@m4ra Yeah, though at that level it's probably a consumer awareness issue too. You should want all of the stuff to be onboard if you really need it reliable. But for lots of things right now that's either unaffordable or even impossible. So legislatively you do have to ask how to ride the edge between protecting people and just forbidding the type of product from existing.

Interchangeable APIs is the best technical solution I've seen (AI Roguelike does this), but still hard to future-proof.

@jonmsterling If you're learning things to use them and not just to be able to say that you know them, then whatever your source of imperfect information you at least have the motive and methodology to eventually refine it.

The challenge from an instructor's point of view is then how to make it clear how the things they are teaching provide affordances for the student to do something they might be interested in doing (which the instructor does not have direct knowledge of).

@jonmsterling IMO you have to ensure that there's a full loop from information to synthesis to application in learning - when learning from instructors, when learning from trial and error, or when learning from an untrustworthy source (like an LLM, but also learning from other students).

Basically - when you go to apply the things and it doesn't work, that's how you filter out the BS.

If the teaching model never asks for that or provides tools for self-check, then mistakes can get locked in.

@lorenschmidt So one thing that makes existing structures unstable to growth is that the traffic needed between two locations that specialize should scale (at least) linearly with their populations which in turn goes as area.

Whereas the bandwidth of roads scales as their cross-section.

So as cities grow, old road architectures fundamentally can't serve them. Specialization may break down and be recreated locally when possible to alleviate the pressure. Etc.

@spiralganglion If you're willing to drop causality there's stuff. In statistical QM, if you allow time to be a complex number, the imaginary part corresponds to temperature (https://en.wikipedia.org/wiki/Imaginary_time). People also solve PDEs with this kind of thing - the keyword is 'kime' - but again it doesn't really behave causally with an 'arrow of time' anymore (see e.g. https://www.sciencedirect.com/science/article/pii/S2666818122000122).

Long story short, the boundary conditions are (part of) the problem.