From: Bradley Thomas (email@example.com)
Date: Tue Oct 13 2009 - 16:15:01 MDT
I'm not saying there necessarily has to be a cyclical pattern, I'm saying
all real programs have to return to a prior state, or halt. If we add input
to the program that is non-cyclic (for example, the digits of an irrational
number) then I agree we can create programs that do not repeat with a
cyclical pattern. But all real programs will still return to an exact prior
state or halt. So real programs can only be regarded as not operating in an
infinite loop by virtue of the fact that an external system (e.g. humans)
snap it out of this loop with unpatterned input data. As I read it, this is
pretty close to what JKC is saying.
Twitter @bradleymthomas, @instansa
From: firstname.lastname@example.org [mailto:email@example.com] On Behalf Of Arets
Sent: Tuesday, October 13, 2009 5:45 PM
Subject: Re: [sl4] I am a Singularitian who does not believe in the
You may be very well aware of that the average value fully depends solely on
the input parameters, i.e. on the given sequence of numbers. So, as long as
the sequences being fed in the algorithm do not follow a cyclical pattern,
equally will not the calculated values. Finiteness of memory in this case
limits only the maximum length of computable sequence (and, technically, the
size of the code representing the algorithm), but not whether the algorithm
is linear or cyclical. More so, it is obviously possible to construct a
cyclical algorithm running for only a finite amount of steps, thus still
escaping any flow infinities.
Of course, most real programs are infinite loops in the sense they will run
indefinitely *unless specific exit conditions are met*; this, I assume,
could very well be true in the case of AI, too; in normal conditions, it
would operate without terminal interrupt, i.e. "pseudo-infinitely". On the
other hand, the Big Bang model predicts a finite Universe with a finite
amount of included matter, therefore, with this, there can be no true
infinities at all outside pure mathematical theory.
On Wed, Oct 14, 2009 at 12:29 AM, Bradley Thomas <firstname.lastname@example.org> wrote:
> Yes, at some point the average value must return to some prior average
> value given finite memory.
> Brad Thomas
> Twitter @bradleymthomas, @instansa
> -----Original Message-----
> From: email@example.com [mailto:firstname.lastname@example.org] On Behalf Of
> Sent: Tuesday, October 13, 2009 5:16 PM
> To: email@example.com
> Subject: Re: [sl4] I am a Singularitian who does not believe in the
> Including clearly linear ones? Can you consider, exempli gratia, an
> algorithm for calculating average value of given sequence of number to
> necessarily be an infinite loop?
> On Tue, Oct 13, 2009 at 11:57 PM, Bradley Thomas <firstname.lastname@example.org>
>> *My point is that any real algorithm in any real computer is
>> automatically in an infinite loop.
>> In the sense that it has to return to a prior state sooner or later
>> (and that state may be halted).
>> Brad Thomas
>> Twitter @bradleymthomas, @instansa
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:04 MDT