Do note that portability is basically a non-issue. If you expect to spend one-twentieth of your typing time on other computers (an insanely high estimate even for a backwards company’s “IT guy”) and we pessimistically assume that learning a new layout will magically reduce your old layout skills to hunt-and-peck[1], the gain in speed only has to be something like 5 wpm before it drowns out the lost speed on old layouts.
[1]: This is not supported by my experiences; two of my close friends have remapped to Dvorak and their QWERTY skills only mildly deteriorated. None of the concerns you have raised have been a problem for them; although one thing that did happen is it became impossible for any of us QWERTY users to type on their computers (something they quite enjoy).
we pessimistically assume that learning a new layout will magically reduce your old layout skills to hunt-and-peck
That basically happened to me. I’ve switched away from QWERTY in 2006 and really only ever use it if I fix something on my mother’s computer. Previously I was typing at 80-90wpm, now I don’t even know where the keys are anymore. I could maybe get back to 30wpm with a week of practice, but most of my QWERTY skill is gone.
I don’t spend 1/20th of my typing time on computers I don’t control, no. But I do think the time I spend on other computers is more likely to be particularly time critical, and I don’t expect to get any real speed gain from the transition as my work is generally more thought or compute bound than typing-bound (data entry might be sped up significantly, but I don’t spend 1/1000th of my time at the computer doing data entry).
[O]ne thing that did happen is it became impossible for any of us QWERTY users to type on their computers (something they quite enjoy).
They aren’t married, are they? My wife complains enough about my running ratpoison...
It doesn’t work like that. Typing does not take conscious thought—it is overlapped with my thinking. Speeding up my typing even tenfold would not significantly speed up my output.
Except when bound by something unrelated. My point is that the gains are minimal, and putting my optimization effort there is ridiculous from a speed perspective—that’s not how you do optimization, if you want to do it well. If there are actually large RSI gains to be had, that’s another matter.
Do note that portability is basically a non-issue. If you expect to spend one-twentieth of your typing time on other computers (an insanely high estimate even for a backwards company’s “IT guy”) and we pessimistically assume that learning a new layout will magically reduce your old layout skills to hunt-and-peck[1], the gain in speed only has to be something like 5 wpm before it drowns out the lost speed on old layouts.
[1]: This is not supported by my experiences; two of my close friends have remapped to Dvorak and their QWERTY skills only mildly deteriorated. None of the concerns you have raised have been a problem for them; although one thing that did happen is it became impossible for any of us QWERTY users to type on their computers (something they quite enjoy).
That basically happened to me. I’ve switched away from QWERTY in 2006 and really only ever use it if I fix something on my mother’s computer. Previously I was typing at 80-90wpm, now I don’t even know where the keys are anymore. I could maybe get back to 30wpm with a week of practice, but most of my QWERTY skill is gone.
I don’t spend 1/20th of my typing time on computers I don’t control, no. But I do think the time I spend on other computers is more likely to be particularly time critical, and I don’t expect to get any real speed gain from the transition as my work is generally more thought or compute bound than typing-bound (data entry might be sped up significantly, but I don’t spend 1/1000th of my time at the computer doing data entry).
They aren’t married, are they? My wife complains enough about my running ratpoison...
Sure, and learning a different layout would make it even more thought- / compute-bound.
It doesn’t work like that. Typing does not take conscious thought—it is overlapped with my thinking. Speeding up my typing even tenfold would not significantly speed up my output.
If your typing speed increased tenfold, you would go from “generally thought/compute bound” to “always thought/compute bound”.
Except when bound by something unrelated. My point is that the gains are minimal, and putting my optimization effort there is ridiculous from a speed perspective—that’s not how you do optimization, if you want to do it well. If there are actually large RSI gains to be had, that’s another matter.