The significant thing about Chrome is that it sets a new way of thinking. It does not mean Chrome will dominate the world. Open standards mean that other companies could provide similar services. It's the 80% scenario. 80% of what we do could be web based and probably will be in the future. It is near 100% for 80% of the population. It does not then make much sense to have everyone running a desktop OS just in case they might happen to want a specialist application that is dependent on that technology. Some people will still need this, but not the majority.
I see the future in the merge of mobile phone technologies and web applications for the vast majority of productivity applications. Mobile phones will move up into the netbook and laptop space by having options to connect to larger screens and keyboards. These phones have operating systems but no-one bothers too much about which operating system. Chrome OS might or might not have a significant role to play in that but one things is sure, the need for Windows, GNU/Linux or MacOS on desktops and laptops is going to diminish. They might take a very long time to disappear altogether just as mainframe computers still exist. The dominant market driver of technology is changing and that is what is really interesting.
Timescales are of great interest because its more a "how long will it take?" than "will it happen"?. It's not a matter of if we move from desktop to web -- it's already past the tipping point; it's how the inertia of billions invested in desktop systems will slow down change.
Take the government's £500 million spent on Curriculum On-Line. Schools got this money ring-fenced to buy education software licenses. The idea was to stimulate software development. Virtually all the software bought with this money could be web based. So why isn't it? Because the proprietary software vendors found it less expensive to just re-sell old products and there was little real incentive to do genuinely new and innovative stuff. Where there are web based applications they are not paid for in perpetuity so the £500 million under-estimates the real cost. The effect of this is to entrench the desktop paradigm. It gets worse. There are about 10,000 hours in the National Curriculum across all the subjects for ages 5-16. Divide 500 million by 10,000. That is £50,000 for every hour - approximately every lesson - in the National Curriculum. If I went to a web developer with a specification for a lesson and gave them £5000 to support it with web based text, graphics animations and links to relevant information, for a tenth of the money spent, we would have free and co-ordinated support for the entire curriculum accessible directly at home and at school. Under a CCSA license all that resource could be shared, improved and developed Web 2.0 style.
So what does this tell us, apart from the fact that political rhetoric about wise spending of tax payers money is not to be trusted? In the longer term, water flows down hill and despite the government spending huge amounts of money in order to hold back progress, there will be mass migration to the web. Not only is it less expensive to develop for the web, it's also massively less expensive in maintenance. We are only at the very beginning and no doubt more expensive mistakes will be made by well-meaning civil servants. Bandwidth will continue to increase, latency reduce. Access technologies will shift to low cost rentals. The irony is that developing countries that have not wasted billions on desktop infrastructure could now go straight to mobile technologies and web applications bypassing the cess-pit of desktop complexity. Sure, some specialists will still use applications that need desktop computers but these will increasingly be niches, and who is to say that in the longer term even these will not move to the web? It's all down to the timescale, in the end water flows down hill.
