Not sure if you guys mean the same thing when talking about keyboard layout, there's physical layout (shape) and logical layout (mapping of symbols to keys). To me it seems clear that the shape of a keyboard is much more important. A "good" logical layout reduces the number of reaches, because frequent keys are easy to reach - so there's less potentially uncomfortable movements, however on an ergonomic keyboards those movements aren't nearly as bad/uncomfortable as on a conventional one.
Don't think there's any papers for it, AFAIK Shai developed it using common sense, analysis of finger distances, same finger occurences, hand balance, consideration of finger strength/dexterity, comfort of common character sequences etc. From what I remember there's loads of bits and pieces of info in the forum, for example why he moved the S key one position to the right. It all sounds very convincing, but it's hard to find. You basically have to look through all of Shai's posts.
Updated: Here are a few examples:
Shai on the Workman layout
Shai on G and P
Shai on B and J
Shai on 2 backspace keys
Shai on S
Shai on Maltron, Capewell, Arensito and Dvorak
From the Colemak FAQ
Obviously I can't speak for anybody but myself, but I look at the "ergonomic" keyboard in two separate ways. What you call physical layout I usually refer to as the "form" or "shape" and the logical layout as the "distribution", so we could be getting confused if I don't clarify.
My own experience is that the form and the distribution are both very important, that together they are worth more than the sum of their parts.
To give an idea of how changing only one factor can make a visible difference, I've linked to three Youtube videos.
http://www.youtube.com/watch?v=pYJtF1I3PRshttp://www.youtube.com/watch?v=o4H931A3BDE&feature=relatedhttp://www.youtube.com/watch?v=hxH7Uith0tQThe first shows the use of a Maltron keyboard under the Malt layout. If you'll observe closely the hands are almost continuously resting on the keyboard, with virtually no "hovering", meaning that there is no "gross" movement of the arms. Striking the keys is achieved almost solely by extending and contracting the fingers. (At the very end of the video the left hand is lifted to click the mouse and you can see the rubber pad stuck on the front of the case where the base of the hand rests.)
The second video shows the same Maltron "form" but using the QWERTY distribution. The hands very seldom rest and there is almost continuous "gross" movement using the arms and shoulders.
The third video shows a Kinesis keyboard being operated under the Dvorak system. Whilst this is not an exact replication of the Maltron form it is reasonably close, and it can be seen that whilst the operator's hands are using fewer and smaller gross movements than shown with Maltron/QWERTY, there is more movement than seen under Maltron/Malt combination.
For the purposes of determining the "ergonomic value" of the key distribution, as I mentioned above, I have some criteria which I use. Based on these criteria I've found a practical measure is the number of "home row words", ie the number of different words which can be keyed in without moving the hands from the home row. The logic behind this is as follows. If a key can be struck without moving the finger from its resting position, then there is no extraneous work being done, including the rather minute effort (on the Maltron keyboard anyhow) of extending or contracting the fingers to the row above or below the home row.
Another reason for doing it this way is that not one on-line "measuring" system that I've tried can be adapted to the Malt system, which has the left thumb on a letter - "E", which means that the Malt has at least one more "home key" than most other distributions. (Some people seem to regard this as "cheating" somehow, but I could never see it myself. :-D)
In this blog link
http://mostergonomickeyboard.blogspot.com.au/I go into great detail as to what I regard as an efficient and ergonomic keyboard and why, but in brief based upon an international Scrabble word list of 172,807 words, the following figures were derived. (The actual word lists are linked below.)
http://proword-keyboardlayoutefficiency.blogspot.com/QWERTY - 198 different words can be typed without taking the fingers from the home keys.
Dvorak - 3126 different words can be typed without taking the fingers from the home keys.
Colemak - 5963 different words can be typed without taking the fingers from the home keys.
Malt - 7639 different words can be typed without taking the fingers from the home keys.
As I say in my blog, I don't regard this as the ONLY criterion, but it's a very simple (and useful) starting point.
I had a read through Shai's links which you put up, and note a couple of points. In respect of not being able to take a Maltron keyboard with a laptop, I'd beg to differ quite significantly, as for many years I was going to different organisations for the purpose of taking a written record of proceedings (not quite verbatim - for that I'd need to make an audio recording for later transcription) and I had absolutely no difficulty with having the Maltron with me ... and I was always riding a motorcycle, so my carrying capacity was extremely limited. Further, as a "Temp" legal secretary, I would sometimes be going to two different legal firms in a week to do word processing, and I always took my Maltron, with no problems at all. Secondly the point about overworking the pinky has never struck me, and I'd refer the reader to the video showing the Malt layout being used. When I bought my first Maltron (1986) it came with a fairly comprehensive book of practice lessons. These days the Maltron website has a very extensive suite of on-line practice typing, for all the different Maltron layouts and different forms, for one or two hands.
Joe