BrtByte

I would not count on the market alone. Like customers do not always choose the technically best or cheapest option

Yeah, I suspect the $200K check would be the cheapest line item in the whole thing

I would just add a reproducibility requirement and avoid keeping the exact same benchmarks for too long

The library analogy works (for me), but the uncomfortable part is that most "open" models are closer to receiving a compiled binary than receiving the library

Funny how the idea only seems to become practical once you have a literal warehouse problem worth of old RAM

The funny part is that a 1000x increase in RAM somehow doesn't make a modern computer feel 1000x more luxurious

The impressive part is not really reusing old RAM, its making the economics work despite the extra chip, software support and operational complexity

I think this is the right distinction

A med student can absolutely contribute useful work, especially with good supervision. The issue is more that inexperienced authors plus publication pressure plus easy tooling is a bad combination

Feels like the minimum standard should be sharing the exact query/design choices and being very explicit about what biases the analysis can and cannot address

That is excellent playground wisdom

There's something very of-the-era about fixing a virus-ridden PC and then "improving" it by installing Prince of Persia

I don't think it was only because you were a kid. The game really did punish hesitation in a way that feels pretty unusual now

A lot of games have time limits, but Prince of Persia made it feel less like an arcade score mechanic and more like part of the story

Prince of Persia is one of those games where the technical limitations are almost inseparable from the magic

Interesting idea. It would basically turn the App Store version into both a discoverability channel and a license anchor for the direct version

The frustrating part is less that Apple has a boundary here, and more that the boundary seems opaque and inconsistently enforced

People will definitely look for alternatives, but that doesn't make regulation pointless

In a way it makes the Temu problem more frustrating

But I still think chargers and children's toys are exactly where the line should be drawn

The fine seems less interesting than the compliance deadline

In an ideal world, the artist's own mailing list or fan community would be the canonical place for this, because it's closer to a direct relationship

A real fan who mostly listens on Bandcamp, buys vinyl, or discovered the artist through live shows may look less "real" than someone who passively streams them every day

It feels like we're replacing one opaque system with another slightly more personalized opaque system

If the second-best lamp is 90% as good and 10x cheaper, most people will use the second-best lamp...

This is a good point, especially the "model is the data" framing

Open weights undercut the absolute cutoff scenario. They don't fully solve the question of who gets the best model first, who gets enough tokens to use it heavily, and who gets to integrate it into sensitive workflows without waiting for permission

I agree the genie is out of the bottle technologically. I'm less convinced that means access stops being politically and economically important. The bottle may be gone but the best lamps are still expensive

The uncomfortable implication is that "AI sovereignty" may end up being less about training your own GPT-class model and more about securing compute, energy, datacenter security and contractual access

In theory that sounds nice, but I suspect it would be much harder to make work in practice than it seems