[02:44:15] In moderate defense of ORES - the fact is not a lot of innovation in Wikipedia has happened since 2018, so the accrued knowledge of ORES from years of community input is still very useful and relevant. For example, the vast majority of low-quality edits in Wikipedia from bored high school students amount to "poop," article blanking, and expletive-laden ALL CAPS edits that has har [02:44:15] dly changed since the early days of en.wp. :) (re @gtisza: I assume this was referring to the Automoderator project, which is intended to be cross-wiki. [02:44:17] [02:44:18] ORES is very costly. Its revert-b...) [06:25:39] Hello everyone, I'm new here, let me make friends, please take care of me [06:26:28] hi. what's your connection to the wikimedia projects? (re @jkylo189: Hello everyone, I'm new here, let me make friends, please take care of me) [07:03:11] Almost certainly spam [07:03:29] yeah, hence my question (re @tehreedy: Almost certainly spam) [13:57:36] and account is now deleted. I downloaded one of the profile pix before, didn't get a hit in tineye. I thought If there was a hit that would seal it. (re @tehreedy: Almost certainly spam) [14:12:48] A lot of innovation happened in spamming tools due to the LLM revolution, though. And also AI is a fast-moving field so something that probably counted as modern in 2015 is could probably be upgraded a lot today just by applying the tools and best practices the field has discovered since. (re @fuzheado: In moderate defense of ORES - the fact is not a lot of innovation in Wikipedi [14:12:49] a has happened since 2018, so the accrued knowledge...) [14:19:17] But in any case, the point is not that ORES is bad, it's that it takes multiple editor-day per wiki to train and for most wikis that is just not feasible. The 'reverted' model doesn't need editor involvement but 1) I don't think it's very good, 2) even that takes manual work by developers, so there is still a scaling issue although a milder one.