4 ms·

I think a big part of the appeal of Dvorak in the dev community is that there's something very satisfying about a keyboard based on real research. That research

by wxs 13y ago

I think a big part of the appeal of Dvorak in the dev community is that there's something very satisfying about a keyboard based on real research. That research was done for 10 finger full-size keyboards though, and doesn't really map to the current generation of keyboards. As I allude to in that footnote, we've actually done some really cool research into optimal keyboard layouts in ambiguous scenarios. We'll probably do a follow up post diving into that! (also stay tuned, there should be a paper coming out soon).

jerf 13y ago
Yes, that's probably true. QWERTY is unlikely to be optimally suboptimal either. ... OK, OK, that was snark, I admit it, but it felt good. Mmmm... oh yeah, that's the stuff. More seriously, it is unlikely that QWERTY is optimal for this use case since it certainly was never designed for this.
- cbr 13y ago
  it is unlikely that QWERTY is optimal for this use case since it certainly was never designed for this QWERTY was designed to minimize jams on manual typewriters. You get a jam when you press two nearby keys in quick succession. This is actually extremely similar to the layout constraint swype operates under.
  - jerf 13y ago
    I was talking about Minuum, not Swype there. For Minuum, "maximizing alternation" is not a priori obviously the same as "maximizing the information theory distinction" between two subsequent keypresses. In fact, the fact that you are trying to maximize separation tends to imply that you are thereby tending to make the next keystroke more predictable than it should be, which is to say that the likely next keystrokes have clumps where too many of the most likely "next letters" are all on one side, which is suboptimal from the information theory perspective. QWERTY is not likely to be optimal for this use case; it may even turn out that the optimal keyboard resembles it as little as QWERTY resembles Dvorak.
OverlordXenu 13y ago
There's been research done on Dvorak? I remember a few years ago, when I was trying to learn it, I ended up looking into it and finding nothing but "this person can type faster on Dvorak than QWERTY!", ie. no actual studies with controls and all that, so I wound up giving up. Got any cites? (I'm not doubting you, I'm genuinely curious.)

by wxs 13y ago

jerf 13y ago

Yes, that's probably true. QWERTY is unlikely to be optimally suboptimal either. ... OK, OK, that was snark, I admit it, but it felt good. Mmmm... oh yeah, that's the stuff. More seriously, it is unlikely that QWERTY is optimal for this use case since it certainly was never designed for this.

cbr 13y ago
it is unlikely that QWERTY is optimal for this use case since it certainly was never designed for this QWERTY was designed to minimize jams on manual typewriters. You get a jam when you press two nearby keys in quick succession. This is actually extremely similar to the layout constraint swype operates under.
- jerf 13y ago
  I was talking about Minuum, not Swype there. For Minuum, "maximizing alternation" is not a priori obviously the same as "maximizing the information theory distinction" between two subsequent keypresses. In fact, the fact that you are trying to maximize separation tends to imply that you are thereby tending to make the next keystroke more predictable than it should be, which is to say that the likely next keystrokes have clumps where too many of the most likely "next letters" are all on one side, which is suboptimal from the information theory perspective. QWERTY is not likely to be optimal for this use case; it may even turn out that the optimal keyboard resembles it as little as QWERTY resembles Dvorak.

OverlordXenu 13y ago

There's been research done on Dvorak? I remember a few years ago, when I was trying to learn it, I ended up looking into it and finding nothing but "this person can type faster on Dvorak than QWERTY!", ie. no actual studies with controls and all that, so I wound up giving up. Got any cites? (I'm not doubting you, I'm genuinely curious.)