Unfortunately the data on this kind of thing is rather small. The only way to be sure which is better is to run two funds side by side with different strategies.
edit: It should be remembered that Viaweb had not only great developers (the functional stuff in Viaweb is cool!) but also a pretty big idea. Viaweb is really a bad example for "ideas don't matter".
Except the initial idea was "put art galleries on the web", called Artix. It morphed into Viaweb later. It's actually a very good example for "ideas don't matter".
I wonder how many startups actually stick to their first idea, and never deviate from that.
Interesting stat, which is obviously an opinion.
Here's a strongly held opinion of mine: top 1% developers != top 1% of business people.