Hi Shai,

Thanks for your interest in my suggestion. One correction, before I discuss some of your criticisms of my suggestion.

The correction is to what I suggested for one template; the second template should have the qwerty "F" key left blank.

With all due respect, I think the reason that most of these alternative keyboard layouts don't catch on is that the steps are too large.  Despite the obvious superiority of many, many, layouts (of which Colemak is only one alternative), NONE have every caught on.  I believe it's because (as has been pointed out here) that too many keys are moved on some iterations for minimal improvement (something the Colemak is designed to counter, yes) AND because the leap is still too large.

With the exception of the keys held in a "dummy position", once the Colemak key "moves in", it never moves after that.  And it actually only takes four different boards, since the final one (the fifth iteration) is your final Coleman board.

Colemak - despite some advantages - is STILL a large SINGLE leap for many people. Enough to be problematic, I think.

Furthermore, this solution is designed to avoid the whole "typing tutorial" nightmare - it might be hard to believe, but many people hate that method of learning anything, and I suspect most qwerties didn't learn their own typing that way. This solution is designed that people can use it on the job, immediately.

As to trying to ensure people can switch back to qwerty - why would you even want that?  Let Colemak rule the world, and leave Qwerty to the luddities.

Anyway, I'm not going to defend the idea any longer. I think it has great merit based on my own personal experience and knowledge of learning styles. So, I'm simply going to try it and report out later.

Thanks for your feedback - it's always helpful to improve an idea or product.

Jay Walker
The Confused Capitalist

Hi DreymaR,

I agree the dummy concept is awkward, but I couldn't seem to make it work (keeping a limited number of keys moving) without it. According to the scottbyrce site on cryptograms, "J" is the third least used letter in English (0.16%, barely ahead of q and z).

I'm not sure about moving more than four or so keys at once (or less than four intermediate boards); having taught kids with learning disabilities I like the idea of mastery of small steps, rather than fumbling around with larger steps for longer. Experience suggests to me that, overall, movement is more rapid when steps are smaller, even if there are a larger number of them. Plus there's the psychological thing that people know they are moving forward - this isn't to be underestimated either. So I have some resistance to the idea of more than four letters. Even then, for example, template two actually moves five letters, if you count 'J' moving into the dummy position. But I suppose that for some people, fewer templates would work better.

You could, for instance, combine the GT move (template Three) into either of the adjacent templates, and that would reduce the steps by one, and the number of interim templates to just three.

By my calculation, using the cryptogram numbers on Scott Bryce site (scottbryce.com), the Colemak moves 76.03% of the English letter usage (frequencies of letters in the English language). Using the suggested template design,  the steps break down as follows: Template One - moves 14.11%, Two - moves 22.57%, Three - 11.21%, Four 13.48%, Five (completion to the Colemak template) - 14.66%.

So Two is really the monster move, because it moves both E [12.51%], and N [7.09%]. Otherwise, most of the moves are relatively balanced, because they generally move one letter in the 6-9% range, plus some other lesser lights.

Your thoughts about 'E' are interesting and, I think, worthy of further consideration. You could of course start off with template Two and continue through to Five, with One becoming the completion template. The idea has some pluses and some minuses.  On the plus side, you get roughly 1/3 of the letter frequencies movement happening on the first template [22.57%/76.03%]. On the minus side, you get roughly 1/3 of the letter frequency movement happening on the very first template! (plus letter 'j' moved into a temporary position).

Generally, I like your 'E' idea, but don't think it quite works with the templates I created, the first step needs to be smaller, I think - not five letters and not 22%. But moving 'E' on the first template is probably an idea that should be explored more; but giving people an immediate taste of success is a design element that needs to be remembered - don't overwhelm them, especially at first.

Anyway, thanks for your thoughts ...

Jay Walker
The Confused Capitalist

Hmmm, interesting as mentioned. I see the point of starting with U>I>L>u (where majuscle denotes a key that moves to the next position in the chain, and minuscle denotes a now vacant QWERTY position). I'm still unsure about the optimal way to change and I guess we're both interested in seeing some trials.  :)

I suppose the tilde should be put on an AltGr combo while its space is occupied? There are some apps that quarrel with this, particularly the Typing of the Dead typing game I've used much while learning. If this is a problem, I guess the tilde could always reside temporarily in whichever QWERTY position is open at the time. It needs to be somewhere for the occasions - albeit rare for many users - it is needed.

I made a few attempts at reducing the number of intermediate steps in a to my eyes acceptable way. Using my notation, first your method for direct comparison (let me know if I made any mistakes):

The Walker 5-step method:
1) U>I>L>u
2) E>K>N>J>~(temp)
3) G>T>F>e
4) D>g   J>Y>O>;>~(temp)
5) ;>P>R>S>d

4-step Walker-esque:
1) U>I>L>u
2) E>K>N>J>~(temp)>e
3) D>G>T>F>e  (~>d)
4) J>Y>O>;>P>R>S>d

Again, it starts gently with the "perfect" loop, then the important E as far as it has to go to hit J which is rare enough to be messed with (if you want really short loops you could've stopped at K!), then one more step before rounding up the rest.

The idea would be that after a gentle start, a very effective 2nd step and a quite effective 3rd step you'll be starting to get the hang of things and also impatient to make "fourth base" and score the "home row run" - if you'll excuse the metaphor.  :)

Rounding up with a little bunch of mostly rare keys (and some common ones that at least only move a short way!) sounds acceptable to me. The "4-step" has the benefit of only using one temp placement, for the J (which I've come to accept as a good compromise). I feel that two temp placements is a lot.

An alternative:
U>I>L>u
E>K>N>J>Y>O>;>e(temp)
T>F>e   (;>~)
;>P>R>S>D>G>t

I'm trying to give only the semicolon the "temp treatment". It's a punctuation sign and as such my mind views it differently. Steps 2) and 3) could be collapsed into one big important step, but then I feel it'd get too long and it'd be better to use the J for a temp key after all (keeping in mind that the step with the E in it constitutes a lot of usage):

1) U>I>L>u
2) T>F>E>K>N>J>~(temp)>t
3) J>Y>O>;>P>R>S>D>G>t

That's as compact as it gets I guess, and the main variation would be whether you use J or semicolon as temp which blows up either step 2) or 3) (I still think that the last step is best suited for a little extra weight). As I said, I'm not sure which is more important: Your wish to do small steps, or Shai's (and mine, mostly) wish to not have a lot of intermediate steps and several temp placements. Time may show, hmmm?

Either way: Best of luck trying it out!