I'm reposting this. I forgot that I have to have the everything-list
email address first, otherwise it bounces from GoogleGroups :(
On Mon, Mar 10, 2008 at 08:56:24PM -0000, Alastair Malcolm wrote:
> I think this issue should be pushed further a little, since there seems to
> be a fundamental misunderstanding somewhere.
It seems a number of people have had difficulties following chapter
4. This is possibly because I haven't explained things well, or even
that there is somthing fundamentally incoherent in it, but I suspect
more likely is that it confronts existing epistemic baggage that
readers have, and that I haven't successfully taken this into
account. It is an important chapter, however, so it is worthwhile
trying to get this straight.
>
> (There appears to be a subsidiary issue of 'many description
> strings to one OM' (what you seem to be saying in your first email para),
> but perhaps that can be dealt with by different definiions of 'OM' - you
> seem to think of it as one OM 'type', whereas I think of it as one OM
> 'occurrence'.)
>
Multiple description strings map to a single OM. This is a relatively
obvious feature of a functionalist theory, which this theory is.
One can certainly think of the OM being an occurrence, but you m ust
also realise that there a multiple identical OMs (an uncountable
infinity in fact) within the everything.
> Perhaps the most direct route to the problem is your statement at the end of
> section 4.2 that (paraphrasing), the problem of induc`tion is not a problem
> for the plenitude provided [either the Schmidhuber solution holds or]
> observers always tend to model reality, find patterns, theories and so on
> that compress the description of the world around them, discarding
> non-model-fitting bits as 'noise'.
>
> The reason that there is no failure of induction (in general), under this
> scenario is, as I stated under (b) earlier (which appears to be the one
> closest to your view), that we are most likely to be in one of the simplest
> universes that supports SAS's, and it is this that determines the complexity
> of the universe that we are in, and provides order for our universe, and not
> how observers behave/operate in it.
>
But where does the premise "we are most likely to be in one of the simplest
universes that supports SASes" come from, if not from the properties
of the observer? That is what 4.1 is all about (coupled with section
5.1 of course).
But section 4.1 does not directly answer the failure of induction
issue. Nothing about Occam's razor says that a universe has to
continue being comprehensible.
> If an observer *didn't* model reality, find patterns etc, the world would
> still continue on its ordered way, even if the observer didn't appreciate
> this. (One could just conceive - and it is hypothetical possibility that is
> the issue here - of some kind of observer that used automatic filtering of
> irrelevant features of the environment together with a one to one mapping of
> the remainder to its internal processor - no actual compression of data as
> such.) Conversely, one could have a world where there were no laws, but a
> freak set of coincidences allowed SAS's to materialise, and (freakly) model
> the reality they (freakly) see - there would still be a failure of induction
> problem (no laws), even though observers model, find patterns etc.
Boltzmann brains are interesting, but seemingly extremely rare
occurrence. In any case, its hard to imagine a Bolztmann brain being
conscious unless its input was of a more sane and comprehensible
variety. However, I don't consider these in my book, and perhaps thats
an area for future work.
One can imagine being thrust into a perfect VR setup by an evil
scientist that connects your sensory input to white noise. I would
contend that the white noise wouldn't remain white noise for long, and
that one would enter a dreamlike state, where the noise was
interpreted as reality. Plus, the absence of correlata with one's own
mental states entailing a lack of self-awareness might lead to loss of
consciousness. I'm not sure how much bearing the sensory deprivation
experiments have on this.
>
> I understand that one can think of the universe as a set of cohering OM's,
> but the minimal specification of them should be something like a TOE
> (otherwise why do we see neurons and brains etc, if we care to look?), so
> 'physical universe' should normally be a relevant concept. And (under the
> above scenario) it is the simplest universe that counts for measure
> purposes, not the simplest OM - this explains the lack of white rabbits, but
> compression of the description of the world and 'random data' being
> discarded by our senses has nothing to do with the WR issue, as far as I
> can see.
>
The idea of cohering OMs forming a universe relates to the Anthropic
principle. By having phenomenal correlata, one can deduce that there
are other OMs that share a reality, and that which remains invariant
under shifts to different OMs become the law of physics. This is, in
essence Vic Stenger's Point of View Invariance (POVI) principle. This
set of laws of physics is the physical Universe inhabited by the
observers - there's nothing else I can imagine it to be.
So, it seems you are suggesting that time translation invariance (of
the same observer, but successive OMs) is enough therefore to banish
the white rabbits. But it does hinge on the anthropic principle, which
we need for other things.
The argument in sect 4.2 does not hinge on the AP, but does rely upon
the property of robustness of the observer, which I haven't completely
formalised, but I have made a couple of attempts in postings on this
list. Unfortunately, I just did a quick search in GoogleGroups, and
can't lay my hands on them just at the minute.
Perhaps there is some connection between robustness and the AP?
I really should write up a formal version of the robustness concept
somewhere more permanent, and also try to develop a formal model of
the process to see what happens in the face of pure white noise. I'm
sure it won't be white rabbits, as the argument in section 4.2 still
seems valid to me, but whether it is anything interpretable as a
universe is another thing.
> It seems then that there is more of a problem with my understanding of
> section 4.2 than I first appreciated!
>
> (Feel free to copy any reply to Everything-list if considered to be of
> general interest.)
