We know that Sean Wrona, current fastest typer, uses custom digraphs and trigraphs to type ludicrously fast. It's impossible to make a perfectly optimized evolved layout that doesn't take these into account, because they directly affect how good a letter position is.
Are there any that try to do so?
I'm going to eventually design a program to try to do this, as a personal project, but it's a ways off as I'm still learning to program.
We are patient. You can take all the time you need.
My fastest typing records http://hi-games.net/profile/4314
My switching experience https://forum.colemak.com/viewtopic.php?id=970
Er... what exactly do you mean with alternate finger digraphs? Are we talking about putting frequent digraphs just on different fingers (as measured by same-finger-ratio), Keys one finger apart such as ring/pointer, same finger but alternate hand, having digraphs or trigraphs on the same hand but in consecutive order (rolls), or putting digraphs on directly adjacent keys (qwerty er, io)?
Since I'm not sure exactly what you mean, I can only direct you to the standard sites, Carpalx and The Keyboard Layout Project.
Cheap Test Drives: Footswitch, Maltron, Vertical Keyboards and The Wide Mod
I mean hitting a certain key with a finger that wouldn't normally hit it if you typed strictly by the books
It lets you type faster, so current evolved scoring models that don't involve this are faulted
Current evolved scoring models have a lot of problems, including that one. ;)
*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***
Oh, I see what you mean now. Interesting. I like it!
Of course, these keyboard analysis programs tend to be based on what is accepted as proper touch typing, standard fingering and whatnot, so as to have the most benefit to the average user. While what Wrona does is fast, it probably takes a significantly longer to learn, and might not be ergonomically advisable. Also, I'm not sure how you would score the effort models. I know The Keyboard Layout Project was once set to optimize for rolls above all else. That was a pretty interesting layout.
Speaking of the failing of layout analysis programs, I would like to see one that has more logical finger movement. For instance, when I type 'there' I do not return my fingers to the home position until the word is done. Most layout analyzers assume I move up for the T then return to home, instead of simply moving to the R. Likewise, there is no need to move the middle finger from D to E then BACK to D, only to go up to E again. Maybe if it could be set to have the fingers rest at the last key pressed until space is pressed?
Also, for the record, PatorJK's Keyboard Layout Analyser now does alternate fingering AND alternate home positions. Very cool.
Edit: while I'm dreaming anyways, I'd also like to see a layout analyser/evolver that takes three or even four consecutive keystrokes into account, not the usual two. Obviously fast typing is limited by finger travel speed. This is where same-finger (or whatever we are calling it these days) comes into play. The thing is, if you are typing fast enough, one keystroke's time is not enough to move between keys.
Example, qwerty 'teach'. Not only does this have a one keystroke gap for the middle finger to move, it is also a row jump, but with that one keystroke on the middle row, it would not be counted by any analyzer that I am aware of. On the other hand, if you are like me, and type the qwerty c with the pointer, then it becomes a two keystroke gap for the same-finger row-jump. Much more comfortable.
Edit 2: Thinking about it a bit more, maybe a finger distance per hand alternation? Or finger distance over time as measured in keystrokes?
Cheap Test Drives: Footswitch, Maltron, Vertical Keyboards and The Wide Mod
"I mean hitting a certain key with a finger that wouldn't normally hit it if you typed strictly by the books"
Do you have a real exemple because I'm not sure to see what you mean. What could be the ratio of hit done with this technic?
The maps of distance and accessibily are changing every time you have to type a di tri quadri gram?
Implicitly, are you saying that alternation of hand, and the home keys are not so efficient, after all?
@cevgar: It's hard to show and understand at least 24576 possibilities in a 2D table. (3!*4³ *4³ if i right)
To see experimentaly the impact of trigrams, I took a text, erased space, determined a list of trigrams and the ratio of use, erased the letter in the middle of trigrams, and analyse the digrams found with classical tools.
If your hand alternation is good, I think it could be a way to increase "rolls" in the same line and decrease the same finger. What I have done with my layout Bvofrak EN V0.5.
But why not to augment rolls with no conventianal fingers if you agree, of course, that alternation of hand could be efficient to type.
