[03:53:48] bmansurov: i saw the patch is ready for deploying. thanks! i quickly scanned it and looked good to me. i'll get up about ahead of deployment so i'm around in case there are questions and to help with testing as needed [10:50:09] bmansurov: In https://phabricator.wikimedia.org/T216750#5279352, I did not get what this part of SparQL query(?article schema:isPartOf ) is searching for exactly. Could you please explain it? [13:22:34] o/ [13:22:48] \o [13:25:52] isaacj: qq. [13:26:34] isaacj: is there a reason that we have Other as an option for the the familiarity and depth of information questions? We left them without Other in the previous surveys [13:26:50] (except the very first ones where we weren't sure if there are actually Other options) [13:29:01] isaacj: and congrats. great to see the responses coming in. :) [13:37:35] thanks leila - I kept other in because it's been a few years so making sure the taxonomy still applies felt important and I find it frustrating as a survey taker when I don't have an other option for more open-ended questions like that [13:37:56] isaacj: you're right. we discussed this. thanks! [13:38:03] isaacj: is everything going well with the surveys? [13:43:51] leila: yeah, as far as i can tell. minor hiccups with russian / norwegian where at first i had accidentally added in logic to the survey ID question that forced it to be a number but that's fixed and responses are coming in for all and looking normal [13:44:31] isaacj: great. :) I see a couple of questions in research-feedback that I'll respond to. [13:44:54] leila: i responded to the one on "impossible to finish the survey" [13:45:15] there was also a blank email. let me know if there are others [13:46:05] isaacj: ok. great. (My email is trying to refresh) [13:47:04] yeah, they come through to my inbox so my responses don't show up for others. i'll switch to using the google group so it's obvious they've been addressed [13:51:45] isaacj: sounds good. [13:52:04] (I'm very excited that this is out of the door. Thanks so much for all your hard work on it.) [18:22:21] starting talks in ten minutes -- really interesting set of talks all around blocks on Wikipedia so hopefully lots of questions / discussion today! [18:29:53] o/ Hi everyone! [18:31:24] hello! [18:32:03] Hi! [18:33:11] o/ [18:36:42] Can we get a link to the report or paper relating to Jonathan's research? [18:37:13] hey @Risker i believe this is it: http://www.cs.cornell.edu/%7Ejpchang/papers/recidivism_online.pdf [18:37:26] (the academic paper) [18:37:45] yep, that's the one [18:38:09] thank you [18:41:31] why was the choice talk page activity? from the community perspective, that's not a very helpful indicator [18:42:21] @Risker -- i'll pass along when we get to Q&A but could you elaborate on why it's not helpful [18:42:37] gah, ignore the @s [18:43:20] too many platforms with different norms [18:44:27] something i've been thinking about with regards to term "recidivism" for framing is that, when I moderated, the point of a block was not always to reform [18:44:27] no worries, isaacj: given the majority of user talk page interactions over the last few years have been templates (both giving and receiving), I would think "engagement" would be better identified by participation in Wikipedia space, or alternately article talk pages [18:45:30] ahh, makes sense -- thanks! [18:45:38] which makes it quite different to a criminal reform frame, because banned users don't necessarily have an incentive to rejoin, and an account is far more disposable than... you. [18:46:34] No questions, but just want to say that this is really interesting [18:46:39] but this isn't a question so don't pass it to Q&A XD [18:46:58] anyway, in my experience, blocks can be for the purpose of explicitly removing someone from a space, or simply to communicate that rules are being enforced, or all of the above + teaching newer mods how to place blocks [18:47:57] hmm...interesting cwylo : any thoughts on to what degree they are used for these different purposes? [18:47:57] would also point out that apologies can sometimes be used as face-saving insults! Something we surfaced in our study on harassment on Arabic Wikipedia. [18:49:05] isaacj: difficult to quantify because a ban can be used for all of those motivations at once; mostly, it's an assumption I've noticed in many papers that study online community moderation, or rather, bans in online community moderation [18:49:26] hmmm. wondering about people who've just plain been around long enough to know what triggers to pull to get