That did seem to be the gist of this post before the comment that Manishearth particularly replied to. But that comment implies that @fourtyseven is interested in cross compiling, and not concerned about trust.
If this is about trust, “Reflections on Trusting Trust” is a great speech, but I think its misconstrued in a lot of discussion about it as raising a security threat that we ought to protect ourselves from.
First, Ken Thompson’s trojan identified the source code of the particular program he was attacking, as well as the compiler, both already available to him - inserting a general trojan that could attack some arbitrary set of programs, including the programs you care about, is a much more challenging problem. There’s a quote I can’t find the source of about the idea of a trojan in the OCaml compiler which inserts vulnerabilities into a program written in Rust - we should hope this vulnerability exists, because it would be definitive proof of time travel.
Second, Ken Thompson’s point was that we can’t guarantee the security of our systems unless we have a complete binary understanding of every program which we execute on them, something that has been impossible for a very long time. He meant that no matter what we do, we have to put our trust in someone else’s code.