>
> Alastair
>
>
>
> ----- Original Message -----
> From: "Russell Standish" <hpcoder.domain.name.hidden>
> To: "Alastair Malcolm" <amalcolm.domain.name.hidden>
> Sent: Thursday, March 06, 2008 10:27 AM
> Subject: Re: White Rabbit query from your TON
>
>
> >Description strings stand in a many-to-one relationship with OMs. The
> >diagram on page 33 summarises this neatly - the description are at the
> >base (syntactic space), and the OMs are in Semantic space.
> >
> >Therefore, of your 3 options, b) is closest to the mark. Only a finite
> >number of bits are relevant for specifying the OM (due to the
> >robustness requirements).
> >
> >It is perhaps misleading to refer to "physical universe". In some
> >sense, the physical universe must emerge out shared meanings from the
> >OMs. Bruno has much more to say on this topic :)
> >
> >The complexity of the OMs are due to an evolutionary process, and I
> >use this as a justification for later modeling observation (or
> >construction of the OMs) as an evolutionary process. The idea of
> >progressively reading in more bits of the description string is akin
> >to random mutations. The selection process is the observer making
> >meaning of the bits read. Heritability corresponds to preservation of
> >information through time, which is why we need to eliminate those
> >white rabbits.
> >
> >Cheers
> >
> >On Thu, Mar 06, 2008 at 09:21:31AM -0000, Alastair Malcolm wrote:
> >>Russell,
> >>
> >>I have recently been revisiting the 'White Rabbit' issue,
> >>and still am rather puzzled by parts of the relevant section (4.2) in
> >>'Theory of Nothing'. I'll focus on the statement 'the complexity of the
> >>world around us is actually the outcome of a very simple process called
> >>evolution' (third last para), and list three possible ways your
> >>'descriptions' can relate to our experiences (call them OM's).
> >>
> >>(a) A maximum of one OM per infinite description bit string, with only a
> >>finite segment being used by the observer - effectively we have 'all
> >>possible OM's' - but with this scenario there is no reason for us to
> >>expect
> >>to see a physical brain (or universe) compatible with that OM (and
> >>Darwinian
> >>evolution seems impossible). In a variation on this idea, each
> >>description
> >>string represents (can map to) all the OM's for a single observer (only),
> >>but this does seem rather artificial and perhaps solipsistic. I sense
> >>that
> >>you don't intend either of these variants, but I am not sure.
> >>
> >>(b) (My preferred option.) A relevant finite part of each description can
> >>represent a physical universe (perhaps a Tegmark level 1/2/3 or similar),
> >>and so can contain mappings to OM's in each universe. We are most likely
> >>to
> >>be in one of the simplest that supports SAS's - there are more bit
> >>combinations along the string beyond those specifying these universes on
> >>any
> >>equal (large) finite string comparison with more complex candidates, and
> >>it
> >>is this latter argument (or an equivalent) that determines the complexity
> >>of
> >>our world, and not evolution, which proceeds as per Darwin, in apparent
> >>contradiction of the quoted statement in my opening paragraph.
> >>
> >>(c) Each description bit string can have one or more mappings to what
> >>are,
> >>or could be, brief experiences. So 'observers will filter their inputs to
> >>more robustly obtain meaning from the data streaming in through their
> >>senses' - the more data we ignore the shorter the 'visible' part
> >>represented
> >>by the input string, and so the less the complexity - hence the
> >>complexity
> >>(at least of what we see) links to evolution, in keeping with the opening
> >>quotation. But note that the data that is excluded still forms part of
> >>the
> >>physical universe (else the observer wouldn't be in a position to reject
> >>it), which in turn implies the description string is split into three as
> >>regards mapping to an OM: this actual mapping, the rest of the mapping to
> >>our physical universe (or other OM's if you prefer), and the rest of the
> >>string, specifying rubbish (or conceivably other universes). (This last
> >>of
> >>the three is as under (b).)
> >>
> >>The reason I am dubious about (c) is that it seems certain that the most
> >>compact description of our universe (akin to a TOE perhaps) will have no
> >>'intra-universe-related' spare (rubbish) bits (the second of the three in
> >>(c)) for evolution to 'work' on (and all bits should mainly correlate
> >>with
> >>micro-level phenomena anyway). (I presume you are not referring to WR's
> >>as
> >>illusions or freak molecular/uncertainty-principle-based assemblages -
> >>this
> >>is not the WR problem.) White rabbits (flying at least) are physically
> >>impossible and so don't occur in our universe - as far as I can see this
> >>should be the context for WR discussions, and not phenomena like shadows.
> >>
> >>So as regards the 'third approach' of your section 4.2 I am not entirely
> >>clear as to which of the above three possibilities you are referring -
> >>probably some fourth one I have missed!
> >>
> >>With thanks for your time
> >>
> >>Alastair
> >
> >--
> >
> >----------------------------------------------------------------------------
> >A/Prof Russell Standish Phone 0425 253119 (mobile)
> >Mathematics
> >UNSW SYDNEY 2052 hpcoder.domain.name.hidden
> >Australia http://www.hpcoders.com.au
> >----------------------------------------------------------------------------
--
----------------------------------------------------------------------------
A/Prof Russell Standish Phone 0425 253119 (mobile)
Mathematics
UNSW SYDNEY 2052 hpcoder.domain.name.hidden
Australia http://www.hpcoders.com.au
----------------------------------------------------------------------------
----- End forwarded message -----
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to everything-list.domain.name.hidden
To unsubscribe from this group, send email to everything-list-unsubscribe.domain.name.hidden
For more options, visit this group at http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---
Received on Tue Mar 11 2008 - 22:55:54 PDT