Just look above here, people still want to do cert reneval manually with long timespans. The most radical thing LE did was not to provide free certs, or even automated certs, but short-lived certs.
The short lifespans are to encourage people to actually automate, not just say they want to.
If the cert lifetime is 5 years, Bill installs "just the first one" manually, solemnly swearing they will automate next time, and in five years nobody even knows who Bill was, Bill left three years ago. How do we replace the certificate?
If the cert lifetime is 2 years. Bill is still there or maybe somebody on the team still knows Bill's email address, but most likely the decision is taken to "just do it manually once because it's urgent" and we just recurse. This would have been the "obvious" choice when ISRG lit up Let's Encrypt.
But because the cert lifetime was 90 days, Bill gets told it's expiring at about the same time he's about to shove that "Actually automate certificates" work onto the back burner. Oh yeah, right, we should actually do that.
Now, it may be that at a startup that "Automation" is initially "Put it on Natalie the intern's TODO list every Monday" but by week six Natalie has learned about cron, or Windows event scheduling, or a dozen other ways to make stuff you always do regularly happen without actual work. Today I expect her LLM will tell her how to do it.
And Natalie is going to discover that some "permanent" solutions are more permanent than others, that you want monitoring and so on. But because Let's Encrypt is so popular unless you're doing something pretty esoteric you won't even be the first person to ask about it this month.
You wouldn't. Let's Encrypt doesn't verify anything more than that the requestor of the HTTPS certificate currently has control over the domain. The equivalent would verify nothing more than that the requestor of the S/MIME certificate currently has control over the email address. This part is actually a fairly standard procedure already, as a lot of websites will send verification emails to fully activate new account signups.
And packaging + distribution is ridiculously easy.
We are moving every script we can to Go.
Compared to bash it's a no-brainer, every dev can understand it, it has approximately .0000000000001% of the possible footguns, a great debugger, etc.
Browsers and TLS infrastructure have been solving that for a while now, via certificate transparency. Browsers can now reject any certificate that isn't publicly logged. So, yes, they could MITM, and burn an entire CA doing it.
If you're dealing with an authoritarian state they don't need to burn anything or care about cert logging. They can:
1. Make it illegal to distribute a browser that distrusts their CA
2. Make it illegal to run a browser that distrusts their CA
3. Block all encrypted traffic that they can't MITM and notify police that you are running illegal software
This is unnecessary, they already have the problem controlled better. They just outright block foreign services, and the domestic ones they can request data from freely.
Doesn’t require cracking crypto or any funny business around forcing people to install stuff.
Sure, a state can do that, and some have tried at various times. But even authoritarian states have a number of competing aims they have to balance. And CT makes authoritarian goals harder; they can no longer do as much surreptitiously.
“Make it illegal to…” that has never in human history prevented anything from happening. Cannot increase the risk? Of course, but laws do not stop humans from humaning.
Furthermore, doing any of the things you listed would isolate all legitimate network traffic as well as any undesirable traffic, fully shuttering all Chinese manufacturing businesses from global requests via the web. This hat would happen then? Phone calls, emails, and even physical mail would become the new norm and most of the Chinese economy would collapse under the weight of not being able to hop on a Zoom with a client that wants tooling made for its aluminum manufacturing molds.
So besides my point of the black market your hypothesis of total control misses all the other pressures that exist that make what you’re proposing infeasible on its face. Only a place like North Korea that is willing to be a pariah state is old be willing to take the economic and social costs associated with your proposal, and they’ve only been able to do that because their abominable regime was in place before the internet existed and they pre-built controls very late in the game.
Tl;dr simply because a country _could_ do something doesn’t mean it’s realistic for reasons outside of basic networking concepts.
Russia's ROSKOMNadzor has been trying to get users to install its own Root CAs in recent years. About 10 years ago everyone in the west removed CNNIC (Chinese counterpart) roots after they were caught MITM-ing.
I tried it on Firefox, but see it failed because it doesn't have WebGPU (on linux). Same thing on Chromium actually:
main.js:145 Myth browser startup failed Error: WebGPU adapter was not available.
at Module.initialize (foster-webgpu.js:400:9)
at async main.js:61:3
(anonymous) @ main.js:145
main.js:146 Error: WebGPU adapter was not available.
at Module.initialize (foster-webgpu.js:400:9)
at async main.js:61:3
reply