The quantifiable improvement and cost benefit is also important. Dvorak shows on average a 1-2% benefit to typing speed, which is basically not worth the time and effort lost to learning a new layout.
Of course, now that I present that argument, I realize it could be applied to text editors shortcuts and plugins, which I have customized extensively on all my main development machines. That said, I almost never do any significant coding on random machines, so the cost/benefit analysis might work out in my favor (I certainly behave as if it does). I also remap all my Caps Lock physical keys to Control/Command, which is pretty common among programmers but very rare among non-programmers. That’s another case where my gut says that the small rare inconveniences of using random computers are outweighed by my regular productivity increases.
However, I don’t have any actual data on this, so it turns out my argument in the first paragraph of this comment is really just a guess. I do intuitively feel like QWERTY to Dvorak is a much more significant alteration than Caps Lock to Ctrl or some customized hotkeys in my text editor of choice.
I’m not a particularly fast keyboardist, and I don’t particularly care about the speed differences of dvorak vs. qwerty. But, I do quite appreciate what I believe to be real ergonomic benefits.
My wife switched a few years ago and claims also improved wrist comfort.
But nobody would ever choose a dvorak keymap if they were solely interested in that 1-2% speed benefit (I don’t know where this number came from, but I no reason to doubt it).