I'm not so sure I agree tomlu. You probably know the quote about lies, damned lies and statistics. Look at the bother Dvorak had over his stats.

I don't think that there is much point in promoting Colemak, or providing 'facts' to prove its benefits. People can use it or not. I use it and I enjoy using it. All of my friends, workmates and family know I use it. None of them, not even the ones who can't touch type, are in the slightest bit interested. They don't care. They think I'm a bit odd in fact. They aren't interested in efficiency or comfort. Fine with me. Each to their own.

Some people, such as me, are happy they switched. Others will find fault no matter what. Fast Qwerty typists become frustrated when they plateau. They haven't hit 100wpm in a few months, so Colemak was a waste of time. They forget that the 100wpm they achieved with Qwerty took 10 years!

I think Tomlu's idea is briliant.

Everyone I know is almost exclusively concerned with efficiency/productivity. The first question I get asked when people see my layout is, "Can you type faster on it?" The second question is, "Does is feel better?"

I think disclosure of as much info as possible, comparing Colemak to the alternatives, is a good thing - there shouldn't be any "filtering" of what information to compare and what not to compare. That would end up giving a one sided appearance.

What's the point of comparing some abstract qualities of colemak (more combos, less row jumping, lower hand alteration, lower finger distance travelled, etc.) when real world experiences are better?

We should have something like the following:

Total layouts learned in the past:
Time spent on each layout:
Top speed attained on each layout:
Time taken to achieve top speed:
Relative comfort between layouts:
Worst aspects of each layout:
Best aspects of each layout:
Any additional comments:

And then have an average of all the values on a graph.

Regarding the collection of a representative sample...that might not be possible. But, I think having a poll on the forum and linking it to the main colemak page would be great.

Isn't this pretty much exactly what the colemak site does?

For someone new that visits the page, I think it's reasonable to ask what benefits the layout brings, with proof or quantification if possible. People can use it or not, but fewer will if they think the layout is slower. If it isn't, then that's a shame for those individuals who might otherwise have tried it.

I think this is bit of a hasty conclusion.

First question that needs to be answered is, What's the percentage of Qwerty users that actually type over 100wpm ?

What the number of people that can actually touch type in Qwerty around the world?  is it a billion ? 

are we talking an exceedingly small percentage that has the necessity and discipline to maintain speed over 100wpm?  or is it high,  1 out of 5?   1 in 10?  1 in 100?  1 in 1000?  1 in 10,000? 

Maybe it takes extra work and discipline regardless of how good a layout is or how much experience and speed one has in another layout to go from 60-70 to 110-120 ?   How many adults are up for that ? 

Maybe the Colemak population of users is too small to have more individuals like Heise ?   

Are too many of the rest of us when we get to 60-70 wpm become slackers ? 
Maybe Colemak is self-selective for slackers since Dvorak is too much work to get that rhythm and Qwerty has us working so many more miles across the keyboard?

:-)

Interesting article. Reading it, their 3475-typist sample of best results for each typist on a 5-minute test shows ".17266%" reaching 100+ WPM; that is, 6 typists. If those data were directly transferrable to Colemak data, we could expect about 3 out of 1500 Colemak typists reaching 100+ WPM given enough time for training. (Over time, it's very likely that the 1500 figure will change of course.) The point that most of the QWERTY typists have trained for 10+ years is obviously significant.

These numbers shouldn't be directly transferrable I think. The error weighting is one issue; as I understood it the article didn't really penalize errors but merely sorted an error-free # WPM above one with the same speed but higher error rate. Also, those were typists applying for a job in a company. We in the Colemak community are a somewhat different lot. There's no saying how we will fare over time really.

I wish people would relax a bit with the unrealistic expectations. In that article's sample population, the mean is 40 WPM and 65 WPM is the gateway to the upper decile (10%) of a sample of (self-proclaimed) trained professional typists! Reading that, I realized that while I'm a lot slower than the best on this board I could actually pass one of those strict typing tests for a secretary job. (Well... naaah... I'll just stick to being a scientist for now!)

Well, that certainly makes me feel better about my own typing speed.

I'm currently somewhere around my old QWERTY speed, give or take (I never really cared much about speed, so I haven't made much of an effort to measure either speed), but my QWERTY speed was based on roughly two decades of memorized flailing around, while my two years of Colemak is a slow and steady application of proper touch typing.

And both are right around the mean typing speed, which isn't bad for someone not overly concerned with speed. More importantly (to me) is that typing just feels to much better now.

I hadn't tried the the flash typing tutor before, that seems new. Not rigorous stats by a longshot but interesting nonetheless to look at in light of that study.
    http://www.keybr.com/highscores

I hit 57wpm with zero errors on the random text the first time I tried it even given it seems a bit funky in Safari.  Got to really pay attention because some times it starts as word and then just truncates it.

It work best for me to drag the window so the keyboard is off the screen and all that flashing of the keys is not dancing in my peripheral vision.  I find that very distracting.

Now, if only the sample populations of the TypingSpeed article and our community were comparable one would be tempted to run a test of significance even though p is very small. I can tell right away that the test wouldn't prove a difference between QWERTY speeds and Colemak speeds, but the interesting part would be what one got on a 1-sided test of simonh's null hypothesis.

I dont know why but my querty speed is much better since learning colemak and ive been typing for years. My accuracy now is much higher, my finger dont feel jumbled or dsylexic or they feel fast and accurate like theres not effort.