Re: language, cloning and thought experiments

[Saibal Mitra]

I just send a posting to the FOR list about my article. I did not have the
time to reply to everyone on this list previously. Reading the old
discussion again, I think that it was suggested that the exact quantum
states matter, but they don't. It was only used to illustrate the thought
experiment by Deutsch which would allow one to prove that the MWI is
correct.

This is what I sent to the FOR list:


Some time ago I wrote a small article:


http://arxiv.org/abs/0902.3825


This was recently featured in New Scientist:


http://www.newscientist.com/article/mg20227044.200-avoid-a-future-cataclysm-forget-the-past.html


The idea is that an observer can undo things that have already happened by
resetting its memory, because when you reset your memory to a previous
state, that previous state you are evolving into will be the same in a
sector of the multiverse which evolved from that previous state and then was
reset for any reason. So, if the reseting is triggered for reason A or
reason B, it would lead to the observer ending up in the same state. The
outcome of a new measuremnt to find out why the mempory was reset is then
not pre-determined.

Some details:

The word "state" here refers to the classically describable state of an
observer. In the article, I focus on machine observers. The subjective state
of the observer is then exactly specified by specifying the ones and zeroes
of the bits of the memory. So, I assume that whatever the observer can be
aware of is encoded by the classical state of the bits of the computer and
not the exact quantum state of the computer. The exact state of the computer
has to be specified using a wavefunction of the computer (in fact, the state
of the computer will be entangled with the rest of the universe).


Then, one can write down any generic quantum state of the universe
containing the observer by supplementing the (classical) information stored
in the bits by the extra information you need to fully specify the
wavefunction of the computer and everything else in the universe. One can
then consider the unitary transformations that would represent a memory
backup, memory resetting etc.


After the memory resetting, you are notified why the memory was reset. Since
the relevant things happen in the realm where classical physics applies, the
probabilities are the same as what you would find using purely classical
reasoning. The interpretation of these probablilites is, however, different
from classical physics. When the memory is reset, you evolve to some state
while the rest of the inverse will be in some superposition of states in
which the memory was reset for various reasons. Then, before finding out why
the memory was reset, the outcome of that observation is not pre-determined


--~--~---------~--~----~------------~-------~--~----~
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@...
To unsubscribe from this group, send email to everything-list+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---

[Bruno Marchal]

Accepting QM without collapse, I am not sure you can dump your memory  
in the environment in any truly irreversible way.

Bruno


On 21 Apr 2009, at 15:22, Saibal Mitra wrote:

