[htfu]

2.5 _days_, not years.

A still fuzzy and arbitrary but imo somewhat elegant cutoff would let feasibility be set by the market and technological development, in the sense of something not being worth throwing compute at _now_ because the timescale of work is slow enough that future classical efficiencies will progress (much) quicker.

[AlexCoventry]

For market forces to be meaningful, the calculation would need to provide value to someone. At the moment the desired output is simply "a distribution which is hard to simulate classically," so assessment in terms of market value would be very premature.

[htfu]

But that goes for any raw benchmarking, it’s busywork by design.

[AlexCoventry]

I'm pretty cynical about this specific experiment, but that's taking it to another level.

The ROI was terrible, but it wasn't mere busywork when Galileo was mucking around with balls and ramps.

[htfu]

Haha sorry should've been much clearer, I meant specifically computer power benchmarking.

You perform busywork, if for marketing purposes usually of the kind or at least set up in a way favorable to your product, and on the presumption that it will in some manner transfer to real work.

The more synthetic the less likely that is, but it's still indicative of something and I don't see what you mean is so different here.