Re: A calculus of personal identity

From: Bruno Marchal <>
Date: Fri, 30 Jun 2006 19:42:16 +0200

Le 30-juin-06, à 15:19, Stathis Papaioannou a écrit :

> I have the subjective experience of being a person persisting through
> time because I feel that I know in a 1st person way what I did in the
> past. If I really did know in a 1st person way what I did in the past
> I could not possibly doubt it, just as I cannot possibly doubt that I
> am having my *present* experience.

All right.

> However, I cannot be sure of my memories of the past just as I cannot
> be sure about someone else's experiences: I can only have 3rd person
> knowledge in either case.

I would say 3rd belief, reserving knowledge for the first person (may
be plural).

> It could be, for example, that I have been brainwashed and my memories
> of the past are partly or completely false memories. 

There is no false 1-memories. Only an association between some 1-memory
and some 3-reality can be false. If someone succeeds in implementing
correctly (more than just coherently) false beliefs (like I am Napoleon
just after Waterloo), then I will believe correctly that I am Napoleon
and that I have just lose a battle, almost by definition. I will have
to go in an asylum, sure, but my
1-memory of the past is correct given that they have been correctly

> >
> I agree if you mean by "future" and "past" 3-future and 3-past. 1-
> past 
> >
> and 1-future is not extrapolation thy are feelings continuously lived 
> >
> in a lasting present. I can no more doubt of my feeling of past than I 
> >
> can doubt of a headache (say). Even if time by itself does not exist at
> >
> all (which is the case with comp). The extrapolation would reside only 
> > in some third person projection of that time, space, ... (I think we 
> > agree, the problem could just be the term "illusion").
> I'm not sure if you're saying what I was saying above by
> distinguishing between 1-future/past and 3-future/past.

I think so.

> The relationship between different stages in a person's life - how
> far apart two different experiences can be and still belong to the
> same person - is complicated and necessarily vague. If we allow that
> in principle anyone can change into anyone else, how can you pin down
> this relationship with any rigour? 

To understand the consequence of UDA, I try to no put more rigor than
needed. Eventually those relationship will appear in mathematical form
with the lobian interview. Self-reference through diagonalization will
do the work, but this is needed to extract physics from numbers, not to
understand we have to extract physics from numbers once we assume comp.

> >
> > such as believing themselves to be moments in the life of a single 
> > > individual, having memories or quasi-memories in common, and so on.
> >
> > If I split into two that presents no problem for the 3rd person POV 
> >
> > (there are two instantiations of Stathis extant where before there wa
> s 
> > > one) nor for the 1st person POV (each instantiation knows it is 
> > > experiencing what it is experiencing as it is experiencing it).
> >
> >
> > OK.
> >
> >
> >
> >
> >  A problem does arise when I anticipate the split (which one will I 
> >
> > become?) or look back at the split (*I* was the original!); there is 
> > > no correct answer in these cases because it is based on 3rd person 
> >
> > extrapolation of the 1st person POV, which in addition to its other 
> >
> > failings assumes only a single entity can be extant at any one time 
> > > (only a single 1st person exists by definition, but multiple 3rd 
> > > persons can exist at the one time).
> >
> >
> > This is a little weird. You say there is no correct answer, and then 
> > you give the comp-correct answer.
> > The first person is indeed just NOT first person-duplicable (unless 
> >
> some added artificial telepathic trick, but in general I talk only on 
> > the usual simple teleportation or duplication).
> There is an unambiguous 3rd person descriptive answer, but no such
> unambiguous 1st person answer.

I think there is, once assuming comp.

> We can still talk about 1st person expectations, which I agree is the
> important thing for the subject.

Yes, physics will arise from that.

> >
> > This is not to say that my mind can or should overcome [Lee Corbin 
> > > disagrees on the "should"] the deeply ingrained belief or illusion 
> >
> > that I am a unique, one-
> track individual living my life from start to 
> > > finish,
> >
> >
> >
> Here you really talk about the third person extrapolation, so I agree 
> >
> with you. But the first person is not deceive by its feeling of living 
> >
> uniquely in time and space. It could be dangerous to say so, because it
> >
> leads to (materialism) eliminativism which eventually conclude that the
> >
> whole first person thing is an illusion. This leads to a deeply wrong 
> >
> sense of "human"-
> irresponsibility. Well, it is a negation of the first 
> > person. I can be sure it is wrong, as I bet you can too.
> I would say that the 1st person experience is *not* an illusion in
> any sense of the word.