>
> Yes, I agree, and that's then why we cannot do this in practice. The
> verification of the MWI would have to wait untilk we have artificially
> intelligent observers implemented by quantum computers.
>
> However, ass uming that the MWI is indeed correct, it doesn't matter  
> if you
> undo the measurement. If you just dump your memory in the nvironment  
> in an
> irreversible way, you end up in a superposition like:
>
> |you>[ |universe_1| + |universe_2> ]
>
> As far as |you> are concerned, it doesn't matter if |universe_1> and
> |universe_2> differ by one electron state or the state of 10^23  
> particles:
> the result of a new measurement is not pre-determined in either case.
>
>
> ----- Original Message -----
> From: "Brent Meeker" <meekerdb@...>
> To: <everything-list@...>
> Sent: Sunday, March 15, 2009 08:06 PM
> Subject: Re: Changing the past by forgetting
>
>
>>
>> Saibal Mitra wrote:
>>> If we consider measuring the spin of a particle, you could also  
>>> say that
> the
>>> two possible outcomes just exist and thatthere are two possible  
>>> future
>>> versions of me. There is no meaningful way to associate myself with
> either
>>> of the two outcomes.
>>>
>>> But then, precisely this implies that after a measurement and  
>>> forgetting
>>> about the result will yield a version of me who is in a similar  
>>> position
> as
>>> that earlier version of me who had yet to make the measurement. If  
>>> one
> could
>>> perform measurements in a reversible way, this would be possible to
>>> experimentally confirm, as David Deutsch pointed out. You can  
>>> start with
> a
>>> spin polarized in the x direction. Then you measure the z-component.
> There
>>> then exists a unitary transformation which leads to the observer
> forgetting
>>> about the outcome of the measurement and to the spin to be  
>>> restored in
> the
>>> original state. The observer does remember having measured the
> z-component
>>> of the spin.
>>>
>>> Then, measuring the x-component again will yield "spin-up" with 100%
>>> probability, confirming that both branches in which the observer
> measured
>>> spin up and spin down have coherently recombined. This then proves  
>>> that
> had
>>> the observer measured the z-component, the outcome would not be a  
>>> priori
>>> determined, despite the observer having measured it earlier. So,  
>>> both
>>> branches are real. But then this is true in general, also if the  
>>> quantum
>>> state is of the form:
>>>
>>> |You>[|spin up>|rest of the world knows the spin is up> + |spin
> down>|rest
>>> of the world knows spin is down>]
>>
>> You're contemplating reversing three different things:
>>
>> 1) Your knowledge, by forgetting a measurement result.  Something  
>> that's
> easy to do.
>>
>> 2) The spin state of a particle.
>>
>> 3) The state of what the rest of the world knows.
>>
>> Because of the entanglement, I don't think you can, in general,  
>> reverse
> the spin
>> state of the  particle without reversing what is known about it by  
>> "the
> rest of
>> the world".
>> If it was a known state (to someone) the particle can easily be put  
>> back
> in that
>> state.  But to do so for a general, unknown state, after a  
>> measurement
> would
>> require invoking time-reversal invariance of the state of whole  
>> universe
> (or at
>> least all of it entangled with the particle spin via the measuring
> apparatus).
>>
>> Brent Meeker
>>
>>>
>>> although you cannot directly verify it here. But that means that you
> cannot
>>> rule out an alternative theory in which only one of the branches  
>>> is real
>>> when performing a measurement in this case. But if the reality of  
>>> both
>>> branches is accepted, then each time you make a measurement and you
> don't
>>> know the outcome, the outcome is not fixed (proovided, of course,  
>>> there
> is
>>> indeed more than one branch).
>>>
>>>
>>> ----- Original Message -----
>>> From: "Jack Mallah" <jackmallah@...>
>>> To: <everything-list@...>
>>> Sent: Thursday, March 12, 2009 03:47 AM
>>> Subject: Re: Changing the past by forgetting
>>>
>>>
>>>
>>>
>>> --- On Tue, 3/10/09, Saibal Mitra <smitra@...> wrote:
>>>> http://arxiv.org/abs/0902.3825
>>>>
>>>> I've written up a small article about the idea that you could end  
>>>> up in
> a
>>> different sector of the multiverse by selective memory erasure. I  
>>> had
>>> written about that possibility a long time ago on this list, but now
> I've
>>> made the argument more rigorous.
>>>
>>> Saibal, I have to say that I disagree.  As you acknowledge, erasing
> memory
>>> doesn't recohere the branches.  There is no meaningful sense in  
>>> which
> you
>>> could end up in a different branch due to memory erasure.
>>>
>>> You admit the 'effect' has no observable consequences.  But it has  
>>> no
>>> unobservable meaning either.
>>>
>>> In fact, other than what I call 'causal differentiation', which  
>>> clearly
> will
>>> track the already-decohered branches (so you don't get to  
>>> reshuffle the
>>> deck), there is no meaningful sense in which "you" will end up in  
>>> one
>>> particular future branch at all.  Other than causal differentiation
>>> tracking, either 'you' are all of your future branches, or 'you' are
> just
>>> here for the moment and are none of them.
>>
>>>
>
>
> >

http://iridia.ulb.ac.be/~marchal/




--~--~---------~--~----~------------~-------~--~----~
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@...
To unsubscribe from this group, send email to everything-list+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---

[Saibal Mitra]

That's correct. It is not really irreversible. The point is that it doesn't
matter as you end up in a state where the outcome of finding out what
happened is not pre-determined.

Saibal

----- Original Message -----
From: "Bruno Marchal" <marchal@...>
To: <everything-list@...>
Sent: Tuesday, April 21, 2009 07:27 PM
Subject: Re: Changing the past by forgetting

