It would force the attackers to enter the jurisdiction of a state that will prosecute them if they're discovered, to carry out the attack, or else resort to much more difficult and slower methods (sneaker-net introduction of initial malware infections in the target state, say).

You don't have to enter a specific jurisdiction. There are supply chain attacks, escalation through residential connections, existing international botnets, and a thousand other approaches. And of course, there's always someone out there ready to open an email which will own them.

> There are supply chain attacks

Yes, some relatively slow, difficult, and expensive attacks would of course still be viable. That does not mean that, "it wouldn't change anything in practice."

> escalation through residential connections, existing international botnets

Right—so how are you going to talk to your botnet from outside the target sub-Internet when it won't even route packets you send it, except maybe to some hardened commerce-and-propaganda-only subnet that may have limited or no connection to the rest of the target state/bloc's Internet (and again, even that part existing is a maybe)?

We have a small experiment with that already - corporate NATs which are supposed to achieve the same thing. They still get owned. Even SWIFT which is basically as isolated as it gets in business had bad actors with access.

I don't believe a complete separation would ever be possible. We'd have to put banks on the commerce side. But that means every single business entity would also have to have access to them. But most businesses also need access to the other side and will not do perfect separation.

In malware research labs you can find rooms literally painted in two colours to separate the isolated part and prevent joining networks by accident. Normal companies do not care that much. Most single-person entities will just plug both ends into one computer and go on with their life.

Sure, but there wouldn't be a "bad" network to connect to.

Two things that are surprising me in this thread: 1) "there are ways to work around this, so it's completely useless" (yes, that's... any security), and 2) either not a lot of people are paying attention to future-Internet development, as in actual, technical development and research, or they're not seeing what I'm seeing in them, somehow, as, for example:

https://en.wikipedia.org/wiki/SCION_(Internet_architecture)

There will be plenty of home grown hackers within the country doing it then too. Sure that might make prosecution easier if we can catch them, but it's not like we have anything close to a good track record in solving most crimes and bringing the criminal to justice. It will just be another risk/reward trade off that domestics criminals make every day already.

“The internet interprets censorship as damage and routes around it”

Even if you physically firewalled every connection into a country all it takes is one little node connected via RF (satellite, HF, etc) dropped near an open WiFi hotspot.

The topology for the commercial internet today is not the same as the topology for the early internet which that quote is from. Today's internet is more of a hub and spoke model that is much more susceptible to damage if certain nodes are affected.

Wifi hotspot asks for personal or corporate/server ID of the sender of packets coming from this new node, since it can't route the traffic any farther without that. Gets nothing. Drops that node's packets as either hostile or malfunctioning, and, regardless, useless, since it can't route them anywhere. OK, so maybe you manage to steal an ID. See how this is making attacks harder? Now you're stealing or forging identities just to get any packets routed, and if you do anything suspicious-looking you'll rapidly get your stolen ID on the automatically-managed collective shit-list and it'll stop being very useful. Because the volume of attacks is so much lower, your drop-a-radio-near-a-hotspot trick might even trip enough flags to get someone to come find the device, if you use it very much—and if you can't use it much without "burning" the hardware, then, well, sure seems like it made your job as an attacker a lot harder, right?

There is nothing that guarantees the Internet will keep working the way it does now, and if an open Internet causes enough problems, it will be reigned in. How it works now is a choice, not a law of nature. I'm not happy about it, but that's just how it is. Either these kinds of attacks won't get much worse, or they'll get a lot worse and something like that will be what happens.

Wouldn't it be easier for the vulnerable beef packers and pipelines to simply disconnect from the public internet? What qualities do they have that would force them to burn down the world rather than fixing their shit?

I think between telling big businesses to spend a bunch of money and fix their shit and do things without the Internet, and adjusting the Internet—which would also make state-level attacks and foreign astroturf disinfo/propaganda campaigns harder to carry out—we're going to pick the latter. Again, unless these threats don't keep getting worse, though I expect they will.