Remember though, that the best ideas initially sound like really bad ideas, as explained by PG in his essay (based on pretty good empirical evidence from funding hundreds of them ): http://paulgraham.com/swan.html

So there could well be a few outliers in there. Of course, that is a necessary, not a sufficient condition, i.e, not all bad ideas are going to be successful - but the most successful ones sound like really bad ideas to begin with.