[Cryptography] New SSL/TLS certs to each live no longer than 47 days by 2029

[Jon Callas]

> On Apr 23, 2025, at 12:39, Michael Kjörling <9bf3a7ef93bb at ewoof.net> wrote:
> 
> Another page which is relevant to the question would be <https://blog.apnic.net/2024/05/28/calling-time-on-dnssec/>.
> 

Thanks! Wow, what a good essay. One where I would also say for once, read the comments.

There are a number of excellent points in there about subtle things in network architecture, like that TLS protects an application, DNSSEC an address, and these are not the same.

Also that there's a lot of missed opportunity, or 20/20 hindsight that had DNSSEC been done in the late 90s, and had supporting TLS be a goal, it might have happened. Personally, I think that's far more describing an alternate universe than a missed opportunity, but it's still an interesting thing to think about.

I still think that from our cryptographic viewpoint, there's still the issue that this is comparing one PKI vs another, and there's a whole lot of complication there, particularly when we think about what we could add on to safeguard against an adversary who is an owner of a TLD. What's the analogue of CAA, at least in part? CAA props up weaknesses in the WebPKI via DNS (SEC or not), what would be an analogue?

	Jon