X-Message-Number: 107 From arpa!A.ISI.EDU!TKD Wed Jun 28 09:01:06 EDT 1989 Date: Wed 28 Jun 89 09:01:06-EDT From: Thomas Donaldson <> Subject: Re: CRYONICS #104 - Re: Brain Computational Memory Limits To: In-Reply-To: Message from "" of Tue 27 Jun 89 20:14:55-EDT Message-ID: <> Status: R Dear Ralph: If I understand your reply, you are saying that the SOLE significance of your paper is to provide convincing evidence that human beings will someday be able to make computers capable of vast and imposing thoughts. WILL ANYONE WHO DOUBTS THIS PROPOSITION PLEASE RAISE THEIR HAND? The issue I was raising was that of whether "computational power" is any longer a useful concept. If your brain-imitating computer is actually sequential, then of course we can use calculations about computational power. If we have to duplicate the network anatomy of the human brain to get a similar "power", the CONCEPT needs revision. Nor do I think that however powerful it will obviously follow that this duplicate brain wouldn't be slowed down by bottlenecks (catastrophically slowed down!) if it tries to duplicate a brain with the WRONG network architecture. Underlying this abuse of power lies Moravec's argument. Ten years ago Hans (sorry Hans!) was putting this same position, and predicting that enough "computational power" would be available in 1990. I am doubting that computational power is a good measure either of the time we will take to get AI or the time needed for computations of other kinds. That's all. Incidentally, though, we have brains to make choices, not to make computations. But that is (agreed!) a complex subject bearing on AI and other things. As for your abuse of power, it seems an awful lot of machinery to convince us of something we would already agree on. What audience is your paper designed for? I find that hard to place. Why power? Isn't that perhaps an awfully macho idea? Do we have a lot of BIG MEN going around comparing their muscle size and their computer power? Thomas Donaldson Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=107