You are right that market share is important, but that's the last thing you are right about. In your characterization of the three platforms, you are not remotely evenhanded. So to balance your view, I offer my own:
1. A developer can join the legions of those who create products under the Windows platform -- and unless he defines a whole new genre of software unto itself -- he can look forward to utter anonymity by offering one product among hundreds or thousands of competing products.
2. A developer can join the Linux geek community and code server applets, shell utilities, or scripts in Python, Perl, Ruby, and PHP and the like -- and if he never wants a product to get into the hands of an actual CONSUMER -- he can stay in his geek heaven.
3. A developer can join the ranks of those who develop for the Macintosh -- a platform which combines the best of all worlds: (1) a consumer desktop platform with an established and dedicated installed base, (2) Unix-based underpinnings for the hacker in his soul with robust support for Java, (3) the education market where -- despite some declines -- Apple remains very strong, (4) power users, and (5) loyal customers in the Mac's strong niches of music, graphics, video, and prepress. In this environment a developer can truly distinguish himself without getting lost amid thousands of other competitors. Here a developer can reach multiple markets -- home consumer, home office, small business, education, and corporate -- as well as focusing on both client and server applications.
See how easy it is to skew the landscape? Anyone can tilt things to his own point of view. I normally wouldn't describe a developer's options in such biased terms as I have done above, but I do so here for purposes of making a larger point.