[gutnor]

Ignoring the fact that the majority of the source code is not open sourced and that "impact" biggest variable is going to be chance.

Impact of one open source project measure slightly more than developer talent. It is a combination of a lot of skills, development skill just one among other like marketing, business sense, ability to make video, ... Extreme example would be Zuckerberg: is he the best developer in the world and wouldn't you waste a bit his potential if you hired him as a PHP dev?

And in any case, impact is always going to be a star system. You are going to have a tiny percentage of developers (like a few thousands in the world) that have impact and all the other that have 0. You are back to square one at trying to find a way to differentiate between a loser that can code and another loser that cannot.