Ask HN: Can any computer program be secured against advanced intelligence?
3 pointsby mad_8 hours ago1 comment
  • battery_staple_8 hours ago
    I think this question reduces to: is it possible to prove that a program is bug free, including on one end

    - the hardware it's running on

    - the physical forces acting on that hardware

    and on the other end

    - the feature that the user believes the program implements

    - the user's understanding of what that feature means

    So realistically, no. But if were willing to narrow the question to one which assumes the hardware executes its ABI perfectly, and the contract you wrote for the feature is exactly what the user asked for, then maybe, but still probably no.

    Security is a tradeoff between how much you spend securing it, and how much it's worth to you to delay incursion by an amount of time that allows you to mitigate the incursion.

    • mad_8 hours ago
      This is sort of my thought, but then the question becomes, at some point, has the software been tampered with. Assuming everything can be rewritten and replaced with a backdoor or whatever, can you even verify that the software was not tampered with, and we use hashing to do that, but even that, at some point in the future, could become suspect.
      • battery_staple_7 hours ago
        If this matters to you, the risc-v community may appeal to you. It's the only place (I'm aware of) where the entire stack from lithography to software can be source-available. (I say source-available instead of open-source because at this level of paranoia, it doesn't even matter if you have to buy a copy of the source, since you're gonna have to spend rather a lot on tokens to do the security analysis.)