These are kind of case-by-case decisions. It depends on the contributor community. In the case referenced, we did have a CLA. Most of the potential (and actual)

These are kind of case-by-case decisions. It depends on the contributor community. In the case referenced, we did have a CLA. Most of the potential (and actual) contributors were academics, supported by their universities or by research grants. AGPL allowed full scientific transparency, without any real downsides. There were other contributors who were turned away by the CLA, but it was not an unreasonable trade-off. We still had hundreds of contributors. And we did release a small piece of the system under Apache, because we wanted an industry standard for an API/format. Having that piece under Apache encouraged competitors to adopt it (which didn't really happen, unfortunately). On project I'm currently leading -- and I can't speak for success yet -- is AGPL without a CLA. Most of the business value comes from network effects. A good analogy would be if Facebook had started out AGPL (or Wikimedia, for that matter). The code has little value independently of the community and partnerships I'm building. AGPL allows for full transparency, which has a ton of value in the market we're targeting (at least for people who think through these things deeply). For people who don't think deeply, having an FSF/OSI-approved license gives name recognition and opens doors. I'm leading two major projects. The other is pursuing a closed model because it makes sense there. I can't speak for any specific case. All I'm trying to say is it's an in-depth case-by-case analysis. Businesses which pick closed because they "want to make money," open "to contribute," BSD to "be more open," or similar knee-jerk reactions are doing it wrong. It is a complex analysis, and my case is different from yours.