Discussion of Open Source Inevitability
Could one perhaps argue that open-source has succeeded by making it cheaper for smaller companies and individuals to build things themselves? In other words, its most disruptive impact has been to democratize access to the means of production (web server infrastructure, scripting languages, compilers like GCC, etc.). It flourishes in distributed networks and among people who need a great, free digital toolbox.
I certainly do not think the success of open source is inevitable, at least not in all tech markets. It has not succeeded with end-users of personal computers and other consumer and enterprise devices. Obviously, you can't build a proprietary software bundle on top of shared GPL libraries. Thus, on Mac OS X you get gargantuan application packages because each vendor has to reinvent the wheel, packaging all their libraries inside the bundle. (So, in addition to getting expensive applications on OS X, you get bloat.)
When people need a "magical" box, the internals of which are kept hidden from the user, open source has struggled. And owners of intellectual property are understandably suspicious of open source, usually choosing closed platforms (witness Amazon's proprietary ebook format or the recent excitement of publishers about the possibility of using iPad apps to charge for content). In these instances, various open source internals are often hidden beneath a closed userspace (witness Netflix, which serves streaming movies from Linux servers and has provided a silverlight plugin for the Linux-based Roku player, but not for other Linux users.)
In other words, the Linux desktop is in a state of perpetual dysfunction, but the open digital toolbox that is GNU/Linux is one of the wonders of the digital world.
-- madalu
In no particular order:
I don't think static compilation (the bundling thing) is actually a huge problem for OS X. They decided that it was better that every application be able to run and contain everything that it needed to run in a single bundle, than it was to: save a little bit of RAM, or save a little bit of disk space. And its easier to statically compile things than it is to get package management right. In all I think it's a fair trade off.
The inevitability that I'm writing to here is basically: given enough time, open source alternatives will become so good and so comprehensive that it will become very difficult
I'm not particularly convinced that in the long run "Free Software" is actually a lot cheaper. You don't have to pay licensing costs, but it's not unreasonable to expect that you'd pay support costs, or that you'd need to have someone in house to administer, develop, and customize the open source product to the organization's needs. I mean, open source technology might come out a little bit cheaper, but that's probably at the accountant stage and not at the decision making stage. Having said that, I think the economic model where you use a piece of free software, and hire one of the developers to do work for your seems much more sustainable and desirable than the whole licensing-based model.
I don't think that the Licensing-based businesses and the proprietary source-code method of software development/creation is going to go anywhere, and I think that there are enough cases where institutional practices haven't caught up to open source models that there's a medium-term future there (at least for enterprises.) But there's a difference between work with existing products and work in existing ecosystems (divergence here about the Apple App Store) and greenfield projects that I think requires a little but more attention.
It's not even so much the "Linux Desktop," as it free/open source software for desktops regardless of platform. I'm counting Firefox and Libre/Open Office for Windows and OS X in this.
-- tychoish