On Facebook today, a friend of mine was leading a discussion about terrible project naming. She asked "Why are you developing software and sharing it if you don't want people to use it?"… For me, there are a lot of reasons: I was developing it for my own use anyway, and put it up to show others what I have done; I'm really just sharing with my friends; I want my students to learn from it; I'm looking for a collaborator who might take it forward. In all of these cases, a poor name might be only a slight impediment.
That said, I spend a lot of effort on naming, and strongly encourage my students to do likewise. In my open source development class I usually spend a full hour talking about project naming. I'm not so concerned about the straight marketability of a name as its brevity, descriptive power, and uniqueness. But oddly, if you get all of those things you probably have a market-ready name too. (The most obvious exception probably being "not in good taste" :-)).
See, for example, the rationale and history for the name of our Nickle programming language. That page is not a retcon: we really did spend that much time on just the name.
It weirds me out that a subculture that would never randomly name a variable in their program "floofram" or "montyPythonReference" will happily do that kind of thing with the project as a whole. (B)