- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Jacques M. Mallah <jqm1584.domain.name.hidden>

Date: Wed, 18 Nov 1998 18:15:06 -0500 (EST)

On Mon, 16 Nov 1998, Wei Dai wrote:

*> On Mon, Nov 16, 1998 at 06:32:23PM -0500, Jacques M. Mallah wrote:
*

*> > However, if we assume all the original programs are running at the
*

*> > same speed, the thing to do would be to look at the total number of
*

*> > conscious computations implemented by each as time goes to infinity. (For
*

*> > more on what I mean by an implementation, which is a nontrivial concept,
*

*> > see my web page on interpretation of QM.)
*

*>
*

*> A big problem with this idea is that the probablities you end up with will
*

*> be very sensitive to the exact nature of the computer you're using to run
*

*> the programs. For example if you use a quantum computer instead of a
*

*> Turing machine, simulations of quantum universes will go much faster and
*

*> the conscious computations in those universes will have larger measures.
*

Perhaps, but if so that's just an example of the point that I

made, that even if all possibilities exist there are still free parameters

in specifying how to count them up in order to make predictions. So it's

really not a problem with the idea I described, which is a simple (thus

interesting) choice for the free parameters, but rather it's a problem for

the more ambitious goal of having no information, which I currently doubt

could be achieved.

Thanks for posting the archive. I looked at it and noticed that

Max Tegmark has never posted to this list. Do you know why not? Does he

read it?

It seems to me that the view that the quantum suicide experiment

could show anything comes from a misconception about measure. I know this

has been discussed before, but I'll give it one more shot.

If there are 900 people named Bob and 100 named Tom, all else

being equal, the effective probability of a typical conscious observation

to be associated with thinking one's name is Bob is 90%. That is the

measure associated with such observations. It does not matter if all the

Bobs are identical; they could be, and that should not alter the measure.

If it did matter, than it would be difficult to explain how, in

the MWI of QM, the probabilities could depend on the amplitude of the wave

function. (See my page on interpretation of QM.)

In terms of computers, and the variable resistance connections

proposed by Hal Finney, there is indeed an objective distinction that

should be made between independent vs. linked computers. It is not a

question of whether the AI 'feels a difference'; obviously, if it runs the

same computation, it must be able to report the same feeling to us.

The distinction is simply a question of how many different

implementations of that computation there are. As explained on my page,

one requirement for an implementation is that causal relationships must be

true between states of the computation, both the actual states and the

counterfactual possible states.

If the computers are wired together, they're no longer independent

implementations; it would be impossible to change the initial conditions

so that some computers would pass through a different set of states from

the others. If however the resistance is high enough and the digital

gates are still able to distinguish high vs. low voltage states without

interference, the computers will behave independently and have more

implementations.

Now, if the average Bob lifetime is 1/36 as long as the average

Tom lifetime due to a propensity of Bobs to commit quantum suicide, the

measure of Bobs drops to 25% of the total measure, and the total measure

itself is greatly reduced (though unobservable).

It works the same across time. Suppose the Bobs have a halflife

of 20 years. The effective probability of an observation being associated

with Bob before the age of 40 is then 75% of Bob's total measure. The

effective probability at age 1000+ is negligable.

Finally, I want to point out that surviving a quantum suicide

attempt does NOT provide any evidence for the MWI. The only way to have

such an observation is by luck, and it makes no difference whether it is

an effective probability or a real probability, it ain't likely to happen

again if you try another suicide.

- - - - - - -

Jacques Mallah (jqm1584.domain.name.hidden)

Graduate Student / Many Worlder / Devil's Advocate

"I know what no one else knows" - 'Runaway Train', Soul Asylum

My URL: http://pages.nyu.edu/~jqm1584/

Received on Wed Nov 18 1998 - 15:16:52 PST

Date: Wed, 18 Nov 1998 18:15:06 -0500 (EST)

On Mon, 16 Nov 1998, Wei Dai wrote:

Perhaps, but if so that's just an example of the point that I

made, that even if all possibilities exist there are still free parameters

in specifying how to count them up in order to make predictions. So it's

really not a problem with the idea I described, which is a simple (thus

interesting) choice for the free parameters, but rather it's a problem for

the more ambitious goal of having no information, which I currently doubt

could be achieved.

Thanks for posting the archive. I looked at it and noticed that

Max Tegmark has never posted to this list. Do you know why not? Does he

read it?

It seems to me that the view that the quantum suicide experiment

could show anything comes from a misconception about measure. I know this

has been discussed before, but I'll give it one more shot.

If there are 900 people named Bob and 100 named Tom, all else

being equal, the effective probability of a typical conscious observation

to be associated with thinking one's name is Bob is 90%. That is the

measure associated with such observations. It does not matter if all the

Bobs are identical; they could be, and that should not alter the measure.

If it did matter, than it would be difficult to explain how, in

the MWI of QM, the probabilities could depend on the amplitude of the wave

function. (See my page on interpretation of QM.)

In terms of computers, and the variable resistance connections

proposed by Hal Finney, there is indeed an objective distinction that

should be made between independent vs. linked computers. It is not a

question of whether the AI 'feels a difference'; obviously, if it runs the

same computation, it must be able to report the same feeling to us.

The distinction is simply a question of how many different

implementations of that computation there are. As explained on my page,

one requirement for an implementation is that causal relationships must be

true between states of the computation, both the actual states and the

counterfactual possible states.

If the computers are wired together, they're no longer independent

implementations; it would be impossible to change the initial conditions

so that some computers would pass through a different set of states from

the others. If however the resistance is high enough and the digital

gates are still able to distinguish high vs. low voltage states without

interference, the computers will behave independently and have more

implementations.

Now, if the average Bob lifetime is 1/36 as long as the average

Tom lifetime due to a propensity of Bobs to commit quantum suicide, the

measure of Bobs drops to 25% of the total measure, and the total measure

itself is greatly reduced (though unobservable).

It works the same across time. Suppose the Bobs have a halflife

of 20 years. The effective probability of an observation being associated

with Bob before the age of 40 is then 75% of Bob's total measure. The

effective probability at age 1000+ is negligable.

Finally, I want to point out that surviving a quantum suicide

attempt does NOT provide any evidence for the MWI. The only way to have

such an observation is by luck, and it makes no difference whether it is

an effective probability or a real probability, it ain't likely to happen

again if you try another suicide.

- - - - - - -

Jacques Mallah (jqm1584.domain.name.hidden)

Graduate Student / Many Worlder / Devil's Advocate

"I know what no one else knows" - 'Runaway Train', Soul Asylum

My URL: http://pages.nyu.edu/~jqm1584/

Received on Wed Nov 18 1998 - 15:16:52 PST

*
This archive was generated by hypermail 2.3.0
: Fri Feb 16 2018 - 13:20:06 PST
*