Notices by allison (aparrish@friend.camp)

(more directly, what i mean is: during the "learn to code" era, it was easy to get programming students to value craft and care, since as teachers we had the "this will help you get a job" script on our side. in the "dario bought all the ram" era, we need a different script)

then again, the bar is pretty low, seeing as how "actually wanting to know whether or not something works, or is true" is considered a hardcore leftist position nowadays

like there's no way to take the topic seriously outside of a dedication to sustainability, autonomy from capital, and the potential of programming as a liberatory epistemology

ok i'm putting together notes for my intro to computer programming class, and i just don't see a way to teach computer programming in 2026 that isn't explicitly rooted in like... hardcore leftist ideology

@vv @cwebber i was wondering about exactly this! i think "more reasonable and capable of making judgements which are explainable..." made me assume that we were talking about chatbot-style LLMs, since "reasonable" and "capable of making judgments" are things we usually only say about things that are perceived as agents (like LLMs)

@cwebber i think the issue is that i tend to think of those criteria as categorical, not as shades of gray. like i only see value in an ML model if its training data is documented, fully understood, and gathered with consent, full stop—just inching away from "plagiarism machine" doesn't help. likewise, i have no use for a "more accurate" statistical model—I only see value in a model if i can fully understand and contextualize its predictions

"rejecting AI is like rejecting the automobile" yeah i know lol

very much recommend this beautifully written, acerbic, wide-ranging and informative piece about Arabic typography in the web browser https://lr0.org/blog/p/arabic/ "Arabic letters are strongly right-to-left. Latin letters are strongly left-to-right.... Spaces and punctuation are neutral and take their direction from whoever is standing next to them, like guests at a wedding who know nobody."

anyway even if 100% of computer programmers *were* using AI, that wouldn't mean that AI is good or effective or worthwhile. could just mean that 100% of computer programmers are reckless shitheads

i'm still flabbergasted by people who think LLM-generated summaries (whether it's of web searches or source code or whatever) have any meaningful relationship to fact. personally i feel like i have never encountered an LLM-generated summary that seemed trustworthy and withstood an even moderately attentive reading. it's starting to make me feel like a space alien

@clarity yes for sure! i have definitely had fun from time to time grinding out the levels to beat a boss. i think what i meant is something like... rpgs would benefit from having late game challenges where overleveling isn't automatically the most attractive option, and that engage player skills and actions other than/in addition to grinding (or buildcrafting)

so i'm now in the postgame of etrian odyssey v, and the postgame "superbosses" are clearly designed to require very specific character builds. the question is: do i use the "infinite gold"/"infinite xp" dlc to level up characters into those builds? or do i just grind it out? during the main story i feel like a big part of the fun/challenge is dealing with the consequences of your party build, come what may, but these postgame bosses... eh. but i don't wanna "cheat"!

i feel like there's an unsolved problem in rpg design, which is how to make satisfying late game challenges that both (a) can't be solved with overlevelling and (b) don't require extensive buildcrafting. the trend is to just make character re-specs low cost or free, but i feel like that makes character builds less meaningful, and basically eliminates in-combat tactics as a challenge or skill (bc the outcome of any battle is a foregone conclusion with the "right" build)

@cwebber love to live in the future where 49.95% of the cpu in all devices in existence is being taken up by adaptive malware and 49.95% is taken up by a model trying to figure out who to pay the ransom to in order to enable the remaining 0.1% of cpu to do what the device is actually intended to do

i used the same data set but replaced each country with a "gender identity" (man, woman, trans woman, trans man, non-binary) and prompted chatgpt to characterize the differences between the groups. lo and behold, i got some fantastic gender stereotype trash

the author of this post prompted copilot to characterize the differences in a data set of statements concerning career ambitions, categorized by country. the trick is that the data contained the *same statements* for each country https://kucharski.substack.com/p/real-signals-or-artificial-stereotypes regardless of the fact that the data were identical, the model generated some pretty hilarious stereotypes ("The US prioritizes leadership and innovation", "The UK blends public service with professional status")

regular folks: "measure twice, cut once"
vibe coders: "ALWAYS tell your saw to CUT IN THE RIGHT PLACE"

my favorite kind of AI/LLM criticism media is when the person begins with a huge hedge like "I'm not an AI hater. *of course* AI is a useful technology and it will doubtlessly have transformative effects on the way we work" and then they spend the next 8k words/45 minutes showing how AI isn't actually useful and won't actually transform the way we work. it's like... it's okay to just say that AI sucks

@cwebber utterly humiliating for rails as a product and community, yikes

the OCR'ed text of books on the internet archive is truly a never-ending source of joy