Man, every part of this blog post sucks.
-
Man, every part of this blog post sucks.
Eternal November — this new influx of users may be better than the last one
The Software Freedom Conservancy provides a non-profit home and services to Free, Libre and Open Source Software (FLOSS) projects.
Software Freedom Conservancy (sfconservancy.org)
This is what happens to discourse when the focus is on whether AI tools "work," necessarily a complex and shifting topic that gives bad-faith actors lots of room to sow confusion, and not on the ethical catastrophe caused by adopting or allowing AI products into OSS development processes.
@xgranade I love how this says nothing about the fact that letting LLM code into copyleft codebases makes value leak out of the project: https://www.quippd.com/writing/2026/04/08/ai-code-is-hollowing-out-open-source-and-maintainers-are-looking-the-other-way.html
We're just supposed to say "it works, don't it?" and look the other way while contributors get robbed.
-
@pathunstrom Yes, this. I would also argue that slop is far worse than spam, if only from a pure ethical perspective. Put bluntly, spam isn't inherently fascist tech, but slop is, and I wish that blog post even *mentioned* the ethical problems with letting fascist tech into OSS.
@xgranade @pathunstrom Spam, the meat product:
1. Full of fat, cholesterol, salt and sodium nitrite
2. Made in a factory that has indulged in union-busting, exploited immigrant labor, and given people autoimmune diseases from inhaling aerosolized pig brains (https://www.motherjones.com/politics/2011/06/hormel-spam-pig-brains-disease/)
3. Foisted on the people because of racism (Spam dominated the Hawaiian diet in WWII because the US barred people of Japanese descent from fishing)Maybe the canned meat product is an even better analogy than the unwanted email....
-
@xgranade oh, you know I agree there! It's just easier to convince boosters that spam is bad than llms are bad, and thus easier to wedge into "so if thing primarily produces spam, and spam bad, we should be heavily skeptical of thing."
@pathunstrom No, of course, sorry for preaching to the choir. I'm just still mad at the original blog post for acting like the problems with AI are at worst the problems caused by spam.
But I absolutely take your point and agree.
-
Look at what happened with Claude Code. We learned via the source code leak that the whole thing is a Rube Goldberg machine of shoddy regexes and Markdown snippets telling Claude to lie, and yet proverbial moments later, Anthropic announces a new product that *totally works this time I swear* and all of a sudden discourse about AI tools "working" is completely reset.
Getting stuck in that discourse loop opens you up to being perpetually distracted from the far more important ethical problems.
@xgranade this is like when equifax got hacked for not doing security basics, then got away with lying to congress about it, then offered "free credit monitoring" to the affected people. if equifax can get away with it like that, and then later facebook with the whole cambridge analytica deal, anthropic has an idea-style howto manual on how to do it next
-
@xgranade They linted the regexes this time.
-
@xgranade This is the pettiest of complaints but _right off the bat_ the author gets their folk history badly wrong.
I have a lot of complaints about the Eternal September framing in general, but forgetting that it was Eternal because it was the _general public_ getting access (through AOL, principally) and only September because it reminded old-timers of the annual new student rush, is appalling.
Grr argh.
-
R relay@relay.publicsquare.global shared this topic
