Thanks for the quick and detailed response.
I'm going to start using Colemak very soon and will report back with my experiences.

Yes - if common digraphs such as st, ca/ac, and ho have to be typed with only one finger this is not good.

I have been typing on a slightly modified version of the bulpkm layout for some time now and it is surprisingly comfortable to type on.  The biggest problem with the unaltered layout is the sequence "st".  I solved this by swapping 's' and 'd', which happens to match the dvorak layout.  This one swap significantly reduces the same finger sequences without introducing any additional typing effort or penalties.  It also lessens the slightly overloaded right index finger.  And of course I set my caps lock button to act as backspace.  The main reason I like my modified BUL layout is because it flows smoother than both dvorak and colemak in my experience.  I have typed Dvorak for over two years (peaked at about 100 wpm), and admittedly not nearly as long with colemak (about 40 wpm) - but enough to really get a feel for the flow of it. 

I don't mean do diminish the contributions of Shai - I just wanted to post my experience with the BUL layout because several people have asked about it and no one seems to have any real experience with it.

Colemak, like Dvorak, has ready-to-use implementations for the major platforms.  The CarpalX layouts lack that.

Also, I think Colemak offers the better trade-off between being ergonomic and familiar; only those keys that make a significant difference are moved.  It really doesn't strike me as elegant to have punctuation and letter keys mixed as in eg. the "XBUL,W" layout.  From all the fully optimized CarpalX layouts, "BULPKM" appeals most to me (at first sight), perhaps precisely because it has the most restrictions applied.

Keep in mind that the creator of carpalX (Martin Krzywinski) is a PhD researcher in Genome Sciences whose primary interest is research.  I'm sure he wasn't out to generate a single "optimal" layout with which to promote and "sway the masses".  His website mentions how subjective a keyboard layout is, and so he developed the carpalX optimizer to let anyone tweak the settings and find a good layout for whatever domain they are working in.  You can easily adjust the penalty/effort parameters in the config files to make it suit your needs.  I have spent months and months doing this - trying to come up with a better layout for my needs.  I took many approaches - for exaple, I initially tried to minimize same finger sequences, and ended up developing a layout that had even less same finger use than colemak does (which was pretty hard to do by the way - colemak has excellent same finger use stats).  I then decided that same finger sequences, while important, is not the most important thing to fast typing - stroke path, or "flow" is more important.  Some same finger sequences do not slow you down at all - in fact, some are very fast to type.  Others will stop you dead in your tracks.

In my experience everything is very closely knit - if you over optimize for one area, you usually end up having to sacrifice somewhere else.  After generating several layouts that are comparable to colemak, I decided that the best route for me was to manually optimize a few things in the BULPKM layout because it flowed under my fingers better than any other layout I've tried.

As for being "available" for all operating systems, I can easily create an installer for windows and linux (I don't use Macs, but I'm sure it can be done there too)

For the curious, here's my modified version:

" B U L  P _ M Y F K Z + |
R I A O H D T E N S ?
: X C V Q G W < > J

perhaps the worst same-finger sequence that remains is "ca", but it hasn't bothered be enough to do anything about it yet.  All in all, the layout is noticeably nicer than dvorak and colemak for my needs.

It changes 17 keys, but unlike Colemak it moves them between hands and fingers much more.

Not sure what exactly makes your scoring system tick, but I'm starting to distrust scoring systems in general as there's so much they don't pick up and so much weird they seem to do. But that's just me I guess.

I've been saying this for months. I've seen people distrust their own subjective conclusions because they do not match the results from some scoring algorithm. In this case, a great algorithm tries to predict the subjective experience - i.e. it is the subjective that is the ultimate authority. It is my experience from years of designing optimising algorithms that computers are stupid (but fast), humans are smart (but slow). Put in another way, when looking at a single layout a human will make a better valuation of its strength than a computer, but a computer can produce and evaluate many, many more and is very useful indeed to find candidates for the human.

If you don't have a lot of time to subjectively try a layout, it can still be useful to be computer guided (eg. tell you of any same-finger problems) in case some problems escape your subjective test corpus.

You maxed out after a week? You mean to say that if you practiced Colemak for another year you would still be at 35 WPM?

Otherwise your layout seems to address the better part of the flaws of qwerty. From a casual inspection I pick up a few same-finger digraphs for de/ed, ce/ec, fr, gr, tr/rt, br (left hand), and lo/ol, nu/un, mu, hu (right hand). Better placement of e, t and perhaps n (your reasons for it staying nonwithstanding) would be adviced if you wanted to further enhance (and complicate) the layout.

What hype did you encounter that made you think a week of learning would be representative? None that I've seen I think. Sounds like you've been reading whatever 'hype' exists with your ADHD glasses on?  ;)

Sounds like you're one of those 'layout shoppers' who try (then generate and try) a multitude of layouts for a few weeks each, post a lot about them before leaving them and then only much later start to wonder what you've been missing. Or never even get that far. Not to judge you harshly, but that's the impression I'm getting here at least?

bsdhacker, don't you feel there is also a problem with the same-finger sequences ol, lo, op, po, ho? Their combined frequencies are more than that of ca/ac. I wonder if it is worth placing a more frequent letter under the pinkie by swapping O and R in order to reduce the same-finger problem in the middle. (That would also make ob/bo and possibly other things a little more difficult to type).

