But try constructing an Independence Day without Will Smith. Or the special effects. Or the soundtrack. Or the distribution chain. Try guaranteeing that it arrives on schedule without making a loss. I think you will not be able to accomplish that with a volunteer effort. Try doing that tens of thousands of times a year (that's for all of what is currently covered by IP) and you're bound to fail. Unlike with Linux, the individual parts of most larger projects involving IP are of no use without the surrounding whole. Unlike Linux, many IP products aren't modular, reusable or decomposable, and so they can only exist if you can find a single source of financing for the whole project. In the case of modular projects, you can rely on overlapping interests to fill in the voids, but most projects aren't like that. Especially if all that the creator gets is the ever-diminishing value of a single copy.
I find myself compelled to advertise :-) http://www.iang.org/papers/task_market.html addreses this, and whilst the above comments are highly applicable, I believe there are ways forward. See also the work of Eric Hughes, John Walker, the AMIX, Robin Hanson and others.
... Why is nobody willing to guarantee kernel stability, even when paid big bucks?
Well, the problem is that you are asking too much of one OS. If you want stability, use FreeBSD (we do). If you want security, check out OpenBSD. If you want portability, try NetBSD. If you want to have fun, then Linux is good, so I hear, but serious business finds itself unsatisfied. Mind you, it is getting a whole lot better!
People do even "grand" things without expecting to be paid (or even worse, expecting to die from it), because they want to.
Well, what stupid people they are. I wouldn't go anywhere as far as gettimg myself killed for the common good. Even paying for software I can just copy is a stretch. What makes you think most people care enough to Do the Right Thing? What makes you think relying on Doing the Right Thing is a good idea? I mean, it's been tried before, and the consequences aren't worth a second look.
We've had a lot of success with open source. My company published the Cryptix java library back in '96, for purely selfish reasons. Since then, a team of volunteers (including our own people) have produced lots of versions, and kept it moderately up to date. (It's a bit quiet now, the lists are down, giving evidence of the unreliability of the open source model ;-) We had to write Cryptix, that was a business requirement as we needed crypto in Java (and Perl) and nobody had done it before. But, once done, I didn't want to pay to keep it going. So we open sourced it. We got the support and the updates for free, mostly, thereafter. I had to pay for bits and pieces, sure, but when I paid about $6000 (all up) to get the OpenPGP library written, it compared pretty well to the licence fees that RSADSI was charging, of $25,000. I got a good price because the source then got opensourced. Everybody won. And, I had to do it anyway, because RSADSI doesn't deliver OpenPGP :-( (Of course there are alternate business models. Baltimore, IAIK, RSADSI are some famous names that decided to sell their crypto software and made a bundle on the stock exchange. But, when it comes down to it, their model failed, because they were seduced into the apparent gold mine of PKI... But that's getting distracted!) There are other benefits to the open source model: most of the people who've volunteered have boosted their CVs and picked up good work because of it. -- iang