Amuseum's model might deserve a new thread of its own.  As for the chart itself, I can't say I completely agree with it.  For me, at least, the middle and ring fingers already hover between the middle and top row, making those positions relatively easy to press; IMO, even easier than QWERTY V.

In one of my Widely Alternating experiments, I did try putting D on the QWERTY V position in accordance to Workman-like models.  I ended up swtiching it back to QWERTY G, however, feeling like it was causing too much... I'm not sure, twisting?

The values are relative. I'm not sure adding a base difficulty makes a difference on reading the chart. But if I were to write a program to analyze layouts, then I would start to calculate more exact values based on various criteria, such as can be found at http://andong.azurewebsites.net/dvorak/ … iency.aspx . Another factor to take into consideration is inter-key combinations and penalties (but probably too complicated to fit on a single chart).

The pinkies are easy to tire out and cumbersome during rolls. OTOH, the ring fingers are stronger and closer to the middle of the hand, thus have better support and quicker during rolls. But since the home keys don't require reaching, the difference isn't that noticeable for single presses. Moreover in practice, there are other more important factors concerning these two keys (such same key presses with top and bottom rows).

I'm confused about the BK vs FU comment. Maybe you misread the chart? The BK keys in the middle columns require stretching the index fingers, thus the higher effort score.

Stretching the pinkies is very different than stretching the index finger. For one, the index finger is longer and stronger. Also the arms simply go in and out to reach the keys, whereas the pinkies requires more twisting of the wrist. However, these motions may be different depending on the keyboard you use. I based the chart on mainly the Kinesis where keys are aligned in columns and the hands are spread far apart from each other. On a standard keyboard, the hands are together and the arms bend towards each other. So the pinkies have easier access to the top row. Nevertheless, many Dvorak typists have complained about the placement of the L in such an awkward position at the top pinky.

Obviously when your hand is moved farther away from the bottom row, the bottom keys become harder to reach. The standard keyboard design does encourage hovering between the home and top rows. However, I don't think that is the most natural or comfortable pose for the arms and fingers. On the Kinesis, my fingers and arms are extremely relaxed; I never swing or twist my arms too much or use much effort to reach with my fingers. I don't have to hover towards top or bottom row; both rows are equally accessible and neither are preferred over another except by the fingers' natural inclination.

Each finger has different lengths and curvature. Just hang your fingers in the air in front of you. You can tell that the middle and ring fingers lean away from you, whereas the index fingers and pinkies lean closer to you. Now if you compare my chart, you will see a similar pattern where the middle and ring fingers favor the top row, while the index and pinkies favor the bottom rows.

As for punctuations, using the knowledge I have now, I would place them at the corners or middle columns, accessed by the pinkies and index fingers. On my chart, they would be placed on the keys with highest difficulties (2.5 and 3.0, the reddish cells.) Period and comma would go on the top pinky keys (QP on Qwerty, Q; on Colemak, 'L on Dvorak) close to the number keys to facilitate typing arithmetic and monetary values and IP addresses. Then apostrophe and double quotes on separate keys, both unshifted. With these four unwanted corner keys taken care of, proceed with optimizing the other 26 keys for the 26 letters.

Arensito has a nice symmetry and intuitiveness at least on the home row. It goes VCVCC from out to in for both hands. Likewise, my layouts tend to go CVVCC from out to in. This symmetry and balance is aesthetically pleasing.

I don't see why QWERTY has to be favored, especially when you're trying to wean away from it. Another popular camp is Dvorak style: all vowels on one hand, punctuations moved to the top. It really depends on your usage. On your own computer at home or work, you can set up your favorite layout without having to resort back to QWERTY. If you say it makes it easier and faster to transition out of, then again it really depends on how much you still use QWERTY for fast typing. I'm an IT guy who helps others with their computer troubles. Of course they still use QWERTY, but I don't need to type fast when I'm at their computers. I can still look down at the keyboard to remind me where the keys are.

These alternative layouts are for the most part purely theory-crafting on the interaction between humans and keyboards. There would be little progress if we keep trying to tweak QWERTY, instead of starting from scratch and working towards a goal. Dvorak's goal is hand alternation and inward rolls, and the result looks nothing like QWERTY. Colemak's goal is to be like QWERTY but improve on home row, among other things. Workman improves on that using a strain/difficulty model. Some programmers also like to optimize their keyboards for their profession. Yet others try to find the layout that scores the highest based on some contrived, nonstandardized metrics. And other designers may have other goals, such as best layout from genetic algorithms, new physical design like Maltron, air typing or touchpad or smartphone (which BTW I have also worked on http://shenafu.com/smf/index.php?topic=88.0) etc. My own goals include not just metrics, speed, and comfort, but also balance, intuitiveness, and aesthetics.

