Inbetween

It seems to me that as we pass into the era of web 2.0, the software platform as we know it today will cease to have significant commercial value.

The principal reason being that the internet and web2.0 is allowing a move from code sharing to instance sharing for software platforms causing the existing network effect mechanism for platforms to fail.

The good news is we can expect new platform models to emerge based on the properties of sharing a single, persistent online instance rather than code sharing and multiple isolated instances (e.g. windows).

Some companies have already hooked into some aspects of this new model. eBay and Amazon as platforms have it, Google as a platform does not. I am of course talking about the architecture of participation becoming the principle network effect mechanisms for web2.0 platforms. That is, if the actions of the users contribute to the shared state of the platform (through which every platform application they may use) in such a way as to enhance the experience of other users,  then their is a strong network effect based upon participation.

It is important to note that the forces enabling this new model are also undoing the previous model.

Here's why (IMHO).
 
Platforms
Over the past 10-15 years, Microsoft demonstrated both the enormous intrinsic and commercial value of software platforms.  We have seen this battle for control of the platform played out over many segments of the software industry and layers in the software stack (Oracle, Syabase at the DB layers, Windows OS/2, IBM Websphere  vs BEA Logic for application servers, smybian etc).

The return on capital invested simply dwarfed other software models and so platform leadership passed into law for many software companies as the one truth strategy for growth. The amazing value creation being principaly driven by two forces; reuse and the network effect.

Reuse: every application built on a platform is saved from having to make the investment to build  features that the platform provides. This massively lowers the cost of production (therefore capital invested) for application developers.

But software libraries and software components do this also, but are not considered platforms. The difference between software platform and software library is the network effect or ecosystem.

Network Effect: Each application built for a particular platform increases the value of having that platform and, by extension, every other application that already uses the platform. So the more applications for a platform, the greater the value of the platform.  So for the owner of the platform, that has a model that can extract commercial value from the platforms massive intrinsic value; the return on capital is a function of the investment that OTHER people have made. Or put another way,  they achieve a return on capital NOT invested by them. Pretty sweet.

But the real questions should be "What causes the network effect in platforms". What is the mechanism by which the investment of application developer A has increased the value of the platform and of application B.

Does that same mechanism hold in the world of web2.0???? My believe is NO it doesn't. And that will have a profound effect on the strategy of software companies over the next 10 years. In fact we are already seeing it.

Traditional cause of the platform network effect

Was the dependency on the users to have purchased and installed the platform in order to use the applications.
Choice of purchase defined which applications you could use, naturally the platform with the better range and quality of software is more valuable (just like in the games console industry).

Web 2.0 removes the need for user purchase of the platform
As functionality moves off the users machine into a standards based cloud, the user choice of application platform effectively disappears.  By definition web 2.o platforms API is web based and implementation neutral.
Consider the Google search APIs. If there is one or 100 applications based on it, the value of the platform is not much enhanced, those applications do not add anything to each other, no network effect. From an ecosystem point of view Web2.0 APIs are much more like software libraries than platforms.

Web 2.0 platform network effect

But web2.0 platforms have a new trick that traditional platforms don't have. They can easily present one shared instance i.e. state to all the different users of all the different applications. This allows the actions of one user using application A of the platform to enhance the experience of another user using application B of the platform. This is the architecture of participation. It is easy to see how both eBay and Amazon increase the power of their content based network effect through open access APIs. It is also easy to see how this doesn't work for Google, the end user of a search app typically can't affect the shared state of the platform.

Open Source Network Effect
Developer still need traditional software platforms though.
So web2.0 platforms allow sharing of the state of the platform. Where as traditional platforms allowed read only sharing of the code.
There is a way that traditional platforms can drive a network effect by allowing participation in the shared code of the platform. They can let users contribute to the code. This can immediately drive a whole new network effect which hugely increases the intrinsic value of the platform. Unfortunately for the existing platform vendors, nobody wants to submit code that somebody else will make money off.  So this can only be done through open source. Linux is hugely valuable, but nobody can make $billions of its commercial sale, at least not directly.
Interestingly, as more open source code is created, it becomes easier to remix the code and create yet more open source software. The more general the software the more valuable for it to have an open source incarnation i.e. platforms are the natural place for opensource to target as we have seen with Linux, MySQL, JBOSS etc.

So for all the reasons above, I am pretty sure that as web2.0 progresses, we will see the rise of a different type of platform and the existing platform players will have a very hard time in holdings onto any serious returns.

Long live the platform of participation.