I’ve worked on a lot of codebases and your “usually” is highly debatable. I don’t think the difference in understandability will be as big as people think as to

I’ve worked on a lot of codebases and your “usually” is highly debatable. I don’t think the difference in understandability will be as big as people think as tooling for ML evolves. Something like Godbolt for neural networks? Can’t be that far away