This is an interesting piece: http://insight.zdnet.co.uk/software/linuxunix/0,39020472,39236745,00.htm. Most noteworthy is that the South African govenment is funding the translation of OpenOffice into South Africa's 11 leading languages. Microsoft with its elderly version of Office can only offer two languages (sort of) for South Africa. The South African government can either procure a massive number of Office licenses that are only partially useful; or they can get a product for free and sink a large chunk of change into making it ideal. That would seem like a waste of money. It's like buying a burger, taking it to a chef and asking him make it into a stew. But that's the problem with burger joint (Microsoft): they'll refuse to make you a stew no matter how many people are lining up asking for it. Government have the size and the money to afford application refinement. Also, when they throw money down a spout, no one questions it. Governments can also serve as vested patrons who can drive an application towards improvement to the benefit of the whole of the user base.
This is where/how Open Source can win with part of its classic model. That part of the model being that consulting is where places like RedHat and so many others make their money. Developers contribute code, bug notices, fixes and documentation (hey, it could happen); as part of paying their dues. They become experts on a particular product. This poises them to branch the code or enhance it for select clients.
I am working through a long and elaborate modfication of phpBB. When I am done, I will be really well suited to help evolve the code or help others modify it for their purposes. If someone obtains freebie code and tries to use it out of the box, they are unlikely to get a glove fit. Spending on modifications still gives you a chance to win as you don't have to pop for the whole development bill. But then, you probably know that already ;)