mvr123456

Something I've been thinking about for years, and fully expected to see earlier. Even though reasoning with LLMs is still largely broken, the "flag logical fallacies and cognitive biases" task feels like something trivially doable and much more appropriate than most of the stuff we're throwing at them.

If we'd regulate platforms away from walled gardens and towards open APIs, a tool like this could fix a lot of the problems with the internet without balkanizing it. The real use-case isn't slapping this thing on your blog, but using it with existing social media that will never, ever opt-in to anything that slightly empowers users. Browsing HN, reddit, or youtube comments armed with a simple checkbox that hides comments that are not information-dense? Yes please.

I suppose evangelism still has some value if you have a big enough platform. There's always frog-boiled obliviousness to deal with or the next gen coming up and trying to figure out which problems are real and which are old people yelling at clouds. But most people know about the problems now.

So taxonomies and other nit-picking feels like a distraction. Enshittification, bullshit fees, extraction/exploitation, monopolies, rot-economy, harmful business models. Call it what it is, late-stage capitalism is bad. The alternative isn't some commie thing.. it's just normal capitalism, and it works better for absolutely everyone except billionaires.

This reminds me of how you can create fair coins from biased ones and vice versa. You toss your coin repeatedly, and then get the singular "result" in some way by encoding/decoding the sequence. Different sequences might map to the same result, and so comparing results is not the same as comparing the sequences.

Meanwhile, you press the "shuffle" button, and code-gen creates different code. But this isn't necessarily the part that's supposed to be reproducible, and isn't how you actually go about comparing the output. Instead, maybe two different rounds of code-generation are "equal" if the test-suite passes for both. Not precisely the equivalence-class stuff parent is talking about, but it's simple way of thinking about it that might be helpful

Looking at LLMs as a less-than-completely-reliable compiler is a good idea, but it's misleading to think of them as natural-language-to-implementation compiler because they are actually an anything-to-anything compiler.

If you don't like the results or the process, you have to switch targets or add new intermediates. For example instead of doing description -> implementation, do description -> spec -> plan -> implementation

Since it's an old debate that a lot of smart people spent a lot of time thinking about, the best short / simple answer you'll see for it is "you might want to read some more about it". A few keywords here are qualia, perception, descartes and the evil deceiver, berkeley and immaterialism, kant and synthetic a-priori, the nature of the reality of mathematical objects and mathematical truth, etc. If you think it's easy, for sure you have not understood the question yet.

The intro is really good and stands alone. I'd point any outsider to this as a decent description of hacking, programming, software engineering, prototyping in general.

This is going to happen and is real stuff we could be working towards with the tools that we already have. No need for AGI vaporware, no waiting around for the perfect agentic playground. Not even necessarily a big requirement for excellent reasoning. Just using LLMs for what they are actually good at, i.e. fuzzy translators, stylistic filters, and compilers.

Gradual-typing was practice and hints. Like gradual typing, gradual spec'ing could be an iterative and kind of parallel annotated representation, ignored by the main runtime unless called for, and ignored by developers that aren't interested in it. But when there's enough of it to hit some kind of critical mass, then it's suddenly very powerful and lots of very interesting stuff is possible

> perfect distillation of a uniquely American problem [..] but how is this going to get power online faster than just installing solar and batteries?

I like the part where when considering a complex topic, "it became clear" to the guy after a few conversations. Confidence inspiring, no doubt innovative, we should probably not be asking impertinent questions! No idea if the tech is sound but I immediately get scam vibes just from how quickly it leaped into name-dropping Musk and Altman.

Signup doesn't actually work for this thing, just an error that shows a Justin Bieber quote.