Huh, I just realized that PCB layout is a graph optimization problem with a bunch of slightly odd constraints.
This is simultaneously the sort of problem that a computer should be able to solve perfectly at small scales, and one that is difficult enough that humans should be able to locally outperform a computer on large scales... Hmmm.
I'm sure computer engineers realized this in like, the 50's...
@starkatt @icefox also unmanageable
often, when doing a design, we'll have to do multiple spins... add some caps here, drop a trace there, replace a WTF88271 clock skew generator with a WTF88274 but still want to possibly swap in the '271 later in production so there's some 0-ohms and DNPs for that...
but we've already validated the high-speed parallel buses running across much of the board, and moving those would be risky...
oh also it's got digital and RF stuff on the same board, have fun