[01:01:14] evening halfak :) [01:01:41] o/ Ironholds [01:01:43] I just discovered a python library that might be awesome. [01:01:45] https://boltons.readthedocs.org [01:02:37] I found a few utilities that I really appreciate. [01:02:42] ooh [01:02:44] How you doin'? [01:03:53] * halfak drools @ https://boltons.readthedocs.org/en/latest/tzutils.html#module-boltons.tzutils [01:04:15] Although, that timezones set needs to be extended to something more reasonable. [01:06:31] halfak, pretty good! [01:06:41] I'm redoing my server so I can host the Party Trick Website [01:06:57] and I realised the wonders of DO is I can just spin up a new VM, build in that, switch the DNS server over when it works, and boom [01:06:58] zero downtime [01:10:56] :) Sounds like what I've been looking for. [01:11:13] How do you access the other VM before you point the DNS? [01:15:41] raw IP! [13:20:01] bleeh [13:31:25] g'morning Ironholds [13:31:33] hey halfak :). How goes? [13:32:24] Ironholds, rough morning. Lots of squash yesterday. I feel like someone beat me with a shovel while I was sleeping. [13:32:40] But otherwise good. How about you? [13:32:40] aww [13:32:52] pretty good, but node errors are making installing shiny on my new machine a pain [13:32:53] * Ironholds sobs [13:32:56] *I want my new website* [13:33:04] Shiney uses node? [13:33:08] *-e [13:33:24] yeah, it's got a JS-based chunk [13:33:32] also a small amount of ruby in the build scripts. ech. [13:33:42] https://github.com/rstudio/shiny-server/ them stats [13:36:20] stats? [13:36:39] Oh. Language proportions [13:36:54] "R 2.3%" [13:42:32] Ironholds, I'm curious what your thoughts are on this: http://www.firstmonday.dk/ojs/index.php/fm/article/view/4901/4097 [13:43:16] I will give it a read! [13:43:38] * Ironholds reads [13:43:46] halfak, from the abstract I'm suspicious [13:43:51] Yup. [13:44:12] because it reads like "I took an Evegeny Morozov book and substituted 'big data' for 'slacktivism' wherever I found it" [13:44:34] like, the abstract could seriously be a review of The Net Delusion with maybe 10 words changed [13:44:38] but, will read, will give thoughts [13:45:05] * halfak is struggling to figure out what slacktivism has to do with population modeling. [13:45:14] Is it just optimizing for lazy + feel good? [13:45:59] oh, no [13:46:04] Morozov writes a lot of books about [13:46:25] "we thought the internet would make us good at everything and solve for democracy but instead it turns out turing-complete tools, by definition, CAN ACTUALLY BE USED FOR ANYTHING! AAAAAAH!" [13:46:43] and his most recent book is about slacktivism, where "anything" is "popular participation in creating change" [13:46:58] he just sort of substitutes the terms out every couple of years and gets another visiting fellowship somewhere [13:47:08] Oh... yeah. I think that in this case, Tufecki is making the "the inevitable consequence of marketing" argument. [13:47:16] aha [13:47:41] But she wants to argue that we should have renewed concern over the effect that marketing has on our social context when it infects the affordances of the software we use. [13:48:04] Which is an argument I believe. [13:48:29] "click optimization" inevitably leads to more clicks -- and it's hard to optimize for the user actually getting value. [13:48:38] So we just optimize the clicks. [13:49:35] I agree with her conclusions. I disagree with her first consequence [13:49:58] "A home judged as �non�voter� can be skipped while the next one will be flooded with campaign material, thus introducing a new form of categorical inequality into the public sphere" - this already happens, just at a broader level [13:50:20] the difference for me between nobody giving a crap about my opinion on politics because I want to set the republicans on fire, and so am already spoken for [13:50:41] and nobody giving a crap about my opinion on politics because I am from MA, therefore assumed by want to set the republicans on fire, and so am already spoken for, is null. [13:51:19] but let's suppose that the inverse is true, right? I'm assumed to vote democratic because I'm in MA, but actually I'm a republican. [13:51:42] at the moment, with high-level targeting, my values or opinions are not factored into discourse or policy at all. They're just assumed from the people in my zip code. [13:51:57] Individual-level targeting could actually increase the number