To be fair, my understanding of these arguments is that they’re about deltas and not about raw numbers.

I don’t agree with them, but I don’t think it was the raw compute power as much as it was maintaining the _delta_ in compute power.

I'm mildly sad you didn't use 2,500 jiggaflops