All right.

> It is the very opposite, in a way: the most real thing, which cannot
> be doubted.


> But extrapolating to other people or other selves in the past, future,
> coming out of the teleporter or whatever, that is another matter.

True, but then we can assume hypotheses and and we can reason, in a
pure third person way, that is doing (refutable and thus doubtable)
science. I guess this is what we are trying to do right now.

> >
> OK but the fourth step of UDA is just asymmetrical duplication without 
> > memory loss.
> >
> I don't count the destruction of the original+its personal diary as a 
> > memory loss, giving that the memory is thoroughly conserved in the 
> > reconstitution (by the hypothesis + default assumptions on the 
> >
> rightness of the substitution level and the reliability of the doctor's
> > work, etc.)
> > What about that?
> > Oh.  I see you answer this below ...
> What I meant was something like this. You are duplicated via
> destructive teleportation so that two copies are produced in separate
> locations. One copy has 40% of its pre-duplication memories missing,
> while the other has 30% of its pre-duplication memories missing and
> 30% of a stranger's memories implanted. What is your expectation of
> what is to happen to you as you enter the machine?

Well, I see that point, but this is precisely why I have not allowed
memory loss in the duplication and multiplication experiments occuring
in the UDA. I say yes to the doctor provided I have reason to bet that
the copy will be "numerically" identical, at the substitution level of
This is Independent of the fact that it is highly probable such amnesy
can help to have an idea of the reason why comp histories can fuse, and
this helps to make already comparison between comp and the quantum
(erasure). For your experiment, I wish the candidate just say "no" to
the doctor.
I am not even sure it makes sense to talk on 40% of 1-memory, unless
you model the memory by sequence of W and M in iterated
self-duplication, so that in some particular simple case, we can begin
to analyze, but then this is not necessary to get the reversal.

> I should clarify, what I meant was not that I disagree with step 7,
> but that I find it difficult to understand. Going over everything you
> have said in this thread, I think the only thing I really disagree
> with is your insistence that we can have 1st person knowledge of our
> past. I don't know that it makes a big difference in the final
> analysis, but I think it is neater, simpler and still in keeping with
> all the facts to say that the 1st person is necessarily tied to the
> present.

Imagine you are drinking a cup of coffee, and just after I implement in
you a 3-false memory of "I have just drink tea". So your incorrigible
apprehension is that you have just drink tea. This is first person
knowledge and it cannot be false, or more precisely it can only be
1-false and it has to be
1-true. It can be 3-false.
If I insist that we can have 1st person knowledge of the 1-past, it is
perhaps due to the fact that I define (in first approximation) the
"consciousness-present" by a memory of some past (like a FINITE
sequence "W W W M W M W M M M W" in 11 repeated duplications: note that
this sequence is a pure first person uncommuncable one, the proposition
"Bruno is this one" admits 2^11 -1 counterexamples, and the proposition
"Bruno-WWWMWMWMMMW is this one" is (here) tautological.
I think I agree with you if you mean 3-past by past. But 1-past is just
the main ingredient for having any reasonable notion of 1-present. I
can not doubt present feelings, but I cannot dissociate them from a
quasi-infinities of "near-present-feelings" which are 1-person building
block of my 1-present.

> >
> Perhaps you have not enough appreciate the importance of the invariance
> >
> of the first person experience when reconstitution delays are added. It
> >
> did help people to explicitly recall the delays in the drawing 5), 6) 
> > 7) (and 8)).
> >
> Unless I am deadly missing something, it seems to me that what follows 
> >
> "We are almost done ..." can easily (?) be deduced from the six first 
> >
> steps. My experience with people who got problems in the seventh step 
> > is that they don't have really appreciate step 5).
> > It would help me if you could tell precisely where is the problem at 
> > the end of step 7; so let me quote it:
> >
> >
> >
> <<We are almost done. Indeed, let us try a simple “physical experiment”
> > like dropping a pen. With comp, when we are in the state of going to 
> > drop the pen, we are in a Turing emulable state.>>
> >
> >
> > All right? This follows directly from the comp hyp.
> Yes, that part is fine.
> >
> <<Our more probable consistent extension is undetermined by the 1-
> comp 
> > indeterminacy on all the “reconstitution” of that similar states 
> > appearing in UD* (the infinite trace of the UD).
> Sorry, I don't understand this sentence.