of people messaged. [13:53:13] but I do agree with her conclusions. [13:53:41] I think she could do more to expand on something she mentioned in her discussion/conclusion, but not so much in the body, and elucidate on that - and that's the proprietary and locked-up nature of these databases [13:53:55] +1 That's my sense too. I appreciate the things that Tufecki calls attention to, but it seems to me that her stuff lacks technical insight. [13:54:05] I think that ties directly into the problems she surfaces around evaluating ideas on their merit and structural inequality but from the other side - from the political side [13:54:37] that is: if all the demographic targeting information is locked up in proprietary databases that can charge ludicrous amounts for exclusive access, it further separates the two major parties from any upsets to that system. [13:54:57] it isolates them even more from popular reform [13:55:14] (I'm reading this with my used-to-campaign-for-a-third-party hat, not my research code jockey hat ;p) [14:34:13] Hello, world of science. [14:34:41] hey guillom! [14:34:46] guillom, european question for you [14:35:04] do you consider the invention of "olives stuffed with $FOO" to be in the top 5 or 10 human creations of all time? [14:35:04] Hey Ironholds [14:35:28] The only thing olives are good for is making olive oil. [14:36:10] ... [14:36:14] THE DISEASE IS SPREADING [14:36:16] BURN OUT THE DISEASE [14:42:43] Nettrom, Nettrom! [14:42:48] finally, a human being of sense [14:43:03] Do you consider the invention of "olives stuffed with $FOO" to be awesome or really awesome? [14:53:05] Stuffed mushrooms, however, are another matter. [14:53:37] I hereby strip you of your status as a european from a nation bordering the Mediterranean [14:53:48] and bequeath unto you the lesser status of "possibly American" [14:53:54] * Ironholds holds out hand [14:53:57] med-card, please. [14:54:38] Considering the amount of cucumber, tomato and feta cheese I usually consume as soon as they're in season, good luck with that card-stripping :p [14:54:57] hmph ;p [14:56:14] well-made olives stuffed with $foo is awesome [14:56:26] the mass-produced ones you usually find in the salad bar: not so awesome [14:56:38] Nettrom, THANK YOU [14:56:43] I'm surrounded by olive-haters [14:56:51] * Ironholds hugs his queen spanish olives with pimento [14:56:57] hating on the olives is ridiculous [14:59:42] BTW Ironholds I'm sorry for having abandoned you a bit this week on the Wikidata analysis. Between tech news and VE (and covering a bit for Sherry & Erica), I don't have a lot of time left. Do let me know if you're blocked by me on anything. [15:00:00] word [15:00:00] guillom, so, on wikidata [15:00:00] Wikidata only has one date/time type, and again it's not explicitly called out and is only defined at the property level >.> [15:00:40] That seems to be a pattern (definition at the property level) [15:00:45] oh, that's fine! [15:00:45] I'm not blocked; I'm gonna finish up everything but the infoboxes by the end of the day [15:00:45] (if you've got infoboxes?) [15:00:46] yeah [15:01:12] to be honest, the conclusion I'm drawing here is that we have pretty much two major blockers: first, the uneven level of localised labels and descriptions [15:01:18] and second, the lack of infrastructure to support nuance [15:01:49] we don't have importance, items only identify themselves as images or dates if you maintain a constantly-updated list of what properties can be of those types (or gain that list with each request) [15:02:09] image and reference usage is pretty much nil :/ [15:02:50] Gotcha; looks like you've got things under control :) I'll put infoboxes on my list for early next week. [15:03:08] *thumbs up* [15:03:17] honestly I can probably write the conclusions without those :/ [15:03:29] That's totally fine with me [15:03:34] like, if it was "I'm unsure" or "I think there are options", I would wait [15:04:06] I can say that I didn't have time to look into them; that's fine I think. We can still look into them later if needed or desired. [15:04:42] Yeah, agreed. The state of Wikidata with regard to infoboxes is pretty much "heh, try DBpedia instead?" [15:06:52] * guillom brb. [15:11:54] :( [15:13:55] Sorry, didn't mean to kill the mood. [15:14:16] * guillom isn't sleeping well these days. [15:14:22] that's okay! [15:14:24] (which makes me