X-Message-Number: 5880
Date: Tue, 5 Mar 1996 15:01:47 -0500
From: <> (Jeffrey Soreff)
Subject: freezing debris, liquid phase tearing?

(This was originally an email response to a posting of Pete's in sci.cryonics.
 He has since suggested that I repost it here, hence this message.)

>I strongly agree that a scientific evaluation of cryonics is necessary
>in order to form any rational opinion about it. But, so far, all the science
>presented in these threads supports the feasibility of cryonics. Can you
>provide any scientific evidence that supports its infeasibility?

Um, have you looked at Mike Darwin's description of the damage produced by
current best available freezing practices?  He has some descriptions of
membrane tearing, loose cellular debris, etc. described in
http://www.access.digex.net/~kfl/les/cryonet/5050.html

While I have signed up with Alcor, and expect that some fraction of the
structural information in the brain will be preserved by current methods,
Darwin's observations of rips, tears, and free-floating pieces of cellular
debris being produced while part of the brain is still liquid implies that
some information is being lost.

I think that the presence of free-floating pieces of neurons approximately
1 micron in size implies that information is being lost for the following
reason:  Even with fully laminar flow, particles are still subject to
brownian motion.  This is basically thermal noise, and produces trajectories
that can't be traced back to their origins.  If I did the calculation right,
I think that a 1 micron cube will move at around a millimeter/second due to
brownian motion.  I'm not sure what the "step size" for the resulting random
walk is (depends on viscosity) but I'd be surprised if micron sized pieces of
debris don't move by many microns during the hours required for full freezing.

Once two pieces of debris have moved far enough to swap positions, we've
basically lost information on which neuron they originally came from.  Note
that this is a *freezing* time information loss, *not* a revival time loss.
Improvement in future technology don't fix it.

Now I'm not claiming that the extent of this information loss is catastrophic.
An axon could have many pieces carved out of it and still leave enough
connected pieces that we can reconstruct what it connected.  If, however,
a bundle of axons gets torn while the torn area is still liquid, and debris at
the tear moves around by a few microns, we can even lose information on
neural interconnection topology, let alone anything subtler.

Unfortunately, this sort of wet tearing that Mike Darwin documented looks like
a new information loss mechanism, different from
a) turbulence, which Ralph Merkle has shown should not occur
b) dry cracking of fully solid tissue, which arguably should not lose any
   significant amount of information.

Comments?

(Pete made a number of comments on ameliorating this information loss.  He
 noted that I've assumed that fragments can't be distinguished by their shape,
 and matched back to their original positions using that information.  I agree
 that I've assumed that, and I agree that using fragment shape data will
 reduce the data loss.  In addition, differences in local chemistry can help,
 at least where the differences in the fragments' components are large enough.
 Unfortunately, I suspect that a lot of the information about a fragment's
 detailed shape will get lost quickly, due to surface tension effects.  If the
 tissue was still so liquid that Mike Darwin observed stretching rather than
 clean fractures, the cell membranes must still be flexible enough to
 redistribute stress.  If they can redistribute stress, and the stresses get
 high enough to break surfaces, the surface boundaries after the break must
 experience enough unbalanced forces to distort their shape well away from
 their form immediately after the break.

 Pete also suggested that Ralph Merkle's rotor machine key extraction methods
 may be helpful in reversing this damage.  I'm less sanguine about how helpful
 these methods will be.  Partly this is an issue of computational cost,
 although here new technologies (Pete suggests quantum computers) will be
 helpful.  Mostly, however, the problem is just insufficient information.  If
 enough microstructure damage has happened that each neuron might have
 connected in several plausible places, you have 10^10 bits of missing data to
 recover.  I don't think that testing the behavior of the plausible
 reconstructions will bound the solution enough to recover many of those bits.
 Even if you add in known personal information about the patient, you will
 basically be trying to reconstruct gigabytes of neural data from perhaps
 megabytes of personal information and general knowledge of what a healthy
 brain acts like.  If we are quite lucky, there may be a lot of redundancy in
 the neural interconnect topology, but my personal guess is that we'll need
 to be able to trace most of a patient's topology in order to revive them.)

                                                      Best wishes,
                                                      -Jeffrey Soreff
standard disclaimer: I do not speak for my employer.


Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=5880