On 24 February 2014 15:50, Bruno Marchal <[email protected]> wrote: > > On 24 Feb 2014, at 02:41, David Nyman wrote: > > On 24 February 2014 01:04, chris peck <[email protected]> wrote: > > *>>This is the same as saying that I will experience all possible futures >> in the MWI - but by the time I experience them, of course, the version of >> me in each branch will be different, and it always seems to me, >> retrospectively, as though I only experienced one outcome.* >> >> Each duplicate will only experience one outcome. I don't think there is >> any disagreement about that. The problems occur when considering what the >> person duplicated will experience and then what probability he should >> assign to each outcome and that seems to me to depend on what identity >> criterion gets imposed. Its a consideration I've gone into at length and >> won't bore you with again. But I will say that where you think that what >> Bruno wants is just recognition that each duplicate sees one outcome, I >> think that he actually wants to show that 3p and 1p probability assignments >> would be asymmetric from the stand point of the person duplicated. >> Certainly for me he doesn't manage that. > > > Correct me if I'm misremembering Chris, but I seem to recall proposing to > you on a previous occasion that Hoyle's pigeon hole analogy can be a useful > way of tuning intuitions about puzzles of this sort, although I appear to > be the sole fan of the idea around here. Hoyle's idea is essentially a > heuristic for collapsing the notions of identity, history and continuation > onto the perspective of a single, universal observer. From this > perspective, the situation of being faced with duplication is just a random > selection from the class of all possible observer moments. > > Well, the "just" might be not that easy to define. > > If the universal observer is the universal machine, the probability to get > a computational history involving windows or MacOS might be more probable > than being me or you. >
But how would "you" remember that? > > I am not sure that the notion of "observer moment" makes sense, without a > notion of scenario involving a net of computational relative states. > > I think the hypostases describe a universal person, composed from a > universal (self) scientist ([]p), a universal knower ([]p & p), an observer > ([]p & <>p), and a feeler ([]p & <>p & p)). > > But I would not say that this universal person (which exist in arithmetic > and is associated with all relatively self-referential correct löbian > number) will select among all "observer moment". > Well, perhaps "eventually" it will select all of them, if we can give some relevant sense to eventually in this context. And I suppose Hoyle's point is that if one imagines a logical serialisation of all such moments, its order must be inconsequential because of the intrinsic self-ordering of the moments themselves. Essentially he is saying that the panoptic bird view is somehow preserved at the frog level, at the price of breaking the simultaneity of the momentary views. > The "hypostatic" universal person is more like a universal baby, which can > split in a much larger spectrum of future 1p histories, but from its first > person perspective it is like it has still to go through the histories to > get the right relative statistics on his most probable universal neighbors. > Won't this still be effectively satisfied by Hoyle's heuristic? ISTM that "going through the histories" is a notion that splits in the 3p and 1p views. I suppose this is equivalent to conceiving observer moments as self-ordering monads in terms of which any random serialisation over the entire class must eventually preserve the right relative statistics. "Eventually" here relies on a similar opacity to delays in continuation as you argue in the UDA, plus the reliance on prior relativisation to some specific spatial-temporal orientation, to get a 1p notion of temporal order. But perhaps this formulation of a discrete observer moment is incompatible with comp? Of course, in the arithmetical reality, it don't get it, it is an indexical > internal point of view. > Perhaps it gets it "eventually", in the sense I outline above? > > The situations of having been duplicated one or more times are then just > non-simultaneous selections from the same class. This gives us a consistent > way of considering the 3p and 1p (or bird and frog) probabilities > symmetrically. That is, it is now certain that I will confront each and > every 3p continuation from a unique 1p perspective, just not simultaneously. > > That said, this approach retains a quasi-frequency interpretation of > probability in the case that there are fungible or equivalent > continuations. For example, if the protocol mandates that I will be > duplicated 100 times and 99 of my copies will be sent to a red room and one > to a blue room, it would be rational to anticipate a higher "probability" > of continuation associated with the larger class, even though each > continuation is individually certain in a different underlying sense. This > is just to say that subjective uncertainty (or the expectation of > probabilistic outcomes) is a function of incomplete knowledge at any given > point in the sequence. > > > OK. > > > I know that Bruno quarrels with Hoyle's idea as being superfluous to, or > possibly even incompatible with, comp > > > I think about it. I try to make sense of it. That might have sense, but > then it remains to look at it in arithmetic. > I mean the relations between a person and the universal person "in her" > is complex, and the splitting between []p and []p & p is part of it. > > but personally I still find it a neat heuristic for pumping one's > intuition on the indeterminacy of first-personal expectations. > > > OK. > It is just that I expect platonism to be counter-intuitive and so > intuition pump must be handled with care. But you know that. I just try to > understand the point. > Absolutely. I appreciate your interest. David > > > http://iridia.ulb.ac.be/~marchal/ > > > > -- > 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 [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/everything-list. > For more options, visit https://groups.google.com/groups/opt_out. > -- 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 [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/groups/opt_out.