So if your goal is maximum comfort and least effort, then it makes sense to move every key to their optimal position regardless of how other layouts have done before. Since that's what you're going to use the majority of the time you're at your own keyboard, you should use the layout that feels best for you. Nevertheless, if someone wanted me to recommend them an alternative layout, I'd just show them Dvorak because it's a standard layout that can be configured almost as readily as QWERTY on many devices. Custom layouts that people make are really for the hardcore enthusiasts who are dedicated and have the time and patience to learn a new layout anyway. Of course it would be nice if others picked up your layout, like Colemak. Even though it's touted as being not much different than QWERTY and performs very well on most metrics, however, that's still not enough appeal for the masses and corporations to change from the industry standard. So in that regard, I don't see much value in retaining QWERTY like features. It's an option that looks good on paper, but in itself is not a deal breaker for people looking for the most comfortable layout. (And if you're thinking about functions like copy and paste, etc. then I highly recommend assigning mouse buttons for these functions in this era of GUIs. You could also reassign unused keys like Insert, Delete, etc. so that you rely less on agility while reducing strain due to reaching and holding down Control. Not to mention that one-press copy or paste is so much better and faster.)

It's a common fallacy (which I myself was guilty of at the start) to assume that layouts similar to QWERTY are worse than those dissimilar to QWERTY.  In reality, Colemak "beats" Dvorak in many analyses.

That said, there are some interesting optimizations if you go all out.  Notably, I've experimented with putting z and q on the number row, replacing their old positions with punctuation.  (My displacement of the low numbers 2/3 probably wouldn't work very well for vim users, however!)

There's a huge positive that you've possibly not noticed: Tarmak. Between Tarmak and preserved/improved shortcuts, I can see Colemak dislodging Dvorak in the long-run.

If you're only worried about copy/paste, maybe your alternative solution could work (though I'm sort of dubious; far-away buttons like Insert aren't usually preferred for common operations).  What about undo, cut, select all, reload, save, new tab, quit, close tab, find, and print (okay, maybe not print), however?  Are you going to assign new keys for all of them?

I'm sure it does. In reality, there is maybe 3 to 10% difference in metrics between these two. And with just a few modifications to Dvorak, it could match the Colemak in the same analyses. Colemak also has the huge benefit of modern computers to assist in finding an optimal layout without having to spend millions of dollars and hiring people to test every permutation over months and years on original research. I'd say Dvorak is pretty successful considering it was created long before computers were invented and he painstakingly did his own research on improving typing practices which we take for granted today. Not to mention a Dvorak typist actually made the Guinness Book of World Records.

I'm not saying QWERTY based are necessarily bad. Since the keys that are kept are at some of the worst positions, those keys make negligible difference in terms of performance. Most of the work is done on the home row, anyway. What I care about is thinking about improving other areas besides typing English text. Such as Dvorak puts comma and period at the top row, which makes typing IP addresses much easier. Such as unshifting double quotes which is used a lot in novels, in newspapers, and in programming.

Not unless Colemak becomes an ANSI standard like Dvorak, which comes with almost every OS by default. So that changing layouts is a simple matter of going into the OS settings or typing a simple command or shortcut to switch between layouts, and that you can use the same layout on every device you encounter without unnecessary steps (like downloading modmaps or KBLs or PKLs or whatever.) Then at that point can Colemak make more headway into mainstream awareness and adoption.

Tarmak may work for some people, but I think it's mostly unnecessary for people who don't touch type to begin with. Typing tutors are not much different than what Tarmak is essentially: learning to type step by step. There are quite a few Dvorak typing tutors out there. Conversely, some people would rather just deal with one layout rather than relearn 4 or 5, even if they are only transitional.