Informally, a consistent extension is a next possible (and living)
observer-moment. I mean, not a culd-sac "world". Now, when you drop
the pen in front of a concrete working and never stopping UD, to
compute correctly with comp your next OM, you need to consider all
computational histories generated by the UD which going through your
"present computational states" (which exists by comp). This gives the
uncertainty space on which you have to isolate a (relative) measure of
uncertainty (possibility, plasuibility, credibility, probability, ...).
And then compute. Of course this is impractical, like quantum field
theory (for such pen dropping problem). But this explain why physics,
once comp is assumed, should be reductible to computer science. Physics
have too emerge from a sort of sum on (infinite) computations.

> > All right? This follows from the definition of the UD. And the UD 
> > exists once we assume Church Thesis, and UD* (the trace of the UD) 
> > exists thanks to Arithmetical realsim. Rememeber I defiune comp by 
> > mainly CT and AR (+ the "yes doctor").
> I think I understand this, but not the sentence to which it is
> referring. Am I right in thinking that the UD is just a way
> to generate all the computable functions (in view of the fact that
> they are not enumerable)?

Almost correct. It is ... "in view of the fact that they are not
*recursively* or *computably* enumerable".
Also, less ambiguously, the UD generates the computations of the (total
and partial) computable functions. This will be perhaps clearer after I
answer Tom and George in the "puzzle thread".
(Hopefully next week!)

> > This follows from 6, and the invariance of the uncertainty measure, 
> > notably for the arbitrary delay---including the null one, and the 
> >
> infinite set of states appearing with a arbitrarily large delay in the 
> > running of the UD[14]. This gives a huge set. >>
> This I understand.
> > All right? You almost just said exactly this in your today's 
> > (29/june/2006) message to Brent Meeker.
> > And from this you can already conclude that comp entails a reversal 
> >
> between physics and computer science/
> number theory. The next sentences 
> > are there just for making easier the interface with the mathematical 
> >
> interview of the lobian machine, which is needed only to make explicit 
> > the manner we have to follow for extracting physics from computer 
> > science.
> The reversal between physics and number theory is something you have
> mentioned many times. Informally, I can see and accept that the
> universe may be Turing-emulable (if this is what you mean)

OK, I think you miss something here. Which explains perhaps my problem
with your use of "past" and "future".
Let me oversimplify my "comp-position". What the UDA is supposed to
show, is that the "real and correct" law of physics emerges from a sum
over infinities of infinite computations. From this it is not even
clear if any notion of 3-universe remains relevant. But more can be
said: whatever such universe could be, it CANNOT be turing emulable. To
emulate it "correctly" not only you would have to emulate in finite
time the whole infinite work of the UD, but you should be able to
collect your first person "appearances" in the UD, which is completely
Only the first person themselves are confronted with such an
uncomputable thing, because they are not aware of the delays, and they
does not know in which computations they appear. They in fact
distributed in many of them, and the physics emerge from that

> but I don't really see how the computational hypothesis about minds
> leads to this result formally.

What I show, is that if "I" am a machine, then my 1-neighborhood cannot
be described by a computation. It is the point of the 1-indeterminacy.
Physics is reduced to the computing of that 1-indeterminacy which
relies by comp on computer science/number theory, exclusively. All comp
3-histories are embedded in arithmetical truth (even in a "tiny" part
of it). The 1-histories emerges in a non computable way.

Another consequence: the observable universe cannot be a quantum
computer. Why? because a quantum computer is turing emulable, and a
priori with comp, the 3-universe cannot be turing emulable.

This does not mean the quantum theory is wrong (a (sub)quantum UD,
generated by the UD, could somehow win the "measure" battle; but then
it will not describe a physical world but an entire "multiverse"
(collection of interfering histories). In that case 1-indeterminacy
would resume into its quantum counterpart. And it is still an open
problem to define what a quantum multiverse really consist in. With
comp, the quantum would just describe the probability of the available
relative computational paths in our neighborhood derivable from our
observations, but that would be 1 person (plural) constructs. Number's
*shareable* (and "parallel") dreams.

I hope this helps. I feel you are close to the reversal. I'm aware it
is tricky.

I must go, so I have not the time to add spelling mistakes :)


You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to
To unsubscribe from this group, send email to
For more options, visit this group at
Received on Fri Jun 30 2006 - 13:43:25 PDT

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:11 PST