labs help to be generally useless. [20:12:11] halfak, YuviPanda: that was pretty much the plan for the data & developer hub documentation: let content creation happen on MW, convert/publish content statically to a different site not running on MW [20:12:27] and hi btw :) [20:12:30] \o/ [20:12:36] I don't hate that idea anymore :) [20:12:52] DarTar: did that go anywhere? [20:12:52] My only concern was letting people fix and contribute wiki style [20:13:03] halfak: oh yeah, source wiki, destination, something nicer [20:13:12] Wiki is nice [20:13:14] Mediawiki is not [20:13:17] I don’t think it’s going anywhere at the moment [20:13:34] it was under Platform but got deprioritized [20:14:04] I still think it could be a fundamental piece in a “future of labs“ world [20:14:11] ^ +1 [21:07:28] Protip: Upgrade your frozen pizzas with little slices of bacon. [21:07:32] It is magnificent. [21:18:24] my mother just asked me how San Francisco is [21:18:45] * Ironholds shakes head [21:20:20] heh. [21:20:23] * halfak googles weather in SF [21:20:42] It's 57F and mostly cloudy [21:21:13] Here, it's -2F and overcast. [21:21:49] That's -19C for you non-mercans [21:22:43] woah [21:22:51] how do you even go outside at -19? [21:23:06] see this is why Minnesotans are so healthy. Smoking involves taking your balaclava off. [21:23:34] Na, you can smoke through it. I've watched people light their cigarette *inside* of their jacket. [21:23:49] dang [21:23:58] still on my list of places to be, but y'all are making it tough [21:24:04] heh. Last year was a really cold one, so I got to enjoy some afternoon trips to the grocery store at -40C [21:24:13] THE FUCK [21:24:16] * Ironholds whimpers [21:24:27] I'll move to Cary. I qualify for residency there now. [21:24:39] they don't really get snow, just rain and heat that punches you hard in the mouth [21:25:13] It's all about being a technologist. [21:25:38] Get yourself a nice jacket and become familiar with how body temperature changes with activity. [21:25:49] I have a hellaciously nice jacket [21:25:53] It's sort of like living under the sea. [21:25:54] it's rated down to -60F! [21:26:05] "How do you breath underwater." Well, we have equipment for that. [21:26:16] Is it down feathers? [21:26:25] * Ironholds checks [21:27:45] halfak, "600 fill power insulation" I don't know what this means [21:28:11] https://en.wikipedia.org/wiki/Fill_power [21:28:13] It's down [21:28:47] 600 is pretty high. [21:29:03] cool! [21:29:09] I prepared properly for the boston winter [21:29:29] them and a set of http://www.sorel.com/mens-conquest-boot-NM1049.html?cgid=men&dwvar_NM1049_variationColor=003#start=11 [21:29:45] Your boots are far more hardcore than mine. [21:29:54] I've been going without boots for years! [21:29:58] Just got a set this year. [21:30:34] http://www.amazon.com/gp/product/B000VX6Y2Y/ref=oh_aui_search_detailpage?ie=UTF8&psc=1 [21:31:03] I don't know why I waited so long. Boots are awesome. [21:31:21] boots are GREAT [21:31:29] neat! [21:35:28] leila, do you know anything about concentration measures? [21:48:02] * Ironholds headscratches [21:48:11] halfak, got some time for a rubber ducking session? [21:48:18] Sure. [21:48:34] Great timing. I'm between tasks ATM [21:48:44] So, I'm thinking about concentration measures. [21:49:35] specifically: what if I have the code not only tell me what the concentration measure was, and whether it was > threshold [21:49:58] but also return the UAs that led to the observed value being > threshold, where applicable? [21:50:20] Is concentration measured at a page level or page/UA level? [21:50:35] page level, but by UA, so it's the Herfindal measure applied to table(user_agents) [21:50:56] BTW, did you look at https://en.wikipedia.org/wiki/Gini_coefficient? [21:51:05] I did; it doesn't actually work here [21:51:17] Sub-linear? [21:51:28] I..don't