There's no "perfect" design for a keyboard layout. Each person has different usage patterns, and different language combinations, different programming languages, different hand sizes, different touch typing techniques, etc.

The decision that many keys would stay in their QWERTY position allowed Colemak to become a reality, without that a final version of Colemak wouldn't exist today. Trying to design a "perfect" keyboard layout without that limitation, makes the quest for perfection a never-ending task. People who are trying to design the "perfect" keyboard layout would face the problem that each little tweak requires a completely new layout. For every aspect you try to optimize, other aspects will suffer, so designing a keyboard layout is always a matter of trade-offs.

Could Colemak be improved if there weren't any limitations on compatibility? I'm sure it could be, but I think that moving a dozen extra keys for a marginal improvement of efficiency is hardly worth it.

What Shai is trying to tell you is without the QWERTY constraint, you are left with too many good solutions. Not just that, you would have what I would call families of good solutions.   It wouldn't have to look anything like Colemak and there are people out there working on such things  (see this example http://www.almaden.ibm.com/u/zhai/paper … Short1.pdf   http://millikeys.sourceforge.net/freqanalysis.html )

Whatever statistical methods you end up using, you are going to find a few keys are the major contributers to the principle components of variance and then it drops off pretty rapidly into higher order noise.  Many lower frequency keys could be shifted around as shallow minima shift in the multidimensional statistical space as you change texts.  You could also imagine families of solutions as you reorder the home row key (take E as the highest frequency key, you have 4 good locations for it alone - middle finger (L or R) or index finger (L or R)).   You have quite a few decent variations of the home row with ETAOINHSRDLU and the many good variations for each home row sequence. 

It's only because Shai insisted on aiming to minimize the differerence between QWERTY AND insisting on the additional constraint of NOT moving in low frequency keys unless there was a very compelling reason (like the needs of high frequency keys, etc.). Only with the bar for movement set high could it be possible to come up with local shallow minima or in other words a single family of solutions of which Colemak's current layout was decided on.  Even then you could have a debate as evident on this forum elsewhere.

I think that if you are not comfortable with his statement about marginal improvement, then the burden is on you to go study the research literature on keyboard design such as it exists over the last century.  If you then decide that it's still a significant unanswered question, then you could offer him a very big donation to spend on all the time it would take to work on your proposal or write a proposal to get him funding from a research grant aimed at addressing the issue.

Why stop there ?  Lets really start from scratch.   I would like to see someone design a non-staggered laptop keyboard layout.

I think the point is what would the keyboard look like if you just considered schoolchildren learning to type.

I think the answer is take your pick. 

A kid can learn QWERTY to achieve in excess of 100 WPM.  A developing brain is highly adaptable.  The trade-offs between Colemak and a optimal layout with no regard for adults would have to be minimal for children considering how well they can adapt and overcome the dramatic positional drawbacks of QWERTY.

I believe that you're looking for one Michael Capewell: http://www.geocities.com/smozoma/projec … 1#Capewell

Unfortunately, he never got around to including Colemak in his comparisons. The Capewell project may be illustrative of Shai's point actually: After making a handful of evolved layouts and from the looks of it putting down some solid hours in the process, it seems he just up and left it all. I guess he got tired of chasing the elusive "holy grail of keyboard layouts" and settled for good enough then lost interest and stamina.

I have seen newer versions of the Maxwell applet including both Colemak and at least one of the "best-so-far" Capewell layouts; where was it, again? Somewhere on these forums - have to look for it.

At the very least, Capewell's page will give you an insight into the complexities involved in generating optimisation parameters. Oh btw: He does keep ZXCV in place. If you're hellbent to move those as well (although I think V would be the only serious candidate as the other three are pretty perfectly placed where they are!) there have been attempts at that too lately. Or you could do the research yourself.

You could also have a look at Amuseum's efforts that he linked to in another thread on this forum: http://sagez.net/amuseum/layout.htm

I agree, KBS. The first thing I thought when reading this thread was "What? Now that Shai has done so much work, why should he start tedious research again for the sake of comparing cmk to another minor (or the perfect) layout?" I can really understand if he's got fed up of Cmk at the moment. There are many things I would like to ask him, but I'm trying to bite my tongue and look for myself or wait (wasn't a 100% successful though, hehe xD). For example, I'm dying to see the new statistics, but I'll just wait for the end of this year when shai will probably give us new ones. I know this sounds like I'm trying to be Shai's little helper or so, but I'm not! I'm just happy that he has actually given up hours and hours of his free time for such a good layout (look at how much thought he's put into it: the 'y' key is directly above 'i', as those hardly come up consecutively in a word, rathere than putting in a letter like 'n', which does indeed very often follow or preceed the 'i') and I've noticed that shai seems to want a *temporary* break from colemak so we should grant him that. I guess, after all he's just a human. (defenitely not a god [yet]) =P (@ Shai: That was just my impression of you and I'm not trying to say that you want to abandon cmk or anything.)

