Certainly the Mac is making progress against Windows. People are happier with the less fiddly OS, they care less about OS compatibility than businesses does. Windows is still like a 1970’s car, for enthusiasts that like to fiddle around under the hood. User friendly OS will gain most traction in the home environment and now almost every family has more than one computer those are big numbers. So I agree that OS share is likely to spread. Consequently cross platform runtimes will become more popular, but they are nothing new, basic interpreters have been around as long as many people working in this industry have lived.
Conceptually the CLR has an edge over the VM. Flavours of CLR will eventually dominate, because they allow specialisation of programming using different languages and that is necessary to bring down development time and costs. Importantly those CLR should also support mobile devices to reduce the costs of developing itinerant computational behaviour. Users have hosts, they don’t want to know some computational behaviour only works on this or that bit or kit, or that they can’t access it here or there. A nice advance I am looking for from a dev environment is its ability to take CLR targeted code and reframe it using a chosen language. That will really accelerate development. As for browsers, they offer a subset of the functionality of a CLR and will be subsumed.

Great article, but a few comments. If the computer desktop is the new TV, then Google is the most popular channel, and Google's search results is the programming.

I think it's impossible to predict the future. And it's very possible computer desktops will be eclipsed by some technology we can't imagine today. Who would've thought 20 years ago, we could be using line of business apps on our telephones? But that's exactly what's happening today with the iPhone and Windows Mobile 6.1.

Regarding your question on which platform is best to bet the future, the answer is simple: The ones that are built on open standards. In 2008, that means HTML, CSS, JavaScript, XML, SQL, and most importantly, IP and HTTP. Adobe and Microsoft's RIA platforms are beautiful, but in my view they're largely, if not totally, proprietary. I'm sure that they'll achieve some level of critical mass, but I don't see them becoming dominant because no IT shop is going to commit to these platforms when open standards are available.

When the company I work for (Alpha Software, a Web application database tools vendor) recently had to make this exact decision and bet on the future that our development tools should target, we chose AJAX. Given the reaction of our customer base, we made the right call.

what are "Google Awards"? :-)