- A lack of marketing to make potential users aware of the existence of mature projects
- Lack of excellent documentation
- Inconsistent or non-existent install scripts
- Lack of knowledgeable and available support staff
- Inconsistent and generally unknowable quality of projects
- Developer-centric designs highlighting feature creep
- Complexities of licensing
I've been meaning to write more on this, particularly in the context of educational software, where more and more people (probably as a reaction to the WebCT/Blackboard/etc. monoliths) are again starting to jump on the open source bandwagon. But the bulk of the enthusiasts pushing this, many of whom I have a great deal of respect for, are prone too much to boosterism, and gloss over the serious pitfalls. It's not to say that those pitfalls can't be addressed (the open outsourcing idea is one suggested approach), but I think it's disingenuous in the extreme to ignore them.
More at a later date, hopefully.
I don't know where I'd put any of the really neat examples, so I dug around on the net and found a copy of GroupKit 3.3 (circa early 1997) that someone had kept (now copied over to