[somewhereoutth]

I'd suggest that a measure like 'density[1]/parameter' as you put it will asymptotically rise to a hard theoretical limit (that probably isn't much higher than what we have already). So quite unlike Moore's Law.