I think it changes the nature of this technique from a cheap simple trick – just one low-precision matmul – to a less general, many times more expensive, more c

I think it changes the nature of this technique from a cheap simple trick – just one low-precision matmul – to a less general, many times more expensive, more complex computation. To me that starts to feel wasteful, even if GPUs can churn through that anyway. If you're merely implementing a color tweak in a game you can come up with plenty of simpler formulas that look nice. I've written oklab and okhsl implementations — that cube amplifies errors greatly. The full-precision cbrt makes it definitely a one-way thing for real-time graphics.