The problems of San Francisco's status as overcrowded tech hub could all be solved if only everyone could decide where the second Silicon Valley was going to be placed. If everyone woke up tomorrow with the belief that in five years some other city was going to be as big a tech hub as San Francisco, then it would happen. Everyone would rush to set up shop there -- startups, VCs, and the random other bits of necessary infrastructure like cat cafes and brunch places where you can queue for an hour and a half. But instead there's no consensus on where the second SF is, so there's defocused overflow in all directions -- to Portland, to Austin, to Seattle, to Boston, and so forth.
Anyway, if you want to build a nerd city in the middle of nowhere, you're somehow going to have to persuade half a million people to move there, and that takes more than some airy-fairy promises about how awesome it's going to be. Cities in real life usually grow slowly and organically over a period of centuries. When they grow fast it's because there's a compelling economic reason (e.g. a gold rush). You can't just make a city out of nothing; even if those half a million people would be happier if they all moved there together, the Nash equilibrium is for them all to individually not move there.
Burning Man is an example where people just decide to pickup and go somewhere, but that's not a realistic model. Nor is starting a city off at 500k population. You are right, cities have to start small and build incrementally by providing value. Like any good system :-)
Any new city has to offer what all the old cities offered: opportunity and a chance at a better life. If they can't do that then they won't flourish. If these benefits can be offered then history shows people will move, especially as home ownership has dipped in recent years.
The idea is that we have an opportunity to build around new resources and new ideas in the same way we built around gold or shipping in the past.
Only gold has been spun into digital.