Happy Birthday Tux
Linux was at first very poor. That's to be expected. In 17 years it has improved some.
Although we call it Linux in most uses the GNU software does play more of a part and should be recognized, as well as the contributions of UCSD and contributors under their BSD license. The terms Operating System and Operating Environment are fairly nebulous even to this day. Still, we need an easy handle to hold it by, and "Linux" will do.
In the beginning it ran on one system with one processor (Linus's). Now it runs on 85% of the top 500 supercomputers, my wireless router and phones and many platforms in between (and beyond!). Hardware support used to be poor. Now Linux supports more hardware than any other system ever offered, period. This has been true for a long time.
Linux conquered the server room first, as is natural. The server room is a hot environment where professionals are classically educated and use real metrics to determine what works and what does not. Technologies are born and live or die in the server room before most of us ever see them. As the World Wide Web grew the power of Linux became obvious to the denizens of server rooms and it matured just in time to be adopted by many of them. In many cases the tech boom of the late '90s was a Linux boom.
Once upon a time it wasn't useful on laptops, but now it's on half of the top selling laptops on a major vendor, Amazon.com. Major vendors used to shun it, in deference to their major partner Microsoft. Now all major vendors offer it preinstalled, some in varying flavors. For a long time Linux was not pretty. Now it's not hard to convince people that a two year old installation of Linux is actually the "next" version of the dominant desktop OS. Linux is now the Belle of the Ball.
Software installation on Linux could still use some work. It's not as bad as Windows. The add/remove programs feature in Windows doesn't actually Add programs, except in certain rare circumstances. The add programs utilities in most Linux distributions links to repositories of software so deep they had to include an internal search engine. Still, the installers in most Linux distributions should also reference a "local" and "Local Area Network" repositories by default, so that applications could add themselves to these repositories as part of their standard installation process. They'll figure this out soon.
The Linux of today embraces the classical scientific rigor of a bygone era, the splashy interfaces marketing thrives on, and the brutal Darwinian winnowing of our current IT environment. It survives and it looks good doing it by doing what it does well.
Most importantly, the direction of the knife has shifted. The cutting edge isn't on optimum performance processing any more. We've seen the trap that that is finally - "Intel giveth, and Microsoft taketh away." Now the huge growth is thin-is-in low wattage desktops and laptops that somehow have gained 20% of the market share in just a year. Here the low overhead requirements of Linux are well positioned to exploit the growth of emerging markets abroad (the third world) and emerging markets at home (nettops, netbooks and thin clients). I believe this is by design -- many of the new generation of developers live in growth markets where building apps that require the juice sucking high wattage platforms of yesteryear is not targeting their optimum market because they just don't have the watts and building out the watts is just not going to happen. I also believe the major vendor is suffering from inertia and failed to make the turn here. In the challenging economic environment before us the power of "free" will gain considerable leverage for the next decade even at home.
There are major applications still which are preventing people from adopting Linux. Photoshop is a major example. As soon as Adobe realizes that their platform partner will not stop trying to kill them with their Adobe Flash replacement Silverlight, they may begin to see the wisdom in diversifying their platforms. Even if they don't, they're a weak anchor. Game vendors follow other platforms too, but their loyalty is transient anyway. The brutal evolution of games is such that the ones that choose the wrong platform die and the life cycle of a game company is about five years.
Linux is only a tiny fraction of the free software revolution, but it can bear the standard for the millions of developers and thousands of projects that march us toward the future. The use, reuse, and improvement of open projects increases the utility for all. For each little bit you put in, you get a million back. This is progress.
Thanks Tux!