Reflections on Trusting Trust (1984) [pdf]

There is also GNU Mes, which is able to build a GCC toolchain starting from nothing but a Scheme interpreter:

https://www.gnu.org/software/mes/

Practical implementation notes can be found on the Guix blog:

https://guix.gnu.org/blog/2019/guix-reduces-bootstrap-seed-b...

https://guix.gnu.org/blog/2020/guix-further-reduces-bootstra...

> Are you protecting against an actor with goals and capabilities, or are you trying to make yourself feel better?

It is difficult to protect against state actors.

But a lot of attackers are not state actors. They are private actors (whether lone individuals or groups) going after the low-hanging fruit. Protecting against this threat is a lot easier.

And for a lot of enterprises, the second category is a bigger threat than the first. Even if the NSA steals all your data, they are unlikely to release it to the general public or use it for extortion. The later group are much more likely to do things like that. If the NSA hacks you, you quite possibly will never know they've done it. If the later group hacks you, they'll make sure you know they've done it by the pain they'll cause.

(Of course, I realise how for companies in certain sensitive industries, such as defence, aerospace, semiconductor fabrication, etc, the threat of state-sponsored cyber-espionage from competing international powers is a real threat – however, I think in those cases one's own country's three-letter agencies are often keen to be helpful.)

It's possible to provably construct a program that will halt. That is the whole point of loop invariants. Similarly it is possible to provably construct a "secure" program up to physical limitations. However in both cases it requires that the semantics of the programming environment are known to the programmer. Even without a malicious adversary there are gotchas at every level from hardware bugs, compiler bugs, programming errors, incorrect configuration, and who knows what else that impair the programmer's ability to understand the actual semantics of the programming environment. When you add to that the potential for maliciously clever modifications such as the one in the Thompson paper along with the undeniably large benefit to any number of parties that might wish to compromise same, the only rational conclusion is that no computing system deserves to be trusted in itself.

I don't trust other levels of the alleged chain of trust either. For example I consider it an absolute certainty that every competent globally active intelligence agency in the world has assets inside Facebook and Google, and most likely others. The cost and difficulty of inserting or recruiting an asset is so low and the potential benefits are so high that it's certain to have happened and in fact already has[1].

By the way while my ex-NSA buddy didn't elaborate on what he meant by attacking the implementation, but I'm pretty sure that includes subverting an implementer.

Edit: I don't think we really substantially disagree, the above is meant to be elaboration not rebuttal.

[1] https://www.washingtonpost.com/world/national-security/nsa-i...