Re: Uploaded thoughts. (p.1)

--- In wta-talk@y..., "John Leppik" <jleppik@k...> wrote:

> The above discussion concerned uploading via the neuron-by-neuron
replacement technique. For me it is the only technique worth
considering
because it holds the highest promise for me of success. The others
would
seem to merely make copies of me, but I would die before or after.
Success
to me means that I would in fact be aware of the successful upload
because
of a continuity in my consciousness. Thus I would retain my
experiences,
know that I was uploaded and my life would go on in some computer.
>

We're on the same page here.

> So how feasible is this process? It would seem that most people
here, and
that includes many very intelligent and well-informed people, think
it is
quite feasible. The item above dismisses the difficulties as "an
engineering problem". Well I am an engineer, so let me try.
>

I understand your point, but perhaps it would make sense to answer
more fundamental question first, i.e. "Is uploading capable of
transferring an original mind onto a new substrate without the loss
of identity?". Only if the answer to this is "yes" it will make sense
for us to focus on the engineering details of this kind of procedure.

It is a good question, but impossible to answer at this point.
Just like we wouldn't be able to answer a detailed design of the
first nano assembler. It would be like trying to prove that since we
don't have the engineering details of that first assembler now,
nanotechnlogy will never be achieved. Remember that in order for you
to show that something is not possible you need to provide a proof
that it will never happen. Can you say, for example, that in 1000
years uploading or nanotechnology will still be a technical
impossibility? If you cannot show that, then it is possible, and
then it's just a question of time before it gets implemented and
becomes real. That's what I mean when I say that something is an
engineering problem. In other words, that something is not impossible
and sooner or later someone will figure out how to implement it.

And yet, without those engineering details, it is possible to come up
with some obvious observations, which serve as a foundation for new
ones, ultimately leading to obvious conclusions. I think this
theoretical bottom-up approach could also be successfully applied to
uploading.

> The hypothesis as I understand it is that somehow we will be able to
replace the neurons in a living, >conscious, working brain one-by-one
with
non-biological neurons, without damage, without loss, without >change.
Neurons are not tiny grains of rice in a bowl. As I am aware, each
has
hundreds or thousands of >miniscule tentacles, intertwined with other
similar entities which function dynamically in a managed
chemo->electro-magneto soup in a manner in which the software cannot
be
separated from the hardware.

I agree with George who says in response to your post that to make
gradual uploading work, one would need molecular machines. Neurons
are not tiny grains of rice in a bowl, however, nanotechnology deals
with machines capable of assembling and disassembling any structure
onto single atoms, and since neurons are made of atoms,
nanotechnology should be able to replicate them perfectly. All
miniscule tentacles are going to be huge compared to the
size of machines that will operate on them.

>What kind of process could possibly detect with certainty what is
there,
how it functions, how it is controlled, >how something non-biological
could
be devised to do the same, replace the existing with a newly
>manufactured
item in a >different technology and all of this without damaging
anything
>or disrupting the >operation?

Please see a hypothetical procedure of neuron replacement known
as "Moravec transfer". This should answer that.

> Even if the neurons could be replaced thus, the brain could not.
We would
now have, say a silicon brain, but how would it be powered and how
would it
interface with the body and at the same time dispense with blood,
electrical
and other juice flows? The replacement would have to be done with a
great
deal of re-engineering of the interfaces.
>

I don't think that the procedure you describe above is really the
optimal way of uploading or even that it is uploading. It's just too
messy, and it violates what one might call functionality equivalence
rule which basically says that in order for uploading procedure to be
successful, all relevant aspects of the mind must be perfectly
replicated in terms of their previous functionality. This means that
what we need first is the FEs (functional equivalents) for everything
that the mind's operation depends upon. Functionality of neurons, and
everything that contributes to the correct behavior of a mind. I
don't think the focus should be necessarily on simulating blood with
a perfect blood equivalent, for example. Just an FE of blood or of
other non-neurons necessary for a proper operation of the mind.
If this kind of approach is followed, then a certain point should
finally be reached when all FEs of all mind "parts" should be
derivable. Only then the process of uploading can really begin.

So no real blood with silicon neurons inside, just a special hardware
or even software, or a combination of both.

>Oh yes, there was that other item of my new brain being disconnected
from
my body. Would it not go as >apoplectic as I presume any brain would
go as
a head is chopped off? It all seems like a wet dream.

All aspects relevant to the operation of a mind need to be
replicated, so if it is determined that a mind cannot go on happily
without some concrete sensory sensations, then we will have no choice
but to provide a mind with this kind of input.

[and then your later post]

[You are right, but this replacement of atoms is an entirely different
process than the uploading replacement process, with nothing in common
except....
With the uploading replacement process you will be picking up this
functionality. How will it dunction in silicon? Deeeeeeh?
Showstopper??] ---

First sentence. It is my position that the replacement of atoms is no
different than the replacement process during uploading as long as
the right FEs form the basic mind units being transferred. Not only
that, atom replacement serves as a form of evidence that a mind is
not dependent on its substrate, and, more importantly, that
consciousness/identity of a person subjected to a constant
flow of those basic units that make up the mind doesn't appear to be
affected by the gradual replacement.

The obvious question to ask now would be whether or not the
phrase "identity doesn't appear to be affected" is enough to conclude
that uploading done this way preserves identity. Probably not. This
would require a definite answer for what identity really is. However,
the issue was not about defining identity, but how to preserve it. My
point is this. I don't know what identity is, but if it exists, I
definitely know where to find it and what to do with it. Let me offer
a perfect analogy that should explain all this mess. Imagine two
glasses, one empty, one filled with totally unknown liquid. The
totally unknown liquid is the identity/consciousness I want to
transfer to the empty glass. I have no clue what the liquid is, but I
know it is in the glass that is full. I do not even care what is the
nature of the liquid, and how I can define it. All I want is to pour
it carefully into the empty glass. When the liquid is transferred, I
still might have no idea about the identity of the liquid, except I
would be certain that whatever it is, it was successfully transferred
to the other glass.

Therefore in this sense the phrase "identity doesn't appear to be
affected" is sufficient to conclude that the procedure does preserve
the identity.

Second sentence. It is not necessary to create FEs associated with the
process of atoms replacing each other since this process isn't
relevant to the proper operation of the mind. If it were, it would
contradict the whole argument made earlier. In fact it was mentioned
earlier that there is some evidence that a constant "flow" of atoms
through the substrate doesn't affect the identity. Thus, the creation
of this particular FEs is unnecessary.

Slawek
Received on Tue Oct 22 13:56:26 2002