Sites in Web 1.0 were connected via nothing more than links. Pre-Google search was based upon text matching. Outside of some chat groups and usenet discussions their was no real interactivity.
Web 2.0 was essentially the rise of Social web. From Wikipedia to Facebook the entire Web 2.0 Event was about creating a more social interactive Web. Today users are in the process of embedding social networks in to the web presence.
Web 3.0 brings 3 technologies that have been around for a while in to a more finished piece of UI: Social, Semantic, and Spatial.
Social, Semantic and Spatial have all been around for a while. Social is dominated by Facebook and Twitter, Semantic is Google and Spatial technology is MapQuest, Google Maps and Microsoft Virtual Earth.
Web 3.0 promises to bring these three ways of connecting technology together. For example a classic Web 3.0 application is Augmented Reality with a browser like Layar. Layar enables you to map a Google search of Twitter search on to a Google map and even insert the information in to a graphic view of your surrondings.
I have been amazed by the potential of such tools. They open up the ability to do Social Mapping or Spatial Mapping which allow us to interact with our environments in radically new ways. We could, for example, now have real time mapping of where all the "fun" is in our local environment, or the "money" is, or where people go to "find love". Many of the more metaphorical uses of the English language may soon have a iPhone App that allows you to use them.
As for Microsfot position in this field: it has none that I can see. This is not really too bad since new technology consistently never comes from the labs of big firms but from the evolutionary battle of start ups.