"There is a tiny little downside here: no local software. Browsers don’t yet do everything, and there are two decades of Windows applications that have been written, performing functions that can’t yet be replicated in a browser."

There are a few problems with that view.

One, the applications that matter to people are changing. For many people, programs like web browsers and iTunes are now opened frequently, whereas a decade ago those apps either didn't exist or weren't nearly as relevant. So who is to say that 2 decades of Windows software adds up to much these days?

Then there's the fact that many desktop apps were not written to coexist, so a legacy on Windows means something much different than on, say, Unix. Unlike Unix, where very old programs are still part of the ecosystem of utilities, Windows apps are just too random and fleeting.

Then there's the fact that desktop programs have grown and grown, and most users don't seem to adopt even a fraction of the features. What took two decades to develop will not take two decades to "replace", because only a fraction of the features will really be required for most people to make the jump.

And most of those huge apps should have been stopped at version 2.0. After awhile, big version numbers are meaningless; even for Apple, what on Earth makes the latest iTunes worthy of a version 8? They're still over-hyping the software, but not much is being added, so these old apps are not really as sophisticated as they appear to be.