[00:44:07] would it be worth digging in the SQL to modify https://issue-tracker.miraheze.org/phame/post/view/19/fsslc_the_actual_next_step_in_miraheze_s_growth/ to show being authored by agent again [00:45:19] we should use the blog ore [00:45:43] we need to get people who are interested enough in writing news stuff [00:45:57] at the least we should do a recap at the end of the year IMO [00:57:07] an EOY recap is planned, yes [00:57:26] WikiTide had one last year but it got nuked when WT Phab went down 😄 [01:01:55] I think Void changed that to me. Could change back to Agent yes. [01:02:43] I miss Unknown Object [01:03:02] It caused a fatal on the blog lol [01:03:29] That is why it was changed. Unknown Object doesn't work for author there. [01:03:34] oooooo interesting [01:06:56] maybe its been patched upstream since [01:06:56] idk [01:07:01] x to doubt [05:27:56] [1/2] this is going to be a long day.. [05:27:56] [2/2] https://cdn.discordapp.com/attachments/1006789349498699827/1293807205899829339/image.png?ex=6708b75b&is=670765db&hm=ae271606dc257888eefbbc941524364bf5027523c6c11cbfc36cbb1c3190d997& [05:52:25] man [06:13:16] new idea: https://possum.city/notes/9srbrg3hc3xy006n [07:22:06] @pixldev what tool do you use for those nice graphs? So I don't have to bother you each time I'd like to make one of my own [07:22:37] I want to see if spambot creation decreases are so significant that they can be noticed on a user account creation graph [12:07:17] I’ll Send you the script later [12:07:26] But it’s really not that complicated [12:07:36] It’s matplotlibs I made with copilot [13:07:19] Thanks! [13:07:25] I'm addicted to graphs now 😄 [13:08:00] I'm really curious about account creation after we introduced hCaptcha and mass banned VPN and proxy registration [13:11:20] RhinosF1: https://issue-tracker.miraheze.org/T12744: wait why configuration? [13:12:00] we always tag renames with db + config [13:12:28] oh okay, til [13:13:05] historically i've seen tech team (mw) + database :p [13:36:36] no it's an infra problem [13:48:11] The data! [13:49:01] I do wanna figure out how to graph the slope of the graph [13:49:39] Huh? [13:49:47] huh? [13:50:19] So isn't it an configuration change? [13:50:49] i've seen wiki rename requests tagged as tech team (mw) + database before, but i suppose the correct way is tech team (infra) + database + config [14:24:43] there is a bit of config to change [14:24:53] but most of the work is moving the database around [17:22:25] @rhinosf1 did you see that [[mw:Extension:AutoModerator]] is starting to be rolled out on WMF (if I recall right) [17:22:26]

[17:27:25] Wanna mess with it [17:27:39] May be able to help supplement CVT [17:27:45] Sure [17:29:51] Do we know anyone deving it @pixldev [17:31:16] Let me see the list [17:31:34] https://www.mediawiki.org/wiki/Moderator_Tools [17:31:45] > Jason Sherman (software engineer), Susana Cardenas Molinar (software engineer), Katy Graessle (software engineer), Dennis Mburugu (engineering manager), Olga Tichonova (designer), Krishna Chaitanya Velaga (analyst) [17:31:47] @theresnotime you have any friends there? [17:31:49] Yeah this is more your circle [17:32:13] I don’t know any WMF people besides James and formerly kunal [17:32:16] That I can recall [17:32:21] I recognise Jason Sherman from phab but never spoke to them properly [17:32:27] @pixldev which James [17:32:57] Forrester [17:33:04] The British one [17:33:35] He’s nice, spoken with him a few times [17:33:56] He gave an awesome tour last time I saw him at the wikinic [17:34:23] literally a picture of Sam Walton in the embed :p you know them surely! [17:36:02] Pfffft [17:36:10] (what's the question though?) [17:36:31] Do I? [17:36:37] I know @freosam [17:36:43] Not sure about Walton [17:36:59] It is like 18:30 and I've been on the go since 7am [17:37:01] Though [17:38:24] You'd need to have a way of using the ML models for AutoModerator though [17:38:52] How much ML [17:39:01] Also why would I know Sam Walton [17:39:07] You're confusing my tired brain [17:39:19] And what type of ML [17:39:23] Just fairly well known in the community ^^ [17:39:49] Oh that I can't remember, me and him had a good conversation about it at Wikimania but I've forgotten since [17:39:52] What's he done [17:39:58] Oh [17:40:06] Can you bring him here to talk about it [17:40:11] The Wikipedia Library guy [17:40:28] Now a manager [17:40:37] Come on south western, provide me good train WiFi please [17:41:56] I know of him then but not him [17:43:40] I think [17:43:41] Not sure I've ever like properly spoke with him on discord / IRC or the like 2 video calls I did in Covid [17:43:41] When we had virtual Wikimedia hackathon [17:43:41] Test [17:43:49] _wonders which of SWR, O2 or Discord are to blame_ [17:44:48] https://cdn.discordapp.com/attachments/1006789349498699827/1293992645684625408/Screenshot_20241010-184430.png?ex=6709640f&is=6708128f&hm=83de8b244e3d62ca639285640d42772c384c95d62d41a335ac5fb77d077b687c& [17:45:06] https://wikitech.wikimedia.org/wiki/Machine_Learning/LiftWing [17:45:13] @theresnotime that sucks [17:45:15] :waaa: [17:45:34] LiftWing is a nice name, no idea about how it is to deploy though [17:45:56] @pixldev you want to train a liftwing model [17:47:01] I'd have thought that a very generic "vandalism" model would be fairly universal [17:47:32] @theresnotime I mean apart from on the vandalism wiki and maybe test wiki probably [17:47:33] But then you'd only catch the very obvious vandalism :AkkoShrug: [17:48:09] I mean the aim is to help with the easy stuff so cvt can focus on the complex ones surely [17:48:20] Ideally prioritise for cvt too [17:59:53] Automatically awesome person Then [17:59:54] Not really. [17:59:59] But I mean [18:00:11] So it’s made for Wikipedia [18:00:39] And? I still don’t see reason we can’t try it as an experiment [18:01:37] Will it work horribly? Quite possibly. Is this a stupid idea? Yes? Will it blow up the farm? Statistically unlikely. [18:02:15] Since when did we try sensible ideas [18:02:26] Sensible is boring [18:02:35] You don't get innovation with sensible [18:02:51] If it's stupid, try and see what happens [18:02:57] Worst case it's a bit of fun [18:03:03] Best case it's stupid and brilliant [18:03:16] I was elected global admin, so not within the past few months [18:03:38] And I personally have no intention of starting now [18:22:11] hmmm [18:22:42] This was basically my idea of making a python program using cluebot lol [18:24:22] @rhinosf1 so the issue would be the model [18:45:15] Right [18:48:51] [1/2] ok, so a little question, I wanted to make a website that logins with my wiki credentials [18:48:51] [2/2] regarding the OAuth thing, what I should do with it? [18:49:21] [[mw:Help:OAuth]] may help [18:49:22]

[18:54:12] Your site must provide a callback url which you must register with the consumer [18:58:52] alr, the only thing I need from that are none other than identify (and maybe email address) [19:01:49] Someone ghost ping me? [19:11:25] ? [19:11:50] the only ping that works was mine [19:15:57] The replt didn't come through [19:16:07] App is super laggy today [23:36:23] alright @rhinosf1 so how would we test out automoderator. Thinking of a basic roadmap [23:36:39] step one: get a model. Step two: test model [23:37:02] AutoModerator doesn’t currently support exporting logs to IRC or smt [23:37:14] so we may still need to do work on a seperate tool [23:38:05] So may need to go back to my first idea of a python system for logging and then move to AM when we’re confident enough in not causing mayhem [23:38:47] would testing be global like AbuseFilter or explicitly opt in cause people dont like AI