I had some vague interest some time back in some of this some time back, the idea of a “zero-admin” network where you could just have random people plug in more infrastructure, install some software package on nodes, and routing and all would just work. No human involvement beyond plugging physical transport in.

Some things to consider:

• People will, given the opportunity, use network infrastructure as a DDoS vector. You need to be strong against that.

• It’s a good bet that not everyone in the system can be trusted.

• Not only that, but bad actors can collude.

• Because transport of data has value, if this is free, you have to worry about someone else who provides transport for existing data just routing stuff over your free system and flooding it.

• If the system requires encryption to mitigate some of the above issues (so, for example, one sort of mechanism might be a credit-based system where one entity can prove that it has routed some amount of data from A to B in exchange for someone else routing some amount of data from C to D – Mojo Nation, the project Bram Cohen did before BitTorrent, used such a system to “pay” for bandwidth), that’s going to add overhead.

• If you want your network to extend to routing data onto the Internet, that’s going to consume Internet resources. Even if you can figure out a way to set up a neighborhood network, the people who, for example, run and maintain submarine cables are not going to want to do that gratis. And yeah, to some degree, you can just unload costs onto other users, the way that it’s common for heavy BitTorrent users to pay the same monthly rate as that little old lady who just checks her email, even though said heavy users are tying up a lot more time on the line. But if you are successful, at some point, this stops flying below the radar and ISPs start noticing that User X is incurring a greatly disproportionate degree of resource usage. I should note that there are probably valid use cases that don’t extend to routing data onto the Internet, but if you don’t permit for that, that’s a very substantial constraint.

If anyone has to do something that they don’t want to do (e.g. run line from saturated point A to saturated point B), then you’re potentially looking at having to pay someone to do something, and then you’re just back to the existing commercial Internet system…which for most people, isn’t that expensive and does a reasonable job of moving data from Point A to Point B.

From a physical standpoint, while different parts of the network can probably use different types of infrastructure, if you want sparse, cheap-to-deploy infrastructure over an area, my guess is that in many cases line-of-sight laser networks are probably your best bet, especially in cities. You can move data from point A to point B quickly through other people’s airspace without paying for it, today. Laser links come with some drawbacks: weather and such will disrupt them to some degree, so you have to be willing to accept that.

The main application that I could think of for regional-only transport, avoiding routing onto the Internet, was some kind of distributed backup system. A lot of people have unused storage capacity. You can use redundant distributed data storage, the way Hyphanet does. You can make systems that permit one user to prove that they are storing a certain amount of data to let them build credibility by requesting hashes of data that they say that they’re storing. It won’t deal with, say, a fire burning down the whole area, but for a lot of people, basically having some kind of “I store your offsite data using my unused storage capacity in exchange for you doing the same for me, and we can both benefit enough to want to continue use of the system” system might be worthwhile. That’s also likely to permit for higher-latency stuff involving encryption and dealing with redundancy. I think that “Internet service for free” off such a system is going to be a lot harder.