Section 1: Hello all. This is my first post to this board, but I've been lurking a while. It seems like this is the best place to be for people interested in keyboard optimizations.
A while back I finally got around to looking into Dvorak. I had heard that it wasn't proven, but that Qwerty was thought to be inferior. Since then, I've been convinced. My first reaction was to go with an alternative that would have some chance of being mainstream, so I found two options: Dvorak and Colemak. I figured that Dvorak would be much harder to learn and so I looked into Colemak for a bit. I installed it, gave it a few practices, but I didn't commit to it very much.
Basically I am a suspicious person when it comes to "optimal." I have no problem spending large amounts of time tinkering, but I don't want to learn a sub-optimal layout. I also am drawn to the sort of tasks involved in tinkering with data and the trial-and-error solutions of making a keyboard layout. So you could say I haven't even tried to learn Dvorak or Colemak, although I typed a few hours on each--I found myself distracted by the opportunity to try to make a "better" layout than either. Suffice it to say, I can type about 100 wpm with Qwerty and am happy with that, but I would like to become able to type in two layouts, the second being something theoretically ideal--a long term thing, so that perhaps some day I will be able to type faster and more easily in my own layout.
Section 2: However, there seem to be several "philosophical" questions about the task of making a "better" layout, as well as some practical ones. All the remarks below assume we are talking about making a keyboard for typing English, but that assumption alone is a difficult one to make (perhaps an additional question could be--ought we try to make a keyboard layout capable of multiple tasks, multiple languages, etc.? For instance, Colemak's decision to keep ZXCV in place is a decision to make the keyboard suitable for more than the main task; to sacrifice part of the main task in order to be useful for another. I will refer to this as question 0.)
1. Speed vs. Safety. Avoiding uneven loads, or loads on the wrong fingers, may be healthy, but it might not always be the "fastest" way to type. This is a broad question; sub-issues would include: Is there such a thing as "safe enough"? Otherwise, ought we optimize one first, the other second? Etc. Clearly, safety optimizations lend themselves towards some speed; however, is it okay to sacrifice some safety, so that the keyboard is "faster"? Should we be content being "as safe as Dvorak" or "significantly safer than Qwerty"; for instance, does Dvorak's pinky load imply that a heavy pinky load is "safe enough" as long as it doesn't exceed Dvorak's? etc. I believe that Dvorak perhaps tried harder to perfect safety than Colemak; Colemak is content to match Qwerty in some of these alleged safety dimensions, so lo ng as you can type fast with it; perhaps consequently, the anecdotal evidence in favor of Colemak's healthiness is not as strong as Dvorak's (perhaps in time it will be). But at the present some users say that Colemak hurts them, when this has not been an issue with Dvorak.
2. Training time vs. Effectiveness. Note that effectiveness isn't a decided issue; it depends on your answer to speed vs. safety. However, we must consider that a layout may be great, but very hard to learn (some say this is true of Dvorak). It is up to us again, to decide whether we should sacrifice some quality so that the required training time is reduced. I believe Colemak and Dvorak represent radically different answers to this question.
There are also practical questions that don't seem definitively answered:
3. What is the fastest keyboard? There is no simple answer; "it depends," we say. We say it depends on how much you weigh "finger rolls", "same finger", "row jumps", and the like. But this isn't really accurate. True speed should be calculable. How much time does a "same finger" cost? etc. An accurate time cost can be put on every problem, and then we can find what is the fastest keyboard. We might find that it hurts your right wrist, and then we will have to think about question 1. We might find that we can't copy-paste very easily or use other common shortcuts and have to refer to question 0. Or we might find it ridiculously hard to remember certain keys, or to find punctuation placed so irrationally and seemingly randomly, and perhaps refer to question 2.
4. What is the safest keyboard? We seem to only have speculations as to what is "helpful"; but this question is as answerable and yet unaswered as question 3. What is required to make a keyboard minimally dangerous? Perfect balance? Between fingers? Between hands? Some kind of symmetry? Minimizing combinations that move the hand or wrist? Until we know the requirements for safety, we can't successfully even think about mitigating safety with speed, or choosing one, etc.
5. There is probably some mystery as to what really adds or removes training time. Colemak seems to have taken an opinion in this regard and run with it, which is more than anyone else. It's a revolution, but we are far from being convinced we have the complete solution. Perhaps this is really two questions, or more. 5a. What keyboard will be easiest for a new typist to learn? 5b. What keyboard will a Qwerty user be able to learn "fast enough" while still gaining improvements over qwerty. Of course the philosphical question 2 remains: How much training time for how much gain? 5c. Could even be for Dvorak users, etc.
Sorry this post is going so long, but I just feel that I have no answers for any of these seemingly necessary questions; yet many people have made layouts, perhaps by guessing at some of these questions, ignoring others. I have done no less. But I wanted to profess my ignorance, before telling you what I did. I doubt my layout is optimal. However, I have just as much doubt for every other layout. Every "authority" seems to come to different conclusions and we have to guess which is closer to the truth, even in issues that should be matters of fact (as explained above).
Section 3. My own assumptions and experience: I think I have been heavily biased by DDvorak site's tool; I realize his algorithm for overall score is no doubt wrong, however, I have largely judged layouts on the individual metrics he offers (though I know they aren't ideal either): homerow jumps and finger repeats, I have looked to minimize, however I have also looked to keep a "reasonable" overall performance score (basically this means keeping "distance" on the pinkies very low), and decided to keep balance in terms of "distance" rather than "frequency."
If all homerow jumps were equal (they aren't), finger repeats were paramount (probably not), a "flow" of outside to inside need only be "acceptable" (questionable), and pounding frequency is acceptable so long as pounding distance is balanced (who knows?), then my layout is good, and I'll stick with it. But there are too many questions, as you can see.
"Flow", the way DDvorak scores it (not ideal), works on the obsevation that going from weaker fingers to strong across the same hand is ideal. No layout besides Dvorak seems to really accomplish this. Therefore I have been content to stick around where Qwerty and Colemak hover. Colemak actually scores worse than Qwerty in this; my layout is slightly better than both, but no where near Dvorak.
Dvorak, however, has a lot of right pinky movement. My layout corrects this.
As far as home row jumps and same fingers, I feel my layout is significantly lower than Dvorak and Colemak in this regard. The only other layout I've found to do this is the Capewell layout; Capewell, however, intentionally keeps you on the same hand for far longer than Qwerty or any layout, while mine switches hands, less than Dvorak but more than Colemak.
I don't really know if sharing my current layout even matters; above are the facts, and questions. What do you think? And what tools do you use to test layouts? Can a windows user run them and tweak them easily?