cranky) [15:14:32] and naw, I was :( at the state of wikidata being that bad. [15:15:26] On the plus side: lots of opportunities for easy micro-contributions, like porting infobox content to Wikidata. [15:15:46] I imagine much can be done automatucally, but some will require a human brain. [15:16:24] yeah [15:16:42] but that's an entirely different game (editing) and set of questions (community fragility and oversight) [15:17:31] yep [15:31:40] guillom, think counts of properties per page and references per page would be useful? [15:33:19] Ironholds: I don't know; what question would that answer? (I'm genuinely asking because I haven't thought about it.) [15:33:59] whether there's even enough information for {placeholders, short descriptions} and whether the information can be "trusted" enough for the community to go for it [15:35:48] I'm not sure the counts of properties is very meaningful, since it's difficult to know which are "important". An item could have 10 properties and non of them telling you what the item actually is (to take an extreme case). [15:36:24] For references, Maybe a better measure would be the proportion of referenced statements, rather than references per item. [15:36:36] none* [15:39:29] ah, yes [15:39:38] that's what I meant by references [15:41:26] I think it would be useful, yes, although I'm afraid you need to have low expectations if you don't want to cry when you see the results :p [15:41:54] yeah, I am ;p [16:04:25] ...huh [16:04:31] guillom, the reference level is actually really good [16:04:50] Oh? Well, that's good :) [16:05:02] But how often is the reference "XX language Wikipedia"? [16:08:43] yeah, point :( [16:57:55] halfak: in the Wiki-Class repository, can the doc/index.rst file be edited, or is it generated from somewhere? [16:58:11] It can be edited. [16:58:23] All of the *.rst files are human edited. :) [16:58:33] cool, was slightly confusing when I looked at it [16:58:49] wanted to updated it with the new model [17:02:39] standuppers [17:02:59] FYI: company-wide meeting running late [17:03:06] We'll probably be ~5 min late for standup [17:03:12] oh you guys are all up there? boo [17:03:43] :P [17:03:50] I'm not up anywhere. [17:03:54] I am in the INTERNETS [17:04:11] you live on 5, I saw you [17:04:33] I still need that pic :P [17:55:45] I missed the standup because CA meeting, right? [18:04:02] CA? [18:04:05] Ironholds, ^ [18:06:39] halfak, community advocacy [18:06:51] I've been cheating on your cabal with another cabal [18:06:58] WHAAAA [18:06:59] k [18:07:00] :) [18:08:41] leila: Do you have some time this afternoon to help me with EventLogging for the apps experiment we talked about on Tuesday? [18:19:50] Deskana: sure. I'll send you an invite. [18:19:59] leila: Thanks. I'm free most of the afternoon. [18:20:11] Which is incredibly rare. [18:21:47] leila: Thanks! [18:21:59] sure, Deskana. see you then. :-) [18:30:34] aaaaufgh [18:30:37] * Ironholds headdesks [18:30:41] guillom, so you know date-times? [18:30:50] All I know is ISO. [18:30:54] they can exist in different calendars. So you can't calculate the relative DT without knowing the calendar. [18:30:57] in wikidata [18:31:00] The rest doesn't exist. [18:31:17] THERE IS ONLY ISO. [18:31:20] so a time is "+00000002014-05-11T00:00:00Z" [18:31:28] and then it links to the Julian calendar so you know when that _actually was_ [18:31:34] or the proleptic gregorian. Or whatever. [18:31:36] guess what? [18:31:44] the logic on how to handle all of that: in a totally different property. [18:32:04] also, dates and times can have varying degrees of precision or range [18:32:10] varying timezones not reflected in the actual timestamp [18:32:20] and all of this is handled by one datatype, for both dates and times. [18:32:32] if you want to just specify a date, you just reduce the amount of precision! [18:32:33] :D [18:32:51] Deskana|Away has the right idea [19:50:02] * Ironholds giggles [19:54:10] hahahah [19:54:11] IT WORKS [20:01:58] halfak, it works! [20:01:59] http://ironholds.org/ [20:02:04] behold: a website written entirely in R. [20:02:16] Nice :) [20:03:53] * Ironholds giggles [20:04:29] Schoonover has nothing on my language fanaticism. [20:04:31] A few months ago, I considered trying to make a website in LaTeX. [20:04:34] His coco obsession is /weak sauce/ [20:04:43] IT was too much bother and too little fun. [20:04:47] It* [20:05:59] :D [20:06:07] * YuviPanda wrote an URL Shortening HTTP server in C once... [20:06:18] YuviPanda, https://twitter.com/zenrhino/status/586620936608096256 is the appropriate reaction [20:06:25] I wrote a website in markdown and R. Show me some love. [20:06:36] :D [20:06:40] Ironholds: <3 [20:12:36] hahahah [20:12:44] DarTar, wanna see something insane? :D [20:13:24] I always want to see something insane [20:13:37] DarTar, http://ironholds.org/ see this website? [20:13:51] Go to the "research" tab, see how all the talks are in this beautiful sortable JavaScript table? [20:13:52] lemme guess, entirely generated in R [20:13:55] yep [20:13:59] R and CSS and love and cackling [20:14:05] the entire website is made of R. The entire goddamn thing. [20:14:09] ha ha [20:14:28] reminds me of the good old days when I was beating TeX to produce any kind of possible output [20:14:38] DarTar, guillom was just talking about writing a website in TeX! [20:14:48] and yeah, I love your LaTeX work [20:14:56] "The Beauty of LaTeX" is still one of my favourite things ever. [20:14:57] say, did you get your ORCID? [20:15:07] you should get one [20:15:12] I did! orcid.org/0000-0001-5196-609X [20:15:18] \o/ [20:15:36] alright, gotta run (FR meeting), nicely done with the website [20:15:58] I'm annoyed that you can't have an image for orcid [20:16:16] A photo is important for identifying a researcher. [20:17:12] DarTar, take care! [20:17:44] Speaking of headshots [20:17:50] Ironholds, you should add one to your website [20:18:11] uhh [20:18:18] I would but I haven't worked out how to make images work yet [20:18:23] it's not perfect, but it's pretty ridiculously good. [20:18:26] It's only 50 lines of code! [20:19:02] Plot a figure for your face :P [20:19:33] dev.website(); plot(myface.jpg); dev.off() [20:22:14] halfak, oh, that bit I thought of, it's getting it to size properly that gets problematic [20:25:24] DarTar: Fun fact: you can use Latin Modern fonts as webfonts to make websites look like TeX. (I do it on mine.) [20:25:47] guillom, link? [20:26:12] halfak: to webfonts? [20:26:19] to your website :) [20:26:27] oh [20:26:45] https://guillaumepaumier.com/ [20:27:28] Not nearly as hardcore as Ironholds's. [20:27:40] Looks good though. Nice to read :) [20:28:15] s/hardcore/stupidly overcomplicated as a solution/g ;) [20:28:17] Everything looks good in Latin Modern & Solarized colors. Even IRC :) [20:28:42] Ironholds: But that's the fun, right? Overly complicating things. [20:28:53] yep! [20:28:57] next up is an R-based IRC bot [20:29:29] One day you should do an R workshop. To initiate other people to that Dark Magic. [20:30:54] at the WMF or generally? Because actually I'm doing a talk at Hack/Reduce on the 23rd, and going to CM next month [20:31:01] well, "going to". "Presenting remotely at" [20:31:16] CM? [20:32:01] I was thinking at the WMF. [20:32:12] Like a brown bag maybe, and if interest a full day workshop [20:32:39] We clearly have experts in many fields / tools and we'd benefit from more sharing. [20:33:02] Although I'm also fine with "Just throw all data analysis requests to Oliver and let him do his R thing." :p [20:33:33] well, that sounds pretty unsustainable in the long term ;p [21:06:08] "Editing Editing" http://imgur.com/52T0mi6 [21:26:16] halfak: smart languages use quotes :) [21:29:46] Nemo_bis, what? [21:29:59] * halfak thinks of 5 things this could be referring to [21:30:02] :P [21:31:22] 'Editing "Editing"' [21:43:09] Nemo_bis, ahh yes. That would make more sense :) [21:43:54] Use « French quotes ». They're prettier! [21:44:38] Nope. Not on my keyboard. Doesn't exist. Must ASCIIify. [21:44:43] :P [21:44:46] Better if not » reversed « [21:45:00] halfak: which OS? [21:45:10] Ubuntu 14.04 [21:46:08] halfak: did you try alt-gr + < [21:46:23] What's "gr" [21:46:25] ? [21:47:31] hmm [21:47:36] The right alt [21:47:39] The window key? [21:47:43] no [21:47:46] Nemo_bis: There's no alt-gr on US keyboards [21:47:52] Just two alts [21:48:04] Well, ctrl + alt + < then [21:48:15] Ah but that's different [21:48:19] Ok [21:48:26] Hmm... That doesn't seem to do anything. [21:48:33] Sorry, alt + shift + < [21:48:50] I have to do Alt+Z on mine, using a French mapping for my US physical keyboard [21:49:28] Nemo_bis, no effect with alt + shift. [21:49:30] * halfak googles [21:52:12] Ok, enough with pressing random key combinations ^^ [21:52:24] :)