means testing plus algorithmic formulas are two terrible things that depend on one another. the same effects, people paying for social goods based on what they can afford, could be accomplished with less suffering and more fairness by making goods free at point of consumption and supported by progressive taxation ... but instead you get awful things like this: https://www.theguardian.com/global-development/2026/may/04/kenya-ai-healthcare-reforms-driving-up-costs-for-poor
this whole thread is great, but in particular this dicussion of llm code gambling is truly insightful
this is what the inside of my mind looks like, for anyone who doesn't know already
just a good rule to always abide by
@exquisitecorp@merveilles.town recently posted a great piece about no-input mixing that really aligns with my feelings about video and feedback performance (tho my practice is not quite so cheap, even if it also uses old materials): https://leetusman.com/nosebook/no-input
this is always such a jam