I look at problems in the world as a series of solutions built from first principles. So if you want a distributed mesh network, you need a supplier that can make millions of boxes for roughly 1/10 their retail price, or you need software that can run on existing phones/routers/computers. Then you need open software with tests for all the edge cases so you know it's free of bugs. Then you need an algorithm for things like web of trust or onion routing. Then you need a compelling use case (easy, no more internet and/or cell phone bill) and a way to prove it's safe so it goes viral. And so on. No one step of the process is insurmountable, or in many cases, especially difficult.
Once we have that box, other ideas like content-addressable memory (things like IPFS) become feasible and we end up with an internet running thousands of times faster than what we have now, because media can be seeded once to each city and then cached by downloaders.
So of course that's a barrier to entry, because few players in the existing paradigm want to disrupt the status quo. But that's no reason to settle for the wildly expensive and slow technology that we are stuck with now. Pretty much every item in his list works along these same lines.