Software pricing models

I was looking for knowledge bits on pricing and ran into a post on John Clingan's blog: The Clingan Zone. John works at Sun Microsystems, and is one of their evangelists. He also points out their shortcomings, so he's one of the good guys... :-) The question was: Are there any new insights to price new software? Here are some of my thoughts I haven't dabbled with software pricing, but have worked as a programmer (on Unix and Windows), I did have an experience that I'd like to tell you about. Normally, most software vendors assume that the useful lifetime of their software is around 3 years, seen that most people also upgrade their hardware in 5 years, and if it's leased, they upgrade in 3 years. This may not be the case! I visited a client who was running Windows 2000 on some machines and Windows NT on some others. Instinctively, I told her to upgrade - to a non windows platform of course. But on the long ride home, I got to thinking. In a lot of places, especially India, the price of a license of a windows operating system of the "previous generation" which is still supported falls drastically. So when the Windows 2000 system costed x amount, the cost of a Windows NT was x/5! Given that you can find a lot of people who can administer such a system well, their cost falls as well. So upgrading to the older operating system becomes incredibly financially attractive - for people who use it for normal office type applications AT WORK. (Most folks who use it at home in India pirate the latest version.. no question of that.) The second thing was - if you compared feature to feature on a pretty high level - Wordprocessing on a previous platform versus Wordprocessing on a new platform, there isn't that much value addition for a basic user. And given that economic advantage of older platforms, the cost to consumer of a new platform actually increases. So here's my notion - Make newer releases of platform software cheaper (for some customer segments) than the original ones. Stick the higher value add-ons in a separate bundle, which can be purchased if someone buys utility computing or support contracts or so. High value applications should never be introduced in a suite, since a suite automatically becomes a platform. They should be on the utility computing / Application service provider model. Once you find that the rate of subscriber growth on that service has flattened, and has stayed there for awhile, move the service to the suite. Could this perhaps be a case for software companies like Oracle - not to acquire smaller companies?