Guest post by Rabee Tourky
Is there an economic argument for the present market structure in the market for desktop operating systems? Operating systems are programs that tell other programs how to run your computer and interact with the user. There are many operating systems. The famous ones include MS Windows XP, SP2, the various derivatives of the GNU system, like GNU/Linux, systems based on BSD UNIX, like FreeBSD, its good looking but dull child OS X, and its paranoid sibling openBSD.
The present market structure is basically a single operating system that almost completely dominates this market (MS Windows). It is not the best operating system but it is the most popular. It is clear, that if we all changed our operating system, to say openBSD, we would all be better off since it is likely to reduce the cost incurred by viruses and other malicious programs that take advantage of the faulty designs in MS Windows.
Now there are tremendous barriers to entry into this market.
a) Barriers to entry related to getting hardware manufacturers to design drivers for their hardware. Drivers are programs that tell the operating system how to let other programs access the hardware. Anyone who has used FreeBSD will know that it sometimes takes several days to figure out how to get FreeBSD to talk to your Wifi hardware.
b) Getting software manufacturers to write programs for the operating system. This is costly and in many cases is difficult to do. The main cost I think is in keeping two versions of the program up to date and debugging the two programs. For reasons that I don’t fully understand companies that write games that use a lot of graphics are reluctant to write these games for say OSX or FreeBSD or the famous GNU derivatives.
c) These barriers to entry are amplified by network effects arising from the benefits that users get from having many other users use the same operating systems. If one important communications program, like Skype, is not cross platform, then the people using the program would loose a lot if they switched operating systems.
There is also something that is nagging me and that I don’t fully understand. Operating systems are mainly a communications protocol for program writers to use. They are not so much the user experience. They are like the convention of driving on the left or right hand side of the road. I think that there is an argument for emergence of an international standards body in this regard for desktop systems. One family of standards is POSIX, which has been hugely successful in the UNIX (mainly server) end of the market.