tbc I don't feel like I clearly agree with Muratori about what architectures are ideal in what settings; I just really love the deep dive into the history of PL design and the architecture of Sketchpad.

I definitely found his courses helpful in learning to get better at thinking about performance. It's one of the only places where I've seen a bottom-up approach to performance, i.e. making a Fermi estimate of how fast something *could* be and then trying to approach that, as opposed to starting with something existing and profiling it.