Right -- see, I basically maxed out "Mavis Beacon Teaches Typing" in high school. (That is, I practiced with it regularly until it couldn't find any particular problem with my typing, and resorted to giving me practice typing symbols, at which point I got bored and quit.) What prompted me to look at Colemak (and mechanical keyboards) was when some friend/colleagues at work were playing with typeracer.com, and I clocked in at 120WPM on one of the first races. One of my friends didn't really believe it and asked me to repeat the performance in front of him, which I did.

That made me think, maybe if I switched to Colemak and got a better keyboard, I could go even faster. I did both; and though my forearms certainly feel less tired on the odd occasion that I have long bouts of continuous typing to do, I'm neither noticeably faster nor more accurate.

So, both our anecdata seem to match the conclusion of the 1956 study mentioned in this article: That it was the intentional practice that primarily resulted in the improvements, not the keyboard layout; intentional practice on QWERTY would probably yield similar speed improvements to practice on Colemak or Dvorak.

I've had a similar history, and also switched mostly to reduce fatigue & the risk of RSI. Colemak seems easier to use, even if it's not any faster. Of course it's quite possible that the improvement is simply due to the deliberate practice needed to switch layouts; even more than learning to type in the first place switching layouts takes focus and therefore makes it easier to notice & correct bad habits.