>
> Accepting QM without collapse, I am not sure you can dump your memory
> in the environment in any truly irreversible way.
>
> Bruno
>
>
> On 21 Apr 2009, at 15:22, Saibal Mitra wrote:
>
> >
> > Yes, I agree, and that's then why we cannot do this in practice. The
> > verification of the MWI would have to wait untilk we have artificially
> > intelligent observers implemented by quantum computers.
> >
> > However, ass uming that the MWI is indeed correct, it doesn't matter
> > if you
> > undo the measurement. If you just dump your memory in the nvironment
> > in an
> > irreversible way, you end up in a superposition like:
> >
> > |you>[ |universe_1| + |universe_2> ]
> >
> > As far as |you> are concerned, it doesn't matter if |universe_1> and
> > |universe_2> differ by one electron state or the state of 10^23
> > particles:
> > the result of a new measurement is not pre-determined in either case.
> >
> >
> > ----- Original Message -----
> > From: "Brent Meeker" <meekerdb@...>
> > To: <everything-list@...>
> > Sent: Sunday, March 15, 2009 08:06 PM
> > Subject: Re: Changing the past by forgetting
> >
> >
> >>
> >> Saibal Mitra wrote:
> >>> If we consider measuring the spin of a particle, you could also
> >>> say that
> > the
> >>> two possible outcomes just exist and thatthere are two possible
> >>> future
> >>> versions of me. There is no meaningful way to associate myself with
> > either
> >>> of the two outcomes.
> >>>
> >>> But then, precisely this implies that after a measurement and
> >>> forgetting
> >>> about the result will yield a version of me who is in a similar
> >>> position
> > as
> >>> that earlier version of me who had yet to make the measurement. If
> >>> one
> > could
> >>> perform measurements in a reversible way, this would be possible to
> >>> experimentally confirm, as David Deutsch pointed out. You can
> >>> start with
> > a
> >>> spin polarized in the x direction. Then you measure the z-component.
> > There
> >>> then exists a unitary transformation which leads to the observer
> > forgetting
> >>> about the outcome of the measurement and to the spin to be
> >>> restored in
> > the
> >>> original state. The observer does remember having measured the
> > z-component
> >>> of the spin.
> >>>
> >>> Then, measuring the x-component again will yield "spin-up" with 100%
> >>> probability, confirming that both branches in which the observer
> > measured
> >>> spin up and spin down have coherently recombined. This then proves
> >>> that
> > had
> >>> the observer measured the z-component, the outcome would not be a
> >>> priori
> >>> determined, despite the observer having measured it earlier. So,
> >>> both
> >>> branches are real. But then this is true in general, also if the
> >>> quantum
> >>> state is of the form:
> >>>
> >>> |You>[|spin up>|rest of the world knows the spin is up> + |spin
> > down>|rest
> >>> of the world knows spin is down>]
> >>
> >> You're contemplating reversing three different things:
> >>
> >> 1) Your knowledge, by forgetting a measurement result.  Something
> >> that's
> > easy to do.
> >>
> >> 2) The spin state of a particle.
> >>
> >> 3) The state of what the rest of the world knows.
> >>
> >> Because of the entanglement, I don't think you can, in general,
> >> reverse
> > the spin
> >> state of the  particle without reversing what is known about it by
> >> "the
> > rest of
> >> the world".
> >> If it was a known state (to someone) the particle can easily be put
> >> back
> > in that
> >> state.  But to do so for a general, unknown state, after a
> >> measurement
> > would
> >> require invoking time-reversal invariance of the state of whole
> >> universe
> > (or at
> >> least all of it entangled with the particle spin via the measuring
> > apparatus).
> >>
> >> Brent Meeker
> >>
> >>>
> >>> although you cannot directly verify it here. But that means that you
> > cannot
> >>> rule out an alternative theory in which only one of the branches
> >>> is real
> >>> when performing a measurement in this case. But if the reality of
> >>> both
> >>> branches is accepted, then each time you make a measurement and you
> > don't
> >>> know the outcome, the outcome is not fixed (proovided, of course,
> >>> there
> > is
> >>> indeed more than one branch).
> >>>
> >>>
> >>> ----- Original Message -----
> >>> From: "Jack Mallah" <jackmallah@...>
> >>> To: <everything-list@...>
> >>> Sent: Thursday, March 12, 2009 03:47 AM
> >>> Subject: Re: Changing the past by forgetting
> >>>
> >>>
> >>>
> >>>
> >>> --- On Tue, 3/10/09, Saibal Mitra <smitra@...> wrote:
> >>>> http://arxiv.org/abs/0902.3825
> >>>>
> >>>> I've written up a small article about the idea that you could end
> >>>> up in
> > a
> >>> different sector of the multiverse by selective memory erasure. I
> >>> had
> >>> written about that possibility a long time ago on this list, but now
> > I've
> >>> made the argument more rigorous.
> >>>
> >>> Saibal, I have to say that I disagree.  As you acknowledge, erasing
> > memory
> >>> doesn't recohere the branches.  There is no meaningful sense in
> >>> which
> > you
> >>> could end up in a different branch due to memory erasure.
> >>>
> >>> You admit the 'effect' has no observable consequences.  But it has
> >>> no
> >>> unobservable meaning either.
> >>>
> >>> In fact, other than what I call 'causal differentiation', which
> >>> clearly
> > will
> >>> track the already-decohered branches (so you don't get to
> >>> reshuffle the
> >>> deck), there is no meaningful sense in which "you" will end up in
> >>> one
> >>> particular future branch at all.  Other than causal differentiation
> >>> tracking, either 'you' are all of your future branches, or 'you' are
> > just
> >>> here for the moment and are none of them.
> >>
> >>>
> >
> >
> > >
>
> http://iridia.ulb.ac.be/~marchal/
>
>
>
>
> >

--~--~---------~--~----~------------~-------~--~----~
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@...
To unsubscribe from this group, send email to everything-list+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---