X-Message-Number: 7867 From: (Thomas Donaldson) Subject: Re: CryoNet #7857 - #7861 Date: Sun, 16 Mar 1997 10:20:44 -0800 (PST) Hi again! This is Chapter 2 about uploading and cryonics, in case anyone wonders. The first point to keep in mind about real cryonics is that patients are frozen in all kinds of conditions. They may be frozen without cryoprotectant, because we got to them too late. They may also be frozen with the very best techniques available, because the suspension team was set up by their bedside and ready to go when they were pronounced. Not only that, but right now the different cryonics societies (primarily Alcor, CryoCare, and the Cryonics Institute, not necessarily in that order) have been working on better means of suspension. The first two strongly favor vitrification, and there is a very good chance that vitrification can be accomplished in just a few years. (That is what Wakfer's PROMETHEUS is about, and why he feels strongly about this issue. It would be absurd if we fail to do this work and allow ourselves --- OURSELVES --- not to be suspended by the best methods. Among its other features, vitrification completely changes the problem of revival. There are no more cracks, no more damage to cell membranes, etc which might make revival a difficult problem. We may find some chemical damage occurring on the way up or the way down in temperature, and it may not turn out to be easy. I expect a lot of thought will have to go into finding the very best way for vitrification. 21st Century Medicine has funding now to look into cryoprotectants for just that reason. However, it does change the problem. A lot. It should become relatively easy to revive a vitrified brain, among other points. No issue of uploading versus repair need ever arise. Sure, we'll have to provide our neuropatients with new bodies but we can keep them in storage until we solve that problem. I've said frequently that I think it can be best solved by developing means to get the heads (while unconscious) to grow new bodies, and that I don't think this will require megatons of new knowledge: "just" an understanding of growth and development, and how to control it. Does this end the issue of uploading versus repair? Not at all. The problem of those who are NOT found or suspended in time for vitrification to work on them will certainly remain. Sure, parts of that problem can be solved very soon: Hara Ra is working on a system which will call your cryonics society if you stop moving for any reason (just electronics and sensors, available off the shelf). But this issue will remain. Do I want to claim that IN ALL CASES repair will arrive earlier than means to upload someone into a computer? No. It would be foolish to do so, considering the range of possibilities. I do believe, though, that for those suspended even today, for whom their suspension went reasonably well, repair will be a better choice. Uploading raises lots of problems which its proponents have given no sign of confronting. One of the worst problems is that most ideas to do so are destructive and can be done once only --- which means that whatever is done must be done to a very high level of perfection. And right now we have lots of errors etc in our computer programs and our machine design --- which of course we try to solve before we use them, and if we've already started to use them solve them as soon as we can. But considering the amount of information involved, the likelihood of error increases. And the other problem I mentioned still stands: we're not going to be able to upload brains until we understand much better just how brains work. And that understanding will bring with it a much better understanding (and tools, too) of how to repair damaged brains. Finally, about nondestructive reading of a brain (for John K Clark): The problem is to read out the brain at the proper scale. You need to be able to see every individual neuron, and the synapses on it, plus the chemicals involved. The neurons should not show up just as points, they should be large enough to show distribution of synapses and distribution of nerve transmitter chemicals (and possibly receptors, too). This is much more than just a picture of our brains. While we can use X-rays to look at pieces of brain (thin pieces, sliced off) this will become a destructive readout. That is why I suggested what is basically a nanotechnological method for readout. And finally, about modelling brains: This is an amplification of my ideas, I agree with what Arkady said in his shorter message. First of all, a model or simulation is not the same as the thing itself, no matter how good it may be. What we want to make is something which works like a brain, not just something which seems to work like a brain. Yes, I am sure I may get some disagreement here, but I do not expect a simulation of me to be any more conscious than a photograph of me. And for this the computing power simply is not an issue. The main issue I see is that of making devices which work like neurons --- not in a superficial way (we can do that now) but completely, not only in terms of all the different nerve transmitters and their receptors but in terms of an ability to grow and change. One major feature of neurons is just that they do change their connections and even their shape over time. If we had chips for our computers which did this, we would open our computers to find them slowly moving about and changing their connections. I note that a very large and fast computer, with lots of memory, might simulate this to some degree (see comments above), but unlike real neurons there would be limits on just how much movement and change these computed neurons might have (and remember, we've got lots of neurons). Even for just a simulation, you would find yourself programming in all kinds of tricks to save memory, and ultimately you'd still run up against memory limits. That would still be a simulation, too, not the real thing. One major design and programming problem here is simply that even though your neurons don't adopt every possible set of connections and conformation, you must somehow make it possible for them to do so. And connections, in particular, may be added, the number does not stay constant: not only that, but where and to what and the chemistry of the added connection are likely to be important. For simple animals which never learn very much I think this could be done. For human beings I doubt it, simply because of the number of neurons we have. I'll also add that I see no physical reason why devices which do behave like neurons cannot be built, and thus why brains cannot be built. But they would not be computer chips. Perhaps they will be built, if only to produce robots much more like those in science fiction, capable of acting and perceiving like human beings. Long long life, Thomas Donaldson Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=7867