How will this work with chat control?
> "If Ofcom doesn't think this will be enough to prevent significant harm, it can even ask that ISPs be ordered to block UK access."
If you want to enforce stupid laws the burden should be upon you.
which might be the end goal - the internet, with freedom of communication, is a way that the plebs can organize and resist authoritarianism. And as countries are growing increasingly authoritarian (and i include UK here), they may be planning on preventing the old free internet that has enabled so much.
So as technologists here at HN, there needs to be a pre-emptive strike to prevent such an outcome from becoming successful. I would have said TOR, but for most people it's a non-starter. What other options are there?
> And how we burned in the camps later, thinking: What would things have been like if every Security operative, when he went out at night to make an arrest, had been uncertain whether he would return alive and had to say good-bye to his family? Or if, during periods of mass arrests, as for example in Leningrad, when they arrested a quarter of the entire city, people had not simply sat there in their lairs, paling with terror at every bang of the downstairs door and at every step on the staircase, but had understood they had nothing left to lose and had boldly set up in the downstairs hall an ambush of half a dozen people with axes, hammers, pokers, or whatever else was at hand?... The Organs would very quickly have suffered a shortage of officers and transport and, notwithstanding all of Stalin's thirst, the cursed machine would have ground to a halt! If...if...We didn't love freedom enough. And even more – we had no awareness of the real situation.... We purely and simply deserved everything that happened afterward.
The adoption speed is critical, exactly because of what you're saying. It's easy for a wannabe authoritarian to make a decision to "just block all of ECH and QUIC traffic" if that breaks 0.8% of all traffic - but not if that breaks 80% of all traffic.
In such a "splinternet" scenario, it'd be a matter of setting up PTP links across borders. As long as a few people do so, it becomes one big network again.
It should be noted that the Online Safety Act is in fact not international, but UK-only.
We'll have 2 kinds of apps and websites.
One will be super nice products that only work in your country and you can't use it to communicate with outside people.
The other kind will work worldwide but because they would be spending so much more on compliance their product would be a bare minimum ad riddled crap.
Server could have multiple QUIC output nodes to migrate connection in case one of them is blocked. The output node network can be shared by many servers and DoQ endpoints so blocking it entirely would scare government.
This solution still needs to connect to some known IP in order to establish connection first. And the same goes for DoQ. To mitigate this we can use Encrypted Client Hello as other commenter mentioned and connect to a pool instead of single IP.
The option here is to stop trying to solve everything with tech when a lot of the time it's not viable and actively makes things worse. Start putting that time into the non-tech options. Not as fun though, is it?
Not for the masses and not sustainabl,
It's always easier to have a paper say "do this" than finding a tech to circumvent it.
Politics is fundamentally people business and involves lots of people who can't or won't understand the details of what is going on but who may still be interested in the end results.
Chat control (which isn't (yet) a thing) would not in fact lead to the outcome you describe.
Any company would be forced to comply or get the boot from EU market. Apple and Google will happily enforce that and that's probably good enough initially.
US Vendors could also decide to create an EU only version of their services.
So the lack of ability to solve this politically has made technological solution the only out.
The problem is when tech people try apply tech to political problems crudely, without understanding or without caring about the human aspect of it. You need sociologists and political scientists to study what impact a technology will actually have, and normal people to see how they feel about it, not just programmers who may incorrectly assume that e.g. designing an open and secure protocol will automatically and directly map to creating an open and secure society.
For example, in this case, the blunt approach is "How do we design a protocol that can't be censored/monitored?" The answer is TOR, which as parent comment noted, is socially a non-starter. But maybe a better approach could be, "How do we design a protocol which removes the incentives/makes it politically untenable for people to censor/monitor it?"
One way you might approach this is to create a system that's organically useless for bad actors. Clearly different platforms have different levels of "safe" and "awful", due to their structure. Could we design a platform with such strong prosocial incentives that authoritarians are not able to fearmonger about it?
Another approach could be to chain common citizen rights to authoritarian interests. For example, the US government cannot backdoor AES, because doing so would also compromise their own communciations. Can we make it so authoritarians are forced onto the same boat as us for our other communication technologies too, and therefore disincentivized from weakening our privacy because doing so would damage theirs too?
ActivityPub, ATProto, and blockchain could also be seen as technologies that are designed to create a social structure that incentivizes specific political outcomes, with varying degrees of success.
It's people business. So you design around questions like "Where is this technology going to put different types of people, and how are they going to feel about that?"