Analog and "non-exact" computation is not something that sounds very useful. The only reason we've been able to minimise so much is the tremendous noise rejection of binary systems.
Besides, modern computers are spending most of their time on actions of marginal or negative utility, mostly data-bureaucratic rather than computational.
> Analog and "non-exact" computation is not something that sounds very useful.
Floating point operations are non exact in general. Physics simulations work at various degrees of approximation. Non exact computations would be useful as long as we can quantify the "non-exactness".
Besides, modern computers are spending most of their time on actions of marginal or negative utility, mostly data-bureaucratic rather than computational.