Re: The Future of Secrecy

From: Hal Finney (hal@finney.org)
Date: Wed Jun 18 2003 - 22:14:41 MDT

  • Next message: EvMick@aol.com: "Re: like, far out, grahs hoppah"

    In the book The Golden Age, the truth-detection was done by the
    Sophotechs, the super-AIs millions of times smarter than people.
    These AIs were also trustworthy ("friendly"?), so if they learned
    someone's secrets it was OK. It was a sort of literal deus ex machina
    solution to the problem.

    But if the people doing the mind-reading are of only equal complexity and
    intelligence to the minds being read, then it seems like it would be much
    more difficult to untangle the truth about what someone's true beliefs
    and motivations are. It's not even clear that a human level intelligence
    could understand the brain of a human being in sufficient detail.

    As far as the issue of keeping secrets that Wei raised, there might be a
    few possible solutions. One would be to create an AI which would read
    your mind, report on your truthfulness, and then self-destruct. Or,
    given such an AI program, in principle you could use a crypto algorithm,
    a zero knowledge proof, to show that your brain state would satisfy the
    AI program, without actually running it.

    Another possibility would be to standardize on mental architectures which
    have a fairly strict separation between the motivational structure and
    certain kinds of factual knowledge. Then you could expose the one part
    but keep the other secret, and the mind reader could be confident that
    the secret data would not have a significant impact. However I'm not
    sure it makes sense to keep secrets but somehow to have them be unable
    to affect your motivations.

    As a final point, the question was raised about how to allocate resources
    for problem solving: how much to spend working on the problem, versus
    looking for other methods to solve it. Apparently the answer to this
    is known, at least in principle. Juergen Schmidhuber, best known for
    his idea that the multiverse consists of all possible computer programs
    being run at once, has a page describing Levin Search, an algorithm that
    goes back to 1973:

    http://www.idsia.ch/~juergen/optimalsearch.html

    Schmidhuber's team has improved and extended the algorithm. Basically
    the computer spends half its time working on the problem using the best
    algorithm it has found so far, and the other half of the time trying to
    find new algorithms for the problem. This also includes time to try to
    improve the algorithm-search technique. Apparently in the long run this
    works as fast as if you knew the very best algorithm from the beginning,
    to within a small constant factor.

    So that's it, that's the best algorithm for living in the world, I guess.

    Hal



    This archive was generated by hypermail 2.1.5 : Wed Jun 18 2003 - 22:26:06 MDT