Dvorak may seem daunting on the surface, but with an online typing tutor (that's very old but still serviceable; there should be better ones out by now), I learned the layout within a week, spending only an hour per day. Truly, once you learn the home row and how many cool sentences you can type with just those keys and how smoothly it feels without fingers jumping everywhere, it'll boost your confidence in making the switch. The remaining keys will fall in place pretty soon after. Then maybe a month of actual usage (typing emails, message boards, chatting, typing tutor practice, etc.) to bring up my speed and accuracy. All you need is one hour per day to practice to reach a decent speed (note only 30 to 40 wpm is required to pass most official typing certification requirements). As you type more, your speed and accuracy will naturally improve over time.

This is about the same time period as most of your testimonials claim about Tarmak. Actually, did I read someone took a year on Tarmak and still haven't jumped to the last step? That may be a disadvantage of Tarmak: procrastination, or not knowing when to move on to the next step. As you say, most of learning a new layout is about muscle memory. So the sooner you take the dive and learn all the keys--especially the home row--then you spend more time on improving speed and accuracy.

Yes, changing keyboard layouts is hardly trivial.  It's very esoteric.  On the Linux console it requires root access.  Having to fart around with binary blobs isn't much fun either.  I've muttered it before, but I really think it's worth badgering the OS/device makers into making alternative layout adoption and switching between layouts much easier.  You'd be better to push for an agreed portable keyboard layout format (if it doesn't exist already).  I'd love to carry around a tweaked layout, but currently that is way too much effort for me.  And I don't see why Workman users and those of other esoteric layouts shouldn't get access to their desired input method also.  The mainstream vendors should be quite ashamed of themselves.  It really is an accessibility issue.

I used a version or two back of OSX last week, and there were no layout switching options at the login screen.  I'm running Debian stable with Xfce, and the layout switcher continually craps out on me.  It's annoying to have to think and fuss about.  It's an embarrassement and almost enough for me to deter others from even trying an alternative.

Was just reading through this old thread on the Workman layout, as the problem of awkward movement to type D and H was one I also experienced when I first started with Colemak. Although I have largely become adapted it to it now, I would still recognise it as one of Colemak's (very few) weaknesses.

But it occurred to me that if the objective is to de-prioritize the centre-column, as is the aim of Workman, this could be done with a much less drastic change to Colemak. You could for example, switch H - M and also move D > P > G.  This does not change key-fingering at all, but would probably make both D and H easier to type, especially those HE digraphs.

For clarity:

Q  W  F  D  P  J  L  U  Y
A  R  S  T  G  M  N  E  I  O
Z  X  C  V  B  K  H  ,  .

After I had gotten to 80 wpm on Colemak and my right hand and forearm was in a lot of pain (from the dreaded HE combo), I actually tried this mod and immediately liked it.  Unfortunately my pain didn't go away.  To be fair, I only used it for a week, and that probably wasn't enough time to really give it a chance.

But that's in the past.  I am now completely sold on Workman.  It is *WAY* more comfortable to type than Colemak and Dvorak.  Within two days of switching back to Workman, my pain was gone.  I had originally thought that Workman was bottlenecked on speed, but I am up in the 90's now, which is really close to my old Dvorak speed.  Given time, i'm sure I will break my old speed records.

I am more and more convinced that the typing efficiency model that is used to "optimize" so many different keyboard layouts is fundamentally flawed.  When Shai commented on the Workman layout, he cited a bunch of reasons why it was worse than Colemak.  Yet as one who has typed Colemak, Dvorak, and Workman at a high level, I can say with complete certainty that Workman is superior, especially for those who are prone to RSI.

Saying that you should use alternative fingering (re your 'no angle mod') effectively means the same as moving more keys from their original positions.

So, how many changed keys from QWERTY depending on previous technique? I'm not used to seeing it presented as keys kept in place, but as keys moved. That's, ummm, 25 keys you're counting then, which means 14–17 keys moved depending on your ZXC technique?

Sounds like a serious competitor to Workman at least, then. Then again, so is your Colemak-DH right?!

One reason the S was finally uprooted from its old position in the final Colemak was some bigrams really needed fixing. I can see at once that SW is painful if you keep both S and W (and WS too on some keyboards). The whole thing made me wonder why the W wasn't moved instead, but I don't know enough to say.

Good point, Lalop. Indeed, moving fewer keys is a boon but in this case it's a question of whether to move the much-used S (also the site of Ctrl+S) or the much-less-used W (and while Ctrl+W closes stuff it's usually not disastrous – although I did have some swearing bouts over it back in my Dvorak days!).

Workman chart distinguishes QWERTY W/G and H/O.