---
I think most of the efficiency functions that these kinds of efforts tend to optimize for miss the most important factor.
I've thought about this a fair amount drawing on my experience switching to dvorak 15 or so years ago and not getting the typing speed increase I was hoping for, and also from having a close friend with a PhD in piano performance. There's one thing that I've never seen taken into account by these kind of keyboard layout optimization efforts that pianists understand is very important, and that is runs. Most people seem to focus on individual finger movements, a meaningful component to be sure. But I think there's still a lot that could be gained from optimizing a layout for runs. Runs are sequences where the fingers are depressed in sequential order across the hand (possibly skipping one or more fingers). They can be played (typed) not so much by isolating individual finger movements, but by getting the hand into the correct shape in a single movement and then actually depressing keys with a single rotation of the arm/hand/wrist.
The interesting thing about optimizing for runs is that it's directly opposed to the alternating hand optimization that the dvorak layout uses. I would love to see some exploration in this area and think it could yield meaningful speedups. Unfortunately, getting proficient with a layout takes a long enough time that I'm not super likely to take the project on myself.