I'm not complaining I didn't win: the judges clearly thought the other competitors did a better job, and they were probably right. But it's anything but easy money.
TopCoder seems to be overrun with developers from developing (largely Asian, as far as I can tell) countries where salaries (and cost of living) is much, MUCH lower, which is also keeping the prize money low. $500 (before tax) for a week's worth of work is just not attractive to me, I can make (and have made) 3-5x that doing normal consulting. (where the probability of being paid in the first place is a lot higher)
There is, of course, nothing actually wrong with this. This may well be the best way to outsource development to cheap labour: if it's not good enough, don't pay them anything!
There's probably a certain amount of practice to this, over time you figure out what the judges want, and what not to spend time on, but to make a competitive amount of money, you'd still have to have a very high success rate. The article is somewhat misleading in this respect.
That said, if you're in it for the fun, it's great!
Besides, it's what you call a (sustainable) business model.
The dev/design competitions were, for me, quite painful. It was an incredible chore to force my self into using their process, and as I was banking on it for some income, the discovery that solutions were rejected because I was missing absurd amounts of documentation/unit tests meant I don't see myself going back to it any time soon.
The design & dev competitions began to overwhelm the community as well. I haven't played there for a while.