X-Message-Number: 20581
Date: Tue, 10 Dec 2002 16:34:12 -0500
From: Thomas Donaldson <>
Subject: why mol nanotech can't repair us

Why Molecular Nanotechnology Cannot Rescue Us
 
 I will begin here by discussing some of the meanings of "nanotechnology"
itself. One meaning consists of ideas by which we can achieve various goals by

manipulating matter of the size of small molecules. Nanodevices doing this would
do such things as modify complex single molecules to plan. A second broader

meaning suggests that we might make far better machines of all kinds by building
them up from specially designed nanodevices, each of which plays its own
particular role. Though living things aren't really devices, in one sense every

living thing consists exactly of a combination of many kinds of nanodevices each

with their own role. Our future machines will have that same complexity, or even
greater: including such things as a true ability for self repair and changing
their form as needed or requested.
 
 This article criticises only the first meaning for nanotechnology, molecular
nanotechnology. That we will someday achieve many striking technical feats,
including repair of suspensions patients, with devices based on the second
meaning for nanotechnology seems quite clear. Such devices may be of any total
size, but only work because of the nanodevices which make them up. 
 
 For cryonics THE central problem is how to repair damaged brain tissue damaged
by freezing so that it thinks as closely as possible like the original brain.
Without such abilities, even if we "revive" a patient, that patient will not be
the original person but a very well built copy, starting its new life with no
memories or sense of any previous life. However this central repair problem for
patients stored by freezing comes very little from any chemical changes caused
by freezing (1), but instead by the breaking and disorder of neurons and brain
tissues caused by freezing itself. Our tissues, once broken, will not even
remain where they originally lay. In his Appendix B to CRYONICS: REACHING FOR
TOMORROW, G. Fahy describes the "fracture and separation of fractured halves of
cells, axons, dendrites, capillaries, and other brain elements by distances in
the millimeter range..." and "local (not global) ripping, twisting, and fraying
of the ripped ends of the nerve tracts by contraction of the brain cells and by
the push of extracellular ice, creating debris strewn gaps measured in microns
in both length and thickness". 
 
 Note first that neither of these serious injuries occurs on anywhere near a
molecular level. One major discussion of how molecular nanotechnology will help
repair breaks down the problem into 3 subproblems:
 
  1. Where are the atoms?
  2. Where should they go?
  3. How do we move them from where they are to where they 
           should be?
 
 If repair is to be done only by NANOSIZED devices, all 3 questions raise major
problems for nanotechnology. The problem is that those structures damaged by
freezing exist not on a molecular, nanosized level but at a level many times
larger. It's one thing to put together a brain molecule by molecule if we have
answers to Question 2; it's quite another to answer Question 2 itself when
presented with a frozen brain. Moreover, to USE that answer, we will again need
devices larger than nanoscale. Yes, I will emphasize that the devices used for
repair may well be COMPOSED of many nanodevices, but the individual nanodevices
making them up serve only as parts for a larger device. Repair on a molecular
scale ignores the real problems entirely. Nor does any amount of calculation
about the number of molecules involved, and the information amount taken up by
storing the location (and perhaps other features) of each molecule, really give
us any serious answer to the real, central problem of repair. Calculations of
the computer power needed to repair MOLECULES tells us even less.
 
 Here is an analogy for those used to electronic devices. Suppose that 
 someone, perhaps to prevent us from working out its design, takes a CPU chip,
strikes it with a hammer, and then mixes up the pieces by shaking it once. I
assume that the CPU chip is (relatively) quite large. It's true that if we knew
the CORRECT position of every atom in this chip,
 restoring it using nanotechnology becomes a relatively easy task. Yet to work
out just where its atoms should go, from the many pieces we have, raises many
problems. Most of these pieces won't be nanosized,
 but would be worn down so that we could not fit them together like some kind
of 3 dimensional puzzle --- and even doing such fitting together would require
structures larger than nanosized, or many nanosized
 parts connected by computer(2). The problem is worsened by the simple fact

that we can't assume that two pieces connect or not because their shapes connect
or not. Those shapes would themselves have been worn down by damage from the

hammer. And finally, note that pounding the chip with a hammer causes few if any
chemical changes.
 
 If we look only at molecules and their positions, then reassembly becomes far
harder. The computational problem consists then of finding out the original
structures out of which the chip (or the brain) is made given only information
about location of each molecule. The atoms in the broken chip, as different
elements, occur in many different positions and play different roles in each.
Our brains, too, consist of molecules, in much greater variety. Yes, we can
derive the position of nearby molecules, but our problem remains hard. If we
somehow know the position of hundreds or thousands of nearby molecules, then
it's much more tractable, but at the same time we no longer work only on
nanoscales... even if we do this on a computer (hardly likely to be nanosized
because of the computer power needed). Among other problems any chosen molecule
can correctly go into many different locations: if those locations were
calculated independently for each molecule, then somehow we'd have to ensure
against duplication. The molecules surrounding it provide clues at best, since
locations in a patient's brain differ from the correct ones.
 
