Hacker News Reader

Comment by tauneutrin0 | Hacker News Reader

The internet's security relies on an almost impenetrable wall built around it. The king appoints gatekeepers that are each entrusted with a gate they must protect. The gatekeepers must follow a rigorous interrogation process to verify the ID of anyone who wants to pass through their gate. Only then are they permitted to unlock it.

The king was very careful to ensure all the gatekeepers were following the process correctly. It did not occur to anybody that one of the gatekeepers could be using a dirty rag to jam their gate closed instead of a lock.

Once discovered, the wall could be trivially bypassed. The security of the whole internet was immediately broken. Rebuilding the wall to block up the dodgy gate took much effort from very many people.

View on HN

I think part of the issue is that we allow gatekeepers to open any gate.

This system was invented in the 90s when things were looking up. China was opening up to the world, the Russians were suddenly not evil anymore, and the internet was supposed to bring us all together into one happy global society, kumbaya and all that. Obviously that didn't work out.

It's time for a rethink really. Every CA is now a single point of failure.

The entire system (CAs and ICANN DNS) was never good to begin with; it's just become that much worse due to the shifting situation over the years. It's an inherently centralized design with no way to verify that example.com is operated by the same entity from one day to the next. The namespace is global and effectively an exercise in rent seeking.

At absolute minimum any sensible system would have a TOFU element whereby the site operator of example.com changing from one day to the next would result in all the tooling across the entire stack screaming loudly (in other words what ssh does). And that's just the absolute bare minimum change that's needed.

We do at least have CAA records that you can use to tell clients to only accept TLS certificates from specific CAs for your hostnames, but ideally that would be opt-out rather than opt-in.

Now the gatekeeper is saying "I don't want to be a gatekeeper anymore for this one particular gate, because people were mean to me. I'll however continue being gatekeeper for these other gates that drive my business either way."

I think a part of the issue was, everyone living outside the walls did not even try to kick the gates open as everyone assumed a big (cryptographically secure) lock was being used by all gatekeepers.

Its called a Potemkin solution, and there is a serious theory that most nuclear arsenals are exactly that: no-one has ever confirmed they can actually be fired.

In the case of nuclear arsenals that's perfectly fine - MAD simply needs to be plausible, not robustly demonstrated. In the case of CAs I'd argue that the description doesn't really fit since they're getting stress tested by malicious parties on a daily basis. By that logic you could describe linux kernel security as a potemkin village due to the steady stream of zero days. In reality a potemkin village is something intended from the start to deceive everyone else. It's not impossible that's what happened here but that's a very serious allegation to make (ie theorizing that the weak keys were intentional with a conspiracy afoot).

Based on everything with the NSA and backdoors etc, the idea that this was intentional is not inherently implausible.

The reason I agree with you is that it was too obvious i.e. if you were backdoooring something, you would want it to be more difficult to detect than this.

I guess the main takeaway is that the entire Web PKI is a far bigger house of cards than everyone assumes. Ignorance is bliss, I guess.

> no-one has ever confirmed they can actually be fired

that is untrue

https://drmaciver.com/2022/05/learning-to-walk-through-walls...

Citing Terry Pratchett’s “Interesting Times”.

In this analogy I think it would be more appropriate to call them a lockmaker, and a lot of gates have been using this lock. Finding and fixing every lock is where the effort comes from, and some locks may not be found for years to come.

This lockmaker analogy fails to capture one critical aspect: the failure of one CA breaks the security of everyone, even if they did not use that CA's "locks".

This is why I went for a "wall with many gates" instead; any bypass makes all gates redundant.

All analogies break down at some level; the main goal is to clearly communicate the most important ideas.

Reply on news.ycombinator.com