Keith Sawyer, who studies creativity, pushes this in the Huffington Post:
When Krzysztof Klincewicz, a management professor at the Tokyo Institute of Technology, analyzed the 500 top open source projects on SourceForge.net, he found that only 5 of the 500 -- one percent -- were examples of radical innovation.
He's right. (The study is two years old.) Most open source projects are not, in and of themselves, radically innovative. Programs like BitTorrent which break new ground are an exception, not the rule. Ideas like the World Wide Web, which has depended on the free IP model since it started, are rare.
But isn't this true in general? Most new software is not radically innovative, only incrementally so. What open source delivers is a process for innovation and open collaboration. This can be seen in programs like SocialText, an enterprise Wiki. (Wikis are pretty innovative.)