> It's a shame the cost of learning a new layout is so high
Haven't you mean the cost of _creation_ a new layout? Learning is super easy for any non-QWERTY person because since you have managed to realize that default layout is bad - you have all freedoms possible to fix this obstruction. Sometimes the number of freedoms is too high, that's why it is so hard to create a layout which will be really used by anybody. What about learning a new layout per se - the cost of getting rid of QWERTY in my opinion is doubling for every year of touchtyping QWERTY or QWERTY-like layouts. Layout is something that you will be carrying in all your muscles for the rest of your life but after your first switch you will not consider switching layout that hard especially if you really love your new one.
> Another issue is the dataset used for evaluation here.
A beauty of a layout is kind of a beauty of a girl - it must be measured by heart, not by numbers. Layout is something that you will be carrying in all your muscles for the rest of your life. I do not respect any attempts to fix QWERTY (like Colemak and Workman) because QWERTY is so broken that there are no reasons to fix it (coining more and more of fancy metrics usually) except if QWERTY sits too deep in your muscles. And there is not much difference in using three rows of symbols when somebody either programs or writes a letter to his lovely ones.
> You make a good point about finger strength, not something I had considered.
Maybe yes but maybe I haven't disclose my opinion fully. I used to touchtype Dvorak way before some musical instruments gave additional strength for my fingers and I do not remember any desire to rearrange any letters.
There are lots researches going on for enhancing QWERTY, based on rearranging as few keys as possible. They usually are based on: the datasets, somebody's opinions about finger strength, some ideas of alternation of fingers or hands, now AI has joined to the army of fancy metrics. I do not respect any attempts to fix QWERTY (like Colemak and Workman) because QWERTY is so broken that there is no good reason to fix it, except of for users who have QWERTY sitting too deep in their muscle level.
The Halmak's idea of punctuation in the middle of the board is really interesting, I haven't see this idea on any layout I have tried. But my message is that layouts must be created not by complicated calculation's call but by heart's call.
Haven't you mean the cost of _creation_ a new layout? Learning is super easy for any non-QWERTY person because since you have managed to realize that default layout is bad - you have all freedoms possible to fix this obstruction. Sometimes the number of freedoms is too high, that's why it is so hard to create a layout which will be really used by anybody. What about learning a new layout per se - the cost of getting rid of QWERTY in my opinion is doubling for every year of touchtyping QWERTY or QWERTY-like layouts. Layout is something that you will be carrying in all your muscles for the rest of your life but after your first switch you will not consider switching layout that hard especially if you really love your new one.
> Another issue is the dataset used for evaluation here.
A beauty of a layout is kind of a beauty of a girl - it must be measured by heart, not by numbers. Layout is something that you will be carrying in all your muscles for the rest of your life. I do not respect any attempts to fix QWERTY (like Colemak and Workman) because QWERTY is so broken that there are no reasons to fix it (coining more and more of fancy metrics usually) except if QWERTY sits too deep in your muscles. And there is not much difference in using three rows of symbols when somebody either programs or writes a letter to his lovely ones.
> You make a good point about finger strength, not something I had considered.
Maybe yes but maybe I haven't disclose my opinion fully. I used to touchtype Dvorak way before some musical instruments gave additional strength for my fingers and I do not remember any desire to rearrange any letters.
There are lots researches going on for enhancing QWERTY, based on rearranging as few keys as possible. They usually are based on: the datasets, somebody's opinions about finger strength, some ideas of alternation of fingers or hands, now AI has joined to the army of fancy metrics. I do not respect any attempts to fix QWERTY (like Colemak and Workman) because QWERTY is so broken that there is no good reason to fix it, except of for users who have QWERTY sitting too deep in their muscle level.
The Halmak's idea of punctuation in the middle of the board is really interesting, I haven't see this idea on any layout I have tried. But my message is that layouts must be created not by complicated calculation's call but by heart's call.