From: Benja Fallenstein (benja.fallenstein@gmail.com)
Date: Mon Jan 12 2009 - 09:50:35 MST
Hi Norman,
On Mon, Jan 12, 2009 at 10:12 AM, Norman Noman
<overturnedchair@gmail.com> wrote:
> Good lord, it doesn't have to be nearly this complicated.:
You're right. Sorry.
My actual reply to Matt was actually just this paragraph,
>> Matt, as far as I can see you're wrong, though; K(X) = K(Y), no
>> contradiction. Yes, if X "did something besides simulating Y," in a
>> certain sense, then you would have K(X) > K(Y). But by assumption,
>> each AI simulates (eventually) everything the other AI does, so in
>> that sense, Y "does everything" that X does and vice versa. To see
>> that you can have Xs and Ys with such source codes: Let's say that you
>> have X' that takes the source of Y as input and behaves like you want
>> X to behave, and, symmetrically, you have Y' that thakes the source of
>> X as input. Then it seems to be it's basically just an exercise in
>> writing quines to obtain X (using both X' and Y') and to obtain Y
>> (again using both X' and Y'), with K(X) = K(Y).
and what followed was an attempt to show that the simple exercise in
writing quines wasn't just hand-waving. I should've deleted it after
seeing how long it got... I'll try to learn the lesson for next time.
> This is what annoys me about technical explanations...
What I find unfortunate is people throwing their intuitions at each
other without ever trying to address objections on their own grounds,
but since I haven't managed to do better, I must accept the rebuke :-)
All the best,
- Benja
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:03 MDT