What I'm proposing is many individual blockchains, where blocks in any particular blockchain can me produced *only* by nodes with Internet bandwidth originating from a geography particular to that blockchain (for example, from a specific country). Within one country or even a part of a country, dedicated bandwidth prices converge above some price level.
Here's another potential problem though:
What is your proposal to handle the case of cloud service providers all concentrated in one region (e.g Northern Virginia) reselling VMs with gigabit speeds, since a bunch of users can sign up for them and then starve out everyone running a node locally?
You would have to divide the blockchain geography into very small regions which are at most the size of zip codes, just to keep all the big providers congregated together. Is this something you're willing to do?
Dedicated Internet Access at scale is cheaper than using bandwidth from any cloud service provider, isn't it? There would be no competition between customers of CSPs and people using DIAs.
When I said that Internet bandwidth used is a factor (let's call it Proof-of-Used-Bandwidth), I meant the Internet bandwidth used to send/receive data between all participants from a particular geography - I didn't mean that *any* Internet connection within that geography (like sending/receiving data within one data center) is a factor.
Everything is in the whitepaper and documentation on the website.