meekerdb wrote:
Pretty good list.
On 5/6/2015 6:23 PM, LizR wrote:
With profound and sincere apologies to Bruno, some people distinguish
these two items, so I thought it might be worthwhile trying to
marshall the arguments in one place, and give them simple names as per
the objections to the Chinese Room I seem to recall seeing in one of
DRH's books - "The Systems Reply" and so on.
So, why do some folks consider that "comp1" doesn't lead to "comp2" ?
1. "Arithmetical Realism is wrong" - the view that maths is something
we made up. This might be true, but imho it has (so far) failed to
convincingly explain away that pesky "unreasonable effectiveness".
I've never understood the "unreasonable effectiveness" argument. ISTM
that the effectiveness is evidence that we made it up. First, since
/*we*/ are creatures of this world what */we/* make will have some
effectiveness in this world just because of evolution and natural
selection. Second, much of mathematics was invented specifically to
solve problems in this world - so is it any surprise that it's
effective? Third, there is a lot of "game" mathematics that is only
effective in proving things about an invented realm.
I am with you here. I think it was Lee Smolin who said recently that the
effectiveness of mathematics in describing the physical world is greatly
over-rated. As Brent says, mathematics is about as effective as you
would expect for something that we made up to explain our experience of
the world. It is not "unreasonably effective" because, for example,
there is no closed form solution for the simplest three-body problem in
Newtonian gravity.
I would say the best argument for arithmetical realism (especially
arithmetic) is that it is ubiquitous and facts about it seem to be
discovered. But I don't think this is strong argument.
No, it is quite weak because once you have chosen the axioms, the
theorems that can be deduced from these by using the assumed rules of
inference are independent of what anybody thinks about them. It is just
like an invented game such as chess in this regard.
I must admit that I do take exception to Bruno's oft-repeated saying
that "if you accept that 2+2=4 independent of you or me then you believe
in arithmetical realism". That is disingenuous in the extreme, because
he then goes on to use the term "realism" with respect to arithmetic in
ways that do not follow from the simple tautology "2+2=4". Accepting
that tautologies are true independent of people, depending only on the
definition of the terms involved, is a far cry from realism about
arithmetic.
2. "Pronouns" - only one person takes this seriously, and has been
unable to convince anyone else (so far) but in theory there might be
something wrong with using pronouns when talking about matter
duplication (or AI programme duplication, or MWI experimenter
duplication...)
I think there is something wrong with */some/* uses of pronouns. What's
wrong is that they implicitly assume that there is an entity with
temporal continuity that can be identified as "you". But this depends
on a theory of personal identity that become problematic in the presence
of duplicating machines. Bruno seems to say that if M and W have the
same memories about all events prior to t, then M and W are the same
person, or maybe just the same person before t? But do they have to
have ALL the same memories; aren't their degrees of sameness? And is it
really only memories? Duplication is better understood for a computer
based AI, but that opens the possibility of duplicating exactly the same
memory, but having a slight variation in the processor.
However I don't think these seriously impact the argument because
personal identity is really based on bodies and Bruno's theory proposes
to replace our concept of bodies with something based on computation anyway.
But he plans to extract the physical world, with bodies in all their
messiness, from computations anyway. So he cannot simply discard bodies
as of no relevance to the concept of personal identity. Bodies exist in
his world as much as in ours. Basing personal identity solely on the
content of memories is seriously impoverished.
3. "Everything in the light cone" - the view that consciousness is
necessarily an open system, of which any description must take into
account all past influences that may impinge on it. This is more
interesting and persuasive than the first two, imho, but personally I
don't think it's relevant, since if one is going to make a recording
of a series of conscious states one can (in principle) isolate the
brain from the outside world by making a "cut" where signals from the
world turn to nerve impulses. One then has to "merely" record
everything at that interface, and all the computations going on in the
brain in response to that input.
