From: Stathis Papaioannou (stathisp@gmail.com)
Date: Tue Mar 04 2008 - 05:29:05 MST
On 04/03/2008, Lee Corbin <lcorbin@rawbw.com> wrote:
> > You would be relying on an evolved intelligence having at least some
> > similarity with human intelligence, not an unreasonable assumption.
> > But there is no reason why the written language of an alien species
> > might not be, for example, based on the radioactive decay pattern of a
> > sacred stone, in which case to us a message would look completely
> > random.
>
> But if they sent us one of their sacred stones, we would eventually
> find the non-random redundant pattern in the radioactive decay....
> Great example. We would find the pattern because it is objectively
> there, and its Komolgorov complexity is less than that of a random
> string.
I picked radioactive decay as an example of a truly random process
(you can predict the average number of decay events per unit time, but
not the actual number in a particular interval). It is effectively
what is called in cryptography a "one time pad": you would find no
pattern in the sequence, and you would need a copy of the original
radioactive decay pattern to make sense of it, not just the stone.
> > Similarly, an alien computer could seem to us completely random. There
> > need not even be a causal connection between the subcomponents of a
> > computation. A virtual environment in which you experience Monday
> > could be run on computer A, and then Tuesday run on (completely
> > different) computer B in the Andromeda galaxy a billion years later at
> > one billionth speed. As far as you're concerned, you experience Monday
> > followed by Tuesday. You could be in a simulation in which every
> > second of experience is implemented on different hardware, at
> > different rates, different physical locations, in no particular
> > sequence, and you would have no way of knowing it. The shorter the
> > frames are, the more likely that a sequence of physical events
> > isomorphic to whatever hardware/software combination you specify
> > arises by chance.
>
> Isn't this analogous to the code-maker's efforts to change frequencies
> extremely often in order to thwart enemy decryption? But even in
> those cases, if there really is a pattern there, sooner or later others
> break the code.
I'm not sure that there *has* to be a pattern in the functioning of a
computer or whether it could instead be deliberately be designed with
the equivalent of a cryptographer's one-time pad. Suppose for the sake
of argument that there does have to be a pattern, which an intelligent
being after long observation could in theory crack. You could still
increase the randomisation by breaking up a computation into parts,
implementing the parts on computers with different (randomly chosen)
architecture, and shuffling them. The only discernible pattern might
then be within short runs of the one type of computer, and if the
period of the runs could be brought down to the length of one clock
cycle, we might be back to a situation where the one-time pad analogy
holds. In any case, having very short runs exponentially increases the
probability that an isomorphic system will occur somewhere by
accident.
-- Stathis Papaioannou
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT