While everyone else is wasting time and money on the slower, more expensive models, you've found a way to outpace everyone for less money. Everyone else is wrong and you will get rich.

(I don't actually believe the premise is true, I'm just pointing out the logical conclusion to what you're saying so maybe we can reconsider the premise)

This wasn't obvious a year ago, but today CAD literally reduces to Simon Wilson's pelican test, since CAD is largely a matter of functional CSG, and CSG is really not that different from SVG. It's just one more dimension, which it turns out is not a problem.

LLMs consistently one-shot CSG based video game levels with interesting physics puzzles (citing myself). Given this I'm willing to conclude that the frontier models are good at automated CAD if given the correct harness. But I guess a lot of people don't know this yet.

[1] https://github.com/elalish/manifold

It's like when a Youtuber's audience complains about how they're constantly asking you to subscribe. The reason it happens is because the statistics say it works.

The only thing keeping it afloat is the lie that it's social.

AI is really good at making things that look like they work.

This is a steelman of your argument.

I don't understand how to tell the difference, other than what society and the media collectively judges to be genius versus disordered, and where the money accrues.

The people and their personalities are not all that different, other than how palatable/sellable they are.

That's not to say this isn't a real problem; I think the scale of it is much larger than people realize, because society often sees these same personality disorders as something to be rewarded.

I guess this adds indexing and querying but most coding agents have good solutions for this already, and it works automagically for everything, not just memories.

What we could use instead is a file system layout standard, which could subsume memories and a lot more. I don't think that's needed either, but it would probably solve more problems than this.

WSL singlehandedly stemmed much of tide of developers moving away from Windows, but WSL native filesystem performance gave devs that magical experience when they boot into Linux the first time and see that the filesystem doesn't have to be ass. There's always been hacks around this, but for many devs the easiest hack was to ditch Windows.

They should have moved heaven to fix this on day one, there's really no engineering excuse. Linux is open source.

When that happens, will it still be impressive/spontaneous? Will we know the difference?

https://github.com/sparkjsdev/spark

If you mean the technique of splatting specifically, Dreams for PS4 [1] is prior art.

If you mean pre-rendering, there's Myst and games like the original FF7 for PS1.

[1] https://en.wikipedia.org/wiki/Dreams_(video_game)

That's kind of against the usual notion of "open source" but it's the only way this would work in e.g. the game industry, as currently factored.

Studios won't pay people millions of dollars to make games if the return on investment is zero other than helping all of your non-open source competitors.

I do think it's doable, but nobody's done this successfully yet.

Just like the religious leaders of olde, there will be many attempts to rationalize the AI God in the science.

  - many funds owned by the public will buy this, so people will be indirectly invested and could lose money
  - if this affects the economy, it will affect everyone

For a stable simulation you want your initial conditions to be stable, that's what this model optimizes for. But stability tends to be highly dependent on your settings, timestep, etc. Mostly because physics engines are designed for speed, not perfect accuracy.

I've only tackled this problem with shitty low-stakes (AI generated games) but I found heuristic post-optimization to be "good enough" -- essentially just stabilizing the simulation manually during scene bake. And predictably I had to rejigger the v-hacd and constants when I switched physics engines.

So the real usefulness of something like this is how well it generalizes to the heterogeneous simulation ecosystem, but the paper doesn't seem to cover that.

You could have no interruptions, no consent popups, and no tracking, just don't have ads.

Trying to find logic or middle ground on this issue will never make sense, because the ad cartel is based on absurd fiction: that ads are good for people, rather than just being stochastic predation on the fraction who are not successfully able to turn off the malware.

So we arrive at the absurdity that the onus is on the users to defend themselves against the product, and how to present the options to them, so that a sufficient proportion of the prey doesn't do it.

That Stallman's statements were seen as a "red flag", while Bill Gates's dealings with Jeffrey Epstein were seen as perfectly ok.