I imagine a solid base starting with the right kind of "knowledge". From what I observed, and this is very subjective, but people who clump around MS technologies tend to produce lousy communities where not much innovation happens.

I would start with attempting to produce (or translate) good material that has the potential to produce programmers who understand their craft really well and have a really good understanding about systems (to the extent that they can manage a web server, for example, and ssh into it, and edit files with vim, and all that stuff).

Maybe also having a few projects that people can clump around, something that's open source, interesting enough for people to want to use it and contribute to it.

Again I don't really want to clone Silicon Valley per se, what I want to clone is online open source communities. And again, it's not really about cloning, but about using that model as something to aim for, at least initially.