From: Marc Warner (marc1warner@gmail.com)
Date: Wed Oct 14 2009 - 10:56:12 MDT
As someone who's new to the list I just wanted to say that, although you may
not have made your point to JKC, I'd like to thank everyone who argued with
him, particularly Robin and Mu, and give them credit for the measured tone
that they (mostly) used. I might have been lead astray if no one had
responded.
Without wanting to throw fuel on the fire - even if we were to (wrongly, in
my opinion) ignore the arguments of Robin & Mu & co - and try to accept
John's theory at face value, then it still doesn't seem consistent.
John - Why are your arguments about infinite loops limited to fixed goal
minds? It seems that within your framework, a non-fixed goal mind would
still be a Turing machine and susceptible to the same infinite loops. It's
possible that I still haven't understood your point; perhaps you could
clarify the link between a Turing machine and either a fixed or non-fixed
goal mind?
Marc
On Wed, Oct 14, 2009 at 5:05 PM, Robin Lee Powell <
rlpowell@digitalkingdom.org> wrote:
> I'm not going to bother to respond to that insane vitriol, but I
> want to point out to every one else that JKC cut, and failed to
> respond to, almost everything I wrote. Read what I actually wrote
> for yourself, and see if you think there's anything in there he
> should have replied to that he didn't.
>
> -Robin
>
> On Wed, Oct 14, 2009 at 08:59:17AM -0700, John K Clark wrote:
> > On Wed, 14 Oct 2009 "Robin Lee Powell" <rlpowell@digitalkingdom.org>
> > said:
> >
> > > I will not actually reproduce the math here; that would be rude
> > > to the authors, at the very least, and copyright violation.
> >
> > As he has demonstrated numerous times the last thing in the world Robin
> > Lee Powell would want to be is rude, but until this second I had no idea
> > he was such a big fan of copyright law. There for a minute I thought
> > that maybe the reason he didn't explain the math was that it was totally
> > incomprehensible to him and might as well have been written in Chinese.
> > But no, that's not the reason at all, it's the copyright thing. Yes, I'm
> > sure that's the explanation, it's the copyright thing.
> >
> > > Run the algorithm for at most N+1 steps. At every step, mark
> > > the state off in your enumeration. Stop when:
> >
> > > You reach a state (of both the FSM and memory) that Obviously if a
> computer
> > > has a finite memory it would be possible in seen before. Since the
> Turing machine formalism is
> > > entirely deterministic, the algorithm is in an infinite loop through
> this state, and you're done:
> > > does not halt.
> >
> > I've got to admit that makes a lot of sense, but it seems oddly
> > familiar, where did I see that before? Oh yes, now I remember, I said it
> > myself several posts ago:
> >
> > "Obviously if a computer has a finite memory it would be possible in
> > theory to keep a record of every state the machine goes into, and if you
> > found that a state was repeated then you'd know for sure that the
> > machine was in a infinite loop. But the trouble is that the machine
> > needed to do all that checking would have to be much larger than the
> > original machine, and that larger machine would have no way to know if
> > it was itself in a infinite loop unless it was watched over by an even
> > larger machine."
> >
> > > Stop When:
> > > The algorithm runs out of memory.
> >
> > Again I can't disagree with that, but for some reason I am again
> > overcome by a strong feeling of Deja View. I wonder what the cause could
> > be. Maybe it was the post I sent sever posts before the previous one I
> > quoted when I said:
> >
> > "So let's see, if you have a small memory you can prove that a computer
> > will halt when it runs out of memory. Wow what a profound result! I can
> > confidently predict that if you loaded Windows Vista into the original
> > ENIAC from 1946 that machine will halt too."
> >
> > You quote the authors (I guess text is not covered by copyright law,
> > just mathematics) as saying
> > "the abovementioned proof is not novel in theoretical computer science."
> > You also quote them as saying "Unfortunately, N will ordinarily be such
> > a huge number that this result is only theoretical, it does not lead to
> > a practical algorithm". But then, bizarrely, you say "Which was the
> > point many of us have been trying to make".
> >
> > What the hell are you talking about? What's this "many of us" crap? It's
> > the point I was trying to make but nobody agreed with me, that Turing
> > was profound and this stuff about finite memory machines is vacuous,
> > hell even the authors of the paper seem to think so. Since your short
> > term memory is so bad I will repeat something I quoted a few paragraphs
> > back that comes from one of my posts I sent about a dozen hours ago:
> >
> > "Obviously if a computer has a finite memory it would be possible in
> > theory to keep a record of every state the machine goes into, and if you
> > found that a state was repeated then you'd know for sure that the
> > machine was in a infinite loop. But the trouble is that the machine
> > needed to do all that checking would have to be much larger than the
> > original machine, and that larger machine would have no way to know if
> > it was itself in a infinite loop unless it was watched over by an even
> > larger machine."
> >
> > That is almost exactly what the paper said, and the authors of this
> > "paper" admit that it has no practical value and that they are saying
> > nothing new, so it is a bit of a mystery why they bothered to write it
> > in the first place, however there is no mystery why it is just online
> > and not published in a real science journal. But you read (skimmed more
> > likely) this piece of fluff and triumphantly announce "It does, in fact,
> > prove that the halting problem is only undecidable for infinite
> > computers" and then try to peddle the idea that some difficult problems
> > could be solved if only the computer had LESS memory. Should I start
> > pulling memory cards out of my computer now?
> >
> > >And, of course, none of this has anything to do whatsoever with goal
> systems
> >
> > Except that whenever you say "always do X, no exceptions" and X happens
> > to be one of the infinite number of programs that produce infinite loops
> > then your mighty AI turns into a useless lump of metal and silicon.
> > Turing wins again.
> >
> > John K Clark
> >
> >
> >
> >
> >
> >
> >
> > --
> > John K Clark
> > johnkclark@fastmail.fm
> >
> > --
> > http://www.fastmail.fm - Choose from over 50 domains or use your own
> >
>
> --
> They say: "The first AIs will be built by the military as weapons."
> And I'm thinking: "Does it even occur to you to try for something
> other than the default outcome?" See http://shrunklink.com/cdiz
> http://www.digitalkingdom.org/~rlpowell/ *** http://www.lojban.org/
>
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:05 MDT