The comment in the blog about 10,000 hours for expert performance is well put. The research supports 10,000 hours but 10,000 hours in deliberate practice rather than mere repetition. The effort of such practice can only really be sustained for about 4 hours a day. Thus 10,000 hours takes 10 years if done 5 days a week, 50 weeks a year. Just doing the same job for 3 years (even putting in 60 hour weeks) will not reach expert levels.
I read the whole article. It is designed for certain types of task, and it is not appropriate to others. For example, sometimes my tasks involve, "travel to the city and speak with Joe." "Travel to the city" takes 40 minutes. It should not be split into two work units with a break in between. Further, "speak with Joe" would not be splittable. How am I going to say, "Joe, when the timer rings, we are going to get up and not talk to one another for 5 minutes."
However, for desk-work it might help me. I have to stop every 20 minutes and take a 5 minute break anyway due to RSI. Using this interval as an atomic unit for task completion, recording, and estimation may be useful.
Having tried to do the same thing (I didn't realize it had been done already until this morning), I AM quite impressed with the results. They do have varying degrees of success, but, in all, quite amazing.
Further, their scheme to work out the bugs is nice too. They are taking ratings, then if I were them, I'd look at the lowest rated, but still rated models for pictures on which their software does not work too well.