What bothers me in your layout is that XCV<> are kept in their QWERTY places. I would consider switching to a custom layout, but only if it is fully optimized for typing.

"Hype" may be a bit strong of a word, but basically, all the praise and attention the layout was getting (from here, hi-games, keybr, that DDvorak site, etc.), and how it was supposed to be so much more comfortable and elegant to type on than QWERTY, and how it was superior even to Dvorak, and how it had so many "finger rolls," and heck, even a Slashdot story.  This all lead me to believe switching to Colemak was going to be some kind of transcendent experience--that every word would flow sweetly from my fingers like honey and I'd immediately refuse to ever go back to QWERTY because Colemak was just OMG so much better.  What I actually experienced was a layout that was better thought-out than QWERTY, but that still had considerable awkwardness about it.  Common trigraphs like YOU, THE, AND, AST, ION, and ING were all rather awkward to type.  The right pinky finger was hammered (seriously, QWERTY got it wrong by putting A on the left pinky--why compound that by also putting O on the right pinky?  We're talking the 3rd and 4th most common letters on the two weakest fingers, here!  There's more to keyboard optimization than minimizing same-finger!)

Saying one should spend a whole year with something they're unimpressed with after 40 hours sounds like a common trick I've seen employed by fans of certain book, movie, and television series.  The argument is, dissenters don't appreciate the series simply because haven't spent as much time with it as the fans have.  The end goal is either to dismiss all dissenters (thus increasing the series's positive mindshare), or to get the dissenter to invest as much time into the series as the fan has (thus increasing that series's total mindshare).  Most recently, all the tween girls are using this argument with the Twilight series.  "Don't judge the series by the first book/movie!  You have to read all four books several times to really get it."  Yeah, right--let's all go through the motions of being fans of something we're not fans of.  *eye roll*

(Incidentally, I tried guitar for a week and was poor at it.  People served me up heaping pitchers full of that "you can't know after a week" Kool-aid and I, in my youthful naivete, drank it:  I went on to practice it for over a year, often several hours a day.  The end result?  I could (poorly) play parts of a few songs and a couple of box scales.  Technically, I was little better than I was after a week.  To this day I'm unable to reliably play the A-E-D-E open chord progression.  One week was more than enough to prove to me that I didn't have the fingers/touch/coordination/dexterity/whatever to be any good at the instrument, but I thought maybe the people telling me the grass was blue and the skies were green knew something I didn't.  Pfft.)

Personally, I think 40 hours is WAY more than enough time to develop a reasonable opinion about something as simple as a keyboard layout, especially one that's only 17 keys different from one you already know!  Heck, even a few minutes can tell you a lot.  Simply type some example passages as if on the new layout, but still using the old layout.  Then, go back and type the character sequences that resulted, using the layout you know.  That'll give you a pretty good idea of what the new layout would feel like were you proficient with it.  I like to start with the "man from Nantucket" limerick, since it's particularly awkward to type on QWERTY (and that's saying something).

Anyway, I'm not trying to rip on Colemak, here; I'm just trying to explain (and defend) my experience with it.  I have more respect for the layout than is coming through in this post:  If I saw no merit in the layout, I wouldn't be here.  (Although, perhaps like vVv, I think the whole ZXCV thing is a bit smoke and mirrors.  But then, I grew up old-school, when copy was CTRL+INS, cut was SHIFT+DEL, and paste was SHIFT+INS, and we liked it that way, because we selected text with the arrow keys, dammit!  Personally, I still do it that way; trying to select text with a mouse in Windows is like trying to herd cats across a frozen pond, especially when scrolling is involved!)

This does sound like a rationalisation to me. J. Pierpoint Morgan said people often have two reasons for most things - one that sounds good, and the real reason. In sales its called figuring out the hidden objection.

All the trigraphs you mention I think are supremely comfortable although I didn't like all of them at first. My right pinky does not get tired. At the risk of beating a dead horse, your experience is likely because you did not spend sufficient time to allow your motor skills and muscles to get accustomed.

Note that I never suggested, nor do I greatly care if you stick with Colemak. I adviced you should pick the one layout you fancy and stick with it. You're still welcome to post here about it once you have had time to form an opinion on it.

You are certainly right, and well put at that. Trying a layout for a month should be sufficient time to get at least a good idea if it's going to work out for you. Don't waste any more time than that at any single layout! Then pick one - whatever seemed to work best for you - and stick with it! (Or in your case - pick two!)

What I meant is that to become truly proficient in a layout I think you do have to stick to your guns for a year, and you will probably not realise your full potential until a few years in. People are definitely in a position to pick a good choice of layout for themselves much sooner, but if you want to opine about which layout is the best (and not just for you) then I think you had better have a lot more time to soak in the layout so you can say with confidence you know what you are talking about.

I am still picking up speed after a year. Speed seems to come in quanta of about 5 wpm every two or three months or so without me doing any particular practice or anything. I surpassed my (perhaps not so impressive 80 wpm) qwerty speed only fairly recently, and I wouldn't be surprised if I hit 100 wpm in another year or two. At the same time, I am still getting more and more comfortable with the layout. For me, it really does take time for the muscles and brain to mature, and I hazard to guess other people are the same.

I believe Dreymar has had a similar experience. He would have to verify this, but I seem to recall he was stuck at 50 wpm for several months and only recently jumped to 70 wpm.