X-Message-Number: 8558 Date: Sun, 07 Sep 1997 21:10:06 -0700 From: Peter Merel <> Subject: Photocopying Picasso Thomas Donaldson writes, >a Turing machine won't necessarily qualify for practical reasons. I don't believe anyone has proposed building a Turing Machine. The point of the TM discussion is that no evidence suggests brains are not Church-Turing equivalent. If you've got such evidence, now would be a good time. >Since we are quite parallel ourselves, >whatever we are uploaded into should also be at least as parallel. Parallelism can be emulated trivially on a serial machine; non-determinism, however, cannot. It may be that our brains are non-deterministic to some limited extent, but even if this is the case there seems nothing to preclude our building non-deterministic quantum computers of a similar capacity. So it's hard to see that this concern amounts to much. Similarly on complexity: yes, execution time of algorithms is an area of vital concern to engineers. Presuming the mush in our heads is executing algorithms that would not be feasible to emulate in real time on a modern computer, we'll have our quantum processors ready to handle the load in due time. Determining the algorithms and building the QCs are tough engineering problems, sure, but only quibbles when it comes to feasibility. > We want to understand how brains work, > not just have speculations about it based on what we know of computing > We're going to have to get our hands dirty. Yes, the more evidence we can gather about the way our brains work, the better we will be able to both specify and generalize alternative implementations. But Minsky's book hardly detracts from such efforts, and neither does Minsky set out to detract from them, so what's the problem? > Even a little thought about real brains should convince most people that > they don't work like computers. [...] I believe even the > notion of "computer" will have gone through so many changes that we can > forget about our contemporary machines in thinking about this issue. > [...] Well, perhaps the problem there is that those who think it did > a humanlike thing are the ones lacking a good idea of humanity Um, that's fallacy ad populum, then "poisoning the well", followed by ad hominem. To clarify the well poisoning: we might well say that our notion of any theory at all will go through so many changes by some date that we can forget about contemporary results; that doesn't discredit the results. I understand you feel strongly on this issue, but let's keep it clean :-) >But until we have a far better idea of just what a person is and how they >work, and moreover, an idea proven by experiment, we aren't even at the >level of knowing what we want to do. You can take brains apart, analyze their chemicals, look at the interconnections between neurons, the number of dendritic branches, the quantities of neurotransmitters, the firing patterns prompted by various stimuli, etc. but this will get you no closer to understanding "a person" than a study of the mechanics of pigments, brushes, strokes, media and so on will lead to an understanding of an artwork. Furthermore, though a study of pigments, etc. is necessary to construct paintings of your own, to reproduce existing paintings a digital camera and output device are entirely satisfactory. As to what to do, many of us understand ourselves sufficiently to adopt the Merel Criterion as a standard; what uploaders want to do is plainly to create a machine with which they can reproduce their minds to an extent that satisfies the MC. The MC doesn't tell us how to do it, of course. > I suspect those who want to come back > uploaded into a computer think of them as mythical entities capable > of many different things impossible otherwise. Oh yes, trying to do something that's not presently feasible, terribly silly. Chasing the moon, like those silly astronauts. Or like those silly cryonicists. Dangerous Loons, each and every one of them. :-) Peter Merel. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=8558