themselves unblocked [18:49:29] shameless self promo: chp. 4 of my thesis [18:50:00] talks about the social work of moderation, though on Twitch esports moderation, but I think the principles are analogous [18:50:37] yeah, makes sense cwylo . nothing ever fits neatly into classification tasks :) but definitely important when considering what the expected / desired outcomes of blocks are [18:50:49] https://meta.wikimedia.org/wiki/Research:Arabic_Harassment_Netnography re: study on harassment on arwiki [18:51:27] Yeah. I can only speak for my experience, but I've handed out (limited) bans as warnings, as demonstrations, but also as appeasement before [18:51:54] It's tricky. Bans online carry differing proportions of weight depending on the community :/ [18:52:47] Thanks Jonathan....interesting [18:53:22] tl;dr i should write this down and publish somewhere instead of just keeping it on IRC [18:54:22] thanks for the response. [18:56:20] yep, keep the questions coming! [18:57:33] meta page for Lane's project: https://meta.wikimedia.org/wiki/University_of_Virginia/Automatic_Detection_of_Online_Abuse [18:58:57] isaacj: Do you know if Jonathan's slides are available? I'd love to have them for a more detailed reading! [18:59:47] Niharika: i'll check with him and see about updating the mediawiki page (https://www.mediawiki.org/wiki/Wikimedia_Research/Showcase) w/ the link [18:59:55] Thanks! [19:10:22] Question: how is the software used in this study different from the software that was recently shut down that looked essentially like a bad word detector? [19:10:25] Q: how is this toxicity detection system distinguished from e.g. Google Detox? (Asking as a non-data scientist.) [19:10:29] oh hah. same question. [19:10:45] cwylo: snap! absolutely true [19:11:23] I should disclose, though, I did some minor consulting on this, so I can point towards their paper but don't want to put words in Lane's mouth [19:12:05] thanks -- will pass along [19:13:43] "disproportionate weight on decontextualized content of a comment", for a fancy way of saying "bad word detector" :P [19:14:08] The full pre-print paper is on Commons, as well. https://meta.wikimedia.org/wiki/File:Automatic_Detection_of_Online_Abuse_and_Analysis_of_Problematic_Users_in_Wikipedia_preprint.pdf [19:14:57] Relevant details on extra modelling starts on p4, majority on p5 [19:15:14] Risker ^ [19:17:11] Measuring the effectiveness of blocks: https://meta.wikimedia.org/wiki/Community_health_initiative/Measuring_the_effectiveness_of_blocks [19:18:13] thanks, cwylo [19:26:41] Q: What user action is resulting in partial blocks? are they used as a "step-down" from a full block? [19:27:19] isaacj ^ [19:27:28] gotcha - thanks! [19:29:02] many thanks to the presenters, and thanks isaacj for conveying my questions :-) [19:29:30] happily! thanks for asking Risker [19:30:09] A note that both of the papers presented today have not yet been covered in the wikimedia research newsletter / wikipedia signpost yet - volunteers are welcome to write a review or shorter summary: https://etherpad.wikimedia.org/p/WRN201906 [19:34:32] yes, a lot of interesting papers in there right now! [19:36:38] thank you to all the presenters! [19:36:45] Great job, everyone! [19:37:03] thanks everyone! [19:37:17] thanks for the great presentations and thanks for having me! :) [19:37:25] pretty interesting, thanks a lot! [19:38:35] Thanks y'all. Super interesting work! I'd love to get my hands on all the slides! [19:39:11] yes, request was sent -- i'll follow up when the PDF is available [19:44:32] just got mine up on Commons and added the link to the Showcase page. There's a direct link to the PDF: https://commons.wikimedia.org/wiki/File:First_Insights_from_Partial_Blocks_in_Wikimedia_Wikis.pdf [19:45:02] I'll get the graphs up on Commons on Friday [21:30:46] Niharika: https://commons.wikimedia.org/wiki/File:Trajectories_of_Blocked_Community_Members_-_Slides.pdf [21:31:20] isaacj: Thank you! [21:32:05] quite welcome [21:44:37] isaacj: would you be willing to link that on the uh [21:44:47] https://www.mediawiki.org/wiki/Wikimedia_Research/Showcase here? [21:44:57] because people will likely go there first for info [21:45:03] sure, thanks for the reminder apergos [21:45:26] also thank you for these slides and I will be reading the paper! [21:46:18] done! and yeah, it's exciting to see so much work in this area [21:48:59] it's certainly timely [21:50:06] the folks that figure out how to deal effectively with discussion/comment moderation at large scale will be writing history, if it ever happens