During the first year of Colemak's existence, I have actively tried to improve upon Colemak. I must say at this point that I've given up on trying to improve Colemak, and I probably won't be doing any more research of new keyboard layouts. There won't be a "Colemak 2.0".

I'm not definitely not abandoning Colemak, things are just slowing down. Currently I'm working on making a user-friendly Linux installer, and after that I'll be releasing more information about the design of Colemak.

ok kids,  no bitch fights!

Beyond liking the layout, I am liking Shai's approach --

phase 1: a lot of work to come up with a FINAL layout that makes a lot of sense from many angles from general use (various writers), to specialized use (programmers), to multilingual use.

phase 2: make available on as many platforms as possible along tutors and games to learn, build a user base by putting it out there, having a website, winning awards, publishing stats and reasons to exist, active forum, etc.

phase 3: working more user friendly installers, build user base

phase 4: more information about the design of Colemak, build user base

phase 5?

a lot of other layouts seem to be stuck perpetually in phase 1 and will never be anything more than perpetual research projects or abandoned projects.

What makes Colemak important besides the various good reasons for being a serious alternative to Dvorak for those wanting to move away from Qwerty, is that it has actually moved through phase 2 which makes it a viable alternative standard. I am not an early adopter, if it had not moved into phase 2, I would have ignored it and continued to learn Dvorak-Qwerty hybrid on my Macbook because it was easily available.  I am not interested in the next best layout.  I am interested in a layout clearly superior to Qwerty that gets me productive yesterday, easy to install, moves only what makes sense unlike Dvorak, doesn't mess with hotkey combinations more than what is tolerable, and gets me to 60+ WPM without stressing out my 40-something hands and wrists.

Actually, samurai: The Colemak per se isn't so good for programmers I think? The point being that it doesn't improve on the punctuation placements for reasons of consistency so that the average user will have an easier time of it (and international conversion gets easier too since these layouts often have differing punctuation placements). Of course, one major point of Colemak is that being focused on the letter block only it allows the tech savvy users to do whatever they want with the rest! You can remap the letters to Colemak and the rest to something that suits your programming habits if you wish, still call yourself a Colemak user and never look back. This means less support from the official installs and such, so it's for the tech savvy - but programmers are just that. No problem. And I think that a "punctuation-improved" Colemak for programmers could get a place on these forums, for the specially interested. It might not be so easy though since it'd depend on what language(s) you're using. For the most part I suspect it'll remain a largely private matter.

Kaddar: Valid assessments as far as I can see. But keep in mind that people here haven't to my knowledge claimed that it is proven that nothing but marginal improvements can be made! What we've said is that we consider it somewhat unlikely that it's readily doable given the amount of work people have already put into this (although a few of these other attempts may or may not meet your criteria better than Colemak), that proving it thoroughly will most likely be a LOT of work which we can't take upon us to do since we're not employed and paid to do so, and that we consider the other benefits of keeping the pertinent rare keys in place larger than the potential disadvantages.

I will again stress the point of ease and speed of learning. Getting that speed up much faster than with Dvorak is important, regardless of hotkeys. If thousands of people all over the world are to spend time learning something better than QWERTY, the workhours involved quickly add up.

More importantly though, it's about winning people over which making an "elitist" layout (like Dvorak which moves around pretty much everything - including the numbers in the first version!) simply won't succeed with! I consider the history of Dvorak ample indication that this must be the case. Maybe - just maybe - Colemak can fare better than Dvorak over time, not by being marginally faster or more comfortable but by being more acceptable to more people! That would be what I'm hoping for at least. Whether my typing speed at the end will be 100 WPM or 105 WPM (if I'm lucky! - never could get my QWERTY or Dvorak above 60 so I'm wishing here) bothers me minimally.

You didn't ask for such an "elitist" layout to be promoted of course. But I still mention these points to illustrate why there is so little enthusiasm for the project here. If we had a lot of time and resources, yes of course; but we don't so there you have it I'm afraid.  ;)