There has been some debate over the success or failures of Linux on the desktop, mainframe, and in the server market. Linux has finally started to gain some traction, especially in the server market. As for the success of Linux on the desktop, there are all sorts of blogs on the topic.
Some blogs say that Linux is not ready, and some say otherwise. Some arguments are old, some new, and some are relevant, and some aren't. There are gorgeous Linux desktops and then there is the command line. Although the Linux desktop becomes more and more user friendly, it hasn't taken hold of the market the same way OS X has, and certainly not how Windows XP has.
It seems that Windows is successful and Linux isn't. However, I say that this is a mistake. To say this uses install base as the metric for success is to be ignorant on the goals of both systems. They both have achieved the goals of their authors.
The goal of GNU is to free one from usage of proprietary software, and the goal of Windows is to make money. Windows is now so strong that Microsoft can discontinue a successful product to force everyone one to purchase a lesser one. Despite the strong shortcomings of Windows Vista and the sheer success of XP, Microsoft has the audacity to push Vista.
The strength of Microsoft is really in that they have enterprise installations. People might not try to learn a new operating system on their own, but for a job they will. Since Microsoft can sell their product to pretty much any business, workers will be trained for their product. This causes people to purchase Windows for their home computers, something neither Linux nor OS X enjoy.
Linux was developed by a bunch of programmers who wanted to make a free operating system. Generally, programmers are geeks, and will take time just to learn something new. When they talk about technology, what they really say is, "Look at how neat this is." That's why there's so much confusions between geeks and everyone else. Geeks don't realize that people don't value the same things they do, and those other people need to understand that they might actually know something, they don't. (In fact, I use the word 'geek' as a way to relate to those readers who don't know much about the Open Source communities, the real term to be used is 'hacker,' but most people understand, "Evil person who causes all sorts of chaos electronically.")
The geeks succeeded.
GNU/Linux plays music and video, downloads, uploads, plays video games and Windows applications (really, this the only place where an application written for entirely different platform will work), email, browsing, and anything else you'd want to do with a Windows system. One is not forced to use Windows anymore, unless Wine cannot run that one or two vital applications. It's initiative that's holding many people back still, "Oh, the learning curve!" (That's Linux's greatest weakness--it requires initiative).
Now Windows is successful in that it earns Microsoft tons of money, and there's plenty of room to enforce this statement to dismiss even a ridiculous doubt. Linux is free and works.
Usually, when someone talks about success, they usually the context implies which is winning, and that's Windows, although Linux is beginning to gain some traction with the Asus EeePC, Dell Open Source line, Cononical's Ubuntu, and all the little upshot niche markets selling computers preloaded with Linux (usually Ubuntu). So, Linux may kill Windows, it may dominate, and it may be catching up.
Both Windows and Linux are successful because they both have accomplished their respective goals. Windows is winning, although that might not be the case in the future. If one wants to decided which is successful, one must examine the goals because software isn't written just because the author wants to make money. To assume so is either naivety or ignorance.