The "just write this one program" philosophy always just yields one program at most. The "add software to the community" approach might add only 1/10th of a program, but with thousand people doing it you get 100 programs out of it. Which approach looks more reasonable?

Or to give another example how many people can enjoy Game of Thrones because it's written in English? (I'm German and I love it!) If it would have been written in a minority language that is only spoken in one village somewhere in central Asia it might have been more comfortable for the developer, but only few people can ever take joy in it.

They don't need to make the program work on all systems or use a fancy new language. They just need to enable the community and make use of development patterns that are flexible. Java certainly is not a fancy new language, and despite it's claim it only really works on Windows well enough out of the box. Yet Minecraft has an open community where even the developers profit from what others add to the game. When the coders of DF die in an airplane crash tomorrow DF is pretty much dead as well and that's the big problem I think. If I invest my whole life to create an awesome piece of work, then it should surpass me in number of years.