Thank you! You explained it really well.

You’re welcome!

As I understand it, most of the physical infrastructure for data transmission belongs to some private company. These companies sell the usage rights to smaller companies and so on. So these companies are the ISPs.

Usually they are ISPs. If you wanted to buy a route on a fiber optic cable (usually called “dark fiber” in this context), it’s just a different product that they sell. There’s no Internet access on a dark fiber link, not even light from a router at the other side. If you were to purchase a dark fiber route, you’d have to provide routers at either end and light that path yourself.

Whomever owns those ISPs also owns that infrastructure. In liberal countries, it’s usually a for-profit company, but could be a non-profit as well. In less liberal countries, the government may own and operate the fiber.

In this sense, you would either have to build your own network infrastructure (cables, routers, switches) or rent the infrastructure from the owners.

Correct.

The physical infrastructure is basically largely independent of the protocols that run through it (optical cables simply carry light). Couldn’t this infrastructure then somehow be used as a direct connection between two users via a protocol other than IP?

Absolutely. CLNS, IPX, Appletalk, DEC Phase V, and Banyan VINES are all older examples of layer 3 protocols that work perfectly fine over fiber. Once upon a time, IPX was more dominant than IP in businesses. I ran a dual-stack IPX and IP network 25 years ago.

These other layer 3 protocols do not interoperate with IP. For example, on my IPX and IP network, I had computers that could only speak IPX. None of those were able to communicate with IP networks at all. They did not know anything about the Internet, could not use it, could not access it. A web browser on these machines would simply not work.

That’s true for routers as well. If I were to order an Internet circuit, and I enabled IPX on my router, IPX would not work across that link because the ISP doesn’t support IPX. Both ends must agree to route the layer 3 protocol.

But IP won out in the long run. Not because it was government or corporate owned, not because it was centralized, not because it was engineered to be unfair. It won out because it wasn’t any of these things.

The standards for Internet protocols are openly developed by a body known as the Internet Engineering Task Force (IETF). Anyone is free to look at the specs, known as RFCs; anyone is free to write their own implementations of the protocols. This includes IP, TCP, UDP, BGP, DNS, HTTP, HTTPS, and other helper protocols like DHCP and ARP.

That’s why there is a such a diverse, worldwide marketplace of router hardware/software vendors. Each of these vendors have developers that used the RFCs to write their particular implementation of router software.

Going back to the question of neutrality, the RFCs themselves neither encourage nor discourage neutrality. They exist as an agreement on how to exchange information: how packets are constructed, what information is included, how software ought to interpret that info, and so on. It is up to the router vendors to implement these standards, and it is up to the ISPs who buy those routers how to handle those packets.

The ISPs may enforce policies that are handed down from the jurisdiction where they operate. The ISPs may even be owned and operated by the government. China’s Great Firewall is a prime example of anti-neutrality. Iranian Internet, I understand, is also government run and heavily monitored. These providers will use information in the IP packets to throttle or even block that communication.

In more liberal countries, most neutrality proponents are concerned primarily with quality of service. For example, they argue that Comcast Universal should not throttle or de-prioritize traffic to Netflix because it is a competitor to Comcast’s own streaming service.

In either case, the protocol spec itself is not the reason for anti-neutrality behavior by these organizations. And switching to another layer 3 protocol, whether it’s one that already exists or has yet to be invented, won’t facilitate that neutrality by itself.