know what that means [21:51:33] I have been thinking of looking at Thiel, too [21:52:38] Let me put it this way, a gini coef of a logistic function is the same value as a simple diagonal line. [21:52:51] gotcha [21:53:03] * Ironholds grabs an example, actually [21:53:12] Because it measures area under the diagonal [21:53:50] yeah [21:54:26] http://blog.ironholds.org/content/images/2015/Jan/agent_lorenz.png is from some of my initial experiments and may interest you, btw [21:54:51] Ahh... What is this ROC-like-thingie? [21:54:52] * Ironholds generates gini versus herfindal values [21:55:16] Oh wait. I see. [21:55:27] ordinary and generalized Lorenz curves [21:55:47] Ahh... so, I think that gini would work great like this. [21:56:05] let's see what my test says; I might just've cocked it up for the first time [21:56:18] I was using the Gini initially but found some really weird values and didn't know what ot make of them [21:56:46] * Ironholds watches it generate. God bless data.table. [21:57:31] Gotcha. I'll work up a quick example. [22:00:00] * halfak installs the R packages [22:03:49] I seem to be failing too. [22:08:29] WTF gini. Arg! [22:08:46] what's it doing? [22:08:59] Not reflecting the inequality it is supposed to be measuring! [22:10:33] yup! [22:10:44] now use Herfindal and see what you find [22:10:54] conc(x) [22:12:13] https://gist.github.com/halfak/9204ffe666e85bd7ad63 [22:12:17] Got it working. [22:13:06] Ironholds, ^ [22:14:03] huh [22:14:14] my PV test is still running; will see what happens when it's done :) [22:14:31] (running over 31m pageviews was maybe not the best idea) [22:27:10] Ironholds: what partitions are you running on? [22:27:15] , or , how many? [22:27:39] ottomata, I'm not, this is data I retrieved earlier [22:27:44] it's currently running on one of my desktop's cores [22:27:46] oh k [22:39:49] Ironholds, I see that halfak is saving the day. just in case it helps: I know concentration measures in economics theory in very high level. I can help if I read more, but off the top of my head I don't have immediate knowledge to share. [22:40:19] leila, okie-dokes! I may steal some of your time when I hit SF in a few weeks [22:40:42] sounds good. ;-) [22:40:44] WOo! I forgot that we're all going to be in the same spot soon. :) [22:58:33] halfak, you're right, geni does work. That's weird. [22:58:39] I wonder if I was just seeing pages with very few requests.. [23:01:31] \o/ I'm not sure which statistic is *better* here. [23:02:55] hmn [23:02:57] * Ironholds thinks [23:03:00] halfak, the test I'd use is twofold [23:03:12] first, hand-code against a big sample to see if either one is noticeably more accurate [23:03:23] if that doesn't resolve it, see which one draws tighter clusters of values. [23:03:46] IOW, see which one we can use to determine a better threshold, above which/below which the data is assumed to be automata-driven [23:03:53] and if they're both equal at that too, fuckit, use whatever ;p [23:04:04] does that make sense as an approach, to you? [23:13:53] halfak, you don't happen to remember the name of that minnesota public information youtube thing you linked us to, do you? [23:17:14] Ironholds, this one: http://video.tpt.org/video/2365042610/ [23:17:21] yay! [23:17:38] started talking to a Minnesotan and we're discussing cultural differences and similarities between minnesotans and massholes [23:17:59] Ironholds, your approach sounds reasonable *if* we can believe humans are good at detecting robots. [23:18:09] lol @ massholes [23:18:21] hmn [23:18:29] I think we can present them with a lot of summary statistics that would help. [23:18:38] for example, session lengths, average intertimes [23:18:43] and masshole is a thing [23:22:31] related: http://www.wikihow.com/Talk-Minnesotan [23:26:32] All those images of talking people [23:33:39] heh