I suggest you read up something about the raise and fall of US automotive industry: Detroit used to be no less "startupish" than SV is today: thousands of engineering startups have flourished and vanished over relatively short period of time. The history of Detroit is one of several reasons why I don't take Paul's views on special role of SV seriously: Valley isn't something we haven't seen before. If anything, it is declining:
Valley startups no longer represent the cutting edge of computing innovation at large [as they did in the 80s], they have mostly escaped competition by hiding in relatively narrow niche of "English-speaking consumer Internet" very much like Detroit has been hiding behind big SUVs. And "release early, release often" web sites are no different from body-on-frame SUVs: their tech is so well understood and simple that Chinese can produce cheaper and faster copycats of facebooks and twitters in a matter of months within an original launch.
When they'll learn to market them properly, they will build their own data centers in San Antonio, TX and the similarity with Detroit will be complete.
Detroit became entirely dependent on a centralized system powered by a few massive companies. Silicon Valley has some of that, but for the most part is highly distributed. The thousands of startups that feed off the big auto companies can't exist without them. Silicon Valley doesn't die if Google and Yahoo fail to innovate.
That English speaking "niche" currently represents the vast majority the potential revenue. It's a lot harder to innovate than it is to internationalize your software, which virtually every successful company does.
The idea that the value of Twitter or Facebook is in the 5,000 lines of Ruby or PHP that it took to create them is seriously mistaken. Anyone (Chinese or not) can easily clone Twiter or Facebook. Just like almost any competent artist can clone a great painting. That's not the hard part.