 Considering the many different answers to the question of where a molecule
SHOULD be, the computational problems involved in answering Question 2 turn out
to be quite large. For a brain, we wish to recover its original connectivity,
clearly a feature at a much higher scale than nano. Except for certain special
neurons, we do not know a priori even how many synapses it will have (3).
Establishing that one large set of nearby molecules constitutes a synapse gives
only a start to the problem of recovering connectivity. While we might use

nanodevices to examine healthy brains, and from that examination learn much more
about possible molecular signs of connectivity; still, at a minimum, doing so
requires us to use concepts dealing with phenomena on a much higher scale than
nano. Nor could they be easily derived solely from a large list of brain
molecules and their locations. The whole of chemistry is based on but cannot be
derived from the physics of atoms; in some sense, astronomy itself is based on
but cannot be derived  
 from the physics of atoms. And to learn astronomy we begin not with devices
telling us about atoms but with telescopes to tell us about stars and planets.
 
 For brains such points become particularly important because IF our memories
depend on our brains' connectivity, then each of our brains must have DIFFERENT
connectivity than any other. Except for broad connections such as those between
our frontal cortex and our amygdala, the detailed connectivity of a brain would
need to be worked out for each invididual brain. Deriving our connectivity

simply from the locations of molecules does not look like an effective strategy.

While it's easy to work out connectivity in a whole healthy brain, doing so when
many connections are both severed and moved becomes far harder.
 
 I gladly agree that nanodevices could answer Question 1, though in doing so
they would act as parts of a much larger repair device: the HANDS of the
Supervising Computer. To give positions does not require much independent
calculating ability. Even now, very many biological stains exist to show the
location of particular biochemical molecules within a cell, including neurons.
Some of these stains can even work with living cells. Means to report location
may come sooner than many expect. The results of such stains, among other
points, also show vividly just how many locations a given molecule can have.
 
 For Question 2, so far as a powerful and inexpensive computer is required,
nanodevices again will play a role: but again, not singly but as assemblies of

nanodevices, each one with a role as minor as that of single electrical gates in
a semiconductor chip... but of course far smaller than our present electrical
gates (if we still use electrical phenomena in our nanodevice computer). To
answer Question 2 again requires work on a much higher level than molecular.
Just to recognize a synapse, especially if it's been damaged, requires assembly

of information from many nanosized locations. The Supervising Computer must take
all its input information and construct a 3-dimensional map of the damaged
brain. From that map it must then work out the correct locations of the
different brain pieces (I do not refer to molecules here because each of our
molecules will have many different correct locations). Even if it computes with
nanoscale parts, it cannot work at nanoscales in its computations.
 
 Finally, when we come to actually REPAIRING a brain, we want to link together
assemblies of molecules separated from one another by scales larger than any

nanoscale. Supposing for an instant that such devices do their repairs by moving
molecules only(4), they would still have to move those molecules to their new

corrected location, in the midst of many other similar devices working to do the
same thing. They would all need some way to tell their own coordinates and

distinguish one another. Basically, any guiding computer would also work out not
only proper locations but also paths to the repaired, proper location from the
faulty location, for each individual repair device. In no sense could such
devices act independently of one another.This basically makes them again a part
of a much larger device. And again, moving to an assigned location along an
assigned path requires only the computational resources to store that path, and
carry out a fixed algorithm to return to that path if diverted for some reason.
Paramecia come close to having those required abilities. 
 
 Here, too, nanodevices would play a role. Here is a simple way to see just
what kind of role that may be: no one claims that our lymphocytes, including
monocytes, granulocytes, and others, even though they are single cells and can
wander throughout our body act as independent devices. Even though very small,
their activity depends not on any independent systems but on hormones,
antibodies, and other chemicals in their milieu. Without that guidance
lymphocytes do nothing. We might well be repaired by nanodevices, just as
lymphocytes help protect us from diseases, but those devices could only succeed
as part of a much larger repair machine which guides them by chemical,
electrical, or other means. In a sense, the nanodevices putting each part back
to its correct location act here as hands modifying the damaged brain, rather
than hands feeling about in it for locations. 
 
 Those proposing molecular repair by nanodevices do make one important point:
if the structures to be repaired are separated from one another, no special
problem with extra heat will arise. However doing so only increases the problem
of putting together the connected network of neurons which makes up our brain:
from microns we might go up as far
 as centimeters. Since we ourselves do the separation, however, such a system
raises no additional computation problems, only the problems of transport.
 
 I will finally emphasize here that I am not an arguing that repair is
impossible, nor that nanotechnology will play no role in our repair. I am
arguing against a particular THEORY of repair. It's very fortunate that this
theory so far has not influenced the major work on suspension in progress from
other groups. It may turn out, though, that simple confrontation with the
problems of repairing real suspended brains would cause a shift in the ideas of
many about how such repair could be done. 
 
 (1) As for chemical changes, the few that arise may well be repairable by the
cells themselves, once we have put them back together.
 (2) Just what is a nanodevice? Our hands are collections of many nanodevices,

bound together physically and chemically. Is a fine enough probe at the end of a
device to move it a nanodevice? Much more pointedly, is a collection of
nanodevices with activities connected by one large computer to be considered a

nanodevice? What if their activities are connected by chemical "hormones" rather
than a separate computer?
 (3) Currently popular theories of how memory works suppose that our
 memories come from the connectivity of our neurons. This neural connectivity

is necessarily at a far larger scale than nano-. These theories may change, even
soon, but that hardly means that the problem will become easier.
 (4) Given that we are reassembling fragments much larger than 
 molecules, it would be far more efficient to have devices to move the
fragments... even though such devices obviously couldn't be nanosized.

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=20581