The "brain in a vat" scenario. That is always a possibility and I do not
think that Bruno has adequate defenses against solipsism. He criticized
me for being defeatist when I said that we normally just ignore such
possibilities and get on with our lives because we can't do anything
about it anyway. But he can't rule out solipsism either.
But note that the meaning, the reference, of those computations in the
brain depend on the brain not being isolated in the past. Some of them
even depend on the evolutionary history of the species of that brain.
This is true, but can be overcome by enclosing the brain in a vat and
feeding artificial stimuli that mimic some external world.
4. The "Necessity of couterfactuals" - the view that any conscious
being needs to be able to handle counterfactuals. This is true in real
life, of course, but perhaps not if one is trying to create a
recording of a conscious state. Maudlin goes into this in his "Olimpia
and Klara" thought experiment, in which he constructs a Library of
Babel's worth of completely unnecessary machinery to handle the
counterfactuals, then refrains from actually using it. Istm that this
applies to the MGA as well - there is no actual necessity to deal with
cases that don't arise when one is repeatedly running the same
deterministic computer programme, hence this ability isn't needed when
replaying a recording of consciousness.
I don't know which side of this I come down on. But from one point of
view the requirement that consciousness be instantiated by a
computation, not just a playback, leads by extension to the ELC theory.
My feeling is that the whole deal with counterfactuals is just a
diversion. If consciousness can be implemented on a computer, then one
can run the same program with the same inputs any number of times, and
one only ever recreates the same conscious experiences -- there are not
different conscious moments for each run, and counterfactuals play no
part in repeats, they only come into play when you run with different
inputs, which is not the case under consideration. Hence, a recording
(which is equivalent to the original program) will, when replayed, also
generate the conscious experience. The point here being that it has to
be replayed on physical equipment, just as the original program was run
on physical equipment. So MGA does not eliminate the need for the
physical: rather, it reinforces the conclusion that the physical is
indispensable.
Besides, I have always found this insistence on counterfactual
correctness ironic for people who believe in MWI and happily abandon
counterfactual definiteness when they explain the EPR correlations. The
argument in MWI is that experiments that are not performed do not have
definite results. Exactly the same is true for the MGA. Hence
counterfactuals are irrelevant.
5. The "Argument from incredulity" - the whole idea is preposterous,
so there simply must be something wrong somewhere! This is yet to be
proven for all of modern science, most of which has been subjected to
it at some point.
The argument from incredulity has more punch when it's being used
against a reductio. It's like saying accept my incredible theory or
else you'll have to believe this absurdity.
The 'argument from incredulity' was raised against Bruno's
interpretation of the MGA. He claimed a contradiction, but really all
that was established was that he found the results of not believing his
interpretation to be absurd. This is not a valid argument -- what is
absurd to one person may well be quite reasonable to another.
And, for completeness...
6. The "pee-pee" argument, which makes fun of the terminology and
insults the ideas without making any constructive points. I will lump
into this arguments that because we don't have matter duplicators the
whole thing fails, and similar ideas. I think we can dismiss these
without further ado.
For convenience, and because I know we like making up our own
terminology, I have given these all handy abbreviations, namely ARW,
PN, ELC, NOC, INC and TPPA.
Is there anything I've missed?
I think ultrafinitism is another possibility. And it's not often
mentioned, but it's also possible that consciousness is a computation
that depends on the continuum (after all quantum mechanics does).
Yes, that. But I think the strongest argument against the theory is that
it doesn't actually give you a physical world. There is a claim that we
can extract the physical world by "interviewing the machine", but it is
not really clear what this means, and until such interviews give useful
physics, then the model has to be regarded as of dubious utility.
The point is that normal scientific practice requires any theory to be
measured against its empirical success. But if a claimed theory of
everything has no empirical successes at all, that is definitely a
strong argument against it. If the theory doesn't work, there must be
something wrong with it, even if it is not easy to put your finger on
exactly where the theory went wrong.
But I think we have identified enough weak points in the theory that we
need not be too concerned about the final demolition -- the theory will
simply fail to attract any substantial body of believers.
Bruce
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.