Here are the things, that the people dispersing grant monies tell us are true (because they genuinely believe it to be true), but that they actually have no evidence for:

1. People will not build something, unless you pay them to build it.

2. People will build better things, if you pay them to build it.

3. One person building 8 prototypes of a concept, over the span of two months, will invariably result in a superior concept over one that took a different person two months to build.

4. In an unstable protocol environment, you should work sloppily because things change fast.

5. 4 people building 4 implementations of the same concept is better than 4 people working on 1 implementation of a concept.

6. There is no point in investing in a firm architectural, quality control, and environmental basis for your software development. Just hack shit out. Way cooler.

7. Lots of lay users testing the software is always more efficient than one knowledgeable engineer testing the same software.

8. How much automated test coverage should your build server check for? Zero. Stop wasting time. Ship faster. Also, what is a build server?

9. It's normal for things to break, in the next version, that were working in the last version.

10. It is always better for a software developer to work on implementations for one protocol, full-time, in perpetuity. That way, he becomes a Protocol Implementation Expert.