>But if that's why Linux has less usage share, then backwards/forwards compatibility doesn't matter — it's because of lock-in, not lack of backwards/forward compatibility.
That argument is growing old. Mac OS X started from basically the same position as Linux. I.e. no native software and a crufty compatibility layer (Classic Environment vs. Wine) with another major OS.
Yet Mac OS X captured more than 10% of the market, even though it requires overpriced proprietary hardware.
Android gains more users each month then the total number of Linux Desktop users. Even the frigging Windows Mobile has more users then Linux Desktop.
>Sure, many users want programs to "just work" without needing to tinker with them.
Not 'many'. It's 'the majority'.
>But users like those are unlikely to contribute to the development of the programs they use, and the entire purpose of FLOSS is to make it possible for every user to become a developer.
Wrong. I can download VisualStudio Express (for free!) and start developing Windows Phone applications in 3 minutes. With nice tutorials, great help system and one of the best IDEs.
>For example, it's better to have 10000 users where 1000 of them contribute to the software, than to have 1 million users where only 50 of them contribute to the software.
Nope, it isn't. Because 1 million users would be able to sponsor (say) 100 full-time professional developers, artists, testers, help writers. Who are going to produce software that these users really like.
Of course, if your target is to create a playground for developers, then I suggest looking at OpenBSD ( http://lwn.net/Articles/449697/ ):
> "We hack OpenBSD for ourselves. Not for you. Not for the users. If
> the users end up enjoying what we have created for themselves, good
> for them." (c) Theo de Raadt
Well, the only problem is that one day you might find yourself without open hardware capable of running your OS. And nobody would care about you.