Open Source Is Better Than the Closed Stuff (Until You Hit 1 Million Lines)

Would this mean that open source communities would work better for modular systems, since each independent module has a relatively small codebase? Unix & NPM come to mind.

It sounds like they are finding numbers to support the specific application of the bazaar and cathedral models for projects.

I don't completely agree with the methods (static analysis), or the sample size (368 projects), they used, but they were consistant when measuring. (Which is better than most project quality metrics)

I seriously wonder about sample size and representativeness. What projects are they looking at in either group?

Static analysis alone seems like a truly poor way of judging "defects" in any non-trivial system.

Sounds like a good enough heuristic to me.

Doesn't have to be perfect, just indicative.

Not to mention, you could just tune your code base to get good marks on the scanner and get govt contracts.