X-Message-Number: 25569 References: <> From: Peter Merel <> Subject: Technology Failures & The Anatomy Of Breakthrough Date: Tue, 18 Jan 2005 00:46:40 +1100 I agree with most of what Bob Ettinger writes in his response to me. His essay on probability for the most part as well. Bob issues a challenge at the end of it, however: > The important thing is that there have been very many successful > projects of technology, and very few or none that have failed. If you > can think of a failure, that qualifies according to our criteria, > please let us know. I'd suggest most spectacular technology failures are quickly shunned in favour of alternatives and workarounds, so will be under-represented in any history of inventions. Still a few come to mind: Tesla's Wardenclyffe - wireless transmission of power. Dyson's Project Orion - real space travel. O'Neil's Lagrange colonies - real space stations. Fuller's Dymaxion cars, houses, etc. Russell & Whitehead's Principia Pons & Fleischman's cold fusion etc. But for the most part Ettinger is obviously right - we have seen amazing progress in so many technological domains all at once it's foolish to bet against new technology succeeding in rapid order. We may qualify this in AI where the failures have to date completely dominated the successes. Billions of dollars were poured into he American Strategic Computing Initiative and the Japanese 5th generation project without obvious result. The massively funded AI labs at Stanford and MIT have almost completely given up on their original HAL-like development goals, settling into applicably profitable niches like bioinformatics, robotics, and VR. The figure of the wild-haired AI researcher declaring that this time, This Time, he's got it right, it will scale, it will speak, it will Wake Up, has become an icon in cheap SF flicks and a joke among venture capitalists. If we apply Bob's probability methods to AI then the outlook is not rosy. But having beaten that particular dead horse into butter let's consider the anatomy of breakthrough. If AI can't happen on T/VN computers, what have we got that's not T/VN? De Garis's career points the way here. FPGAs - CPUless computing - are an extremely popular technology without obvious Turing equivalence, so classical complexity results may not apply there for novel solution architectures. And quantum computers are indeed qualitatively different from massively parallel T/VN computing arrays, so the same goes there. It may be that the complexity bounds that to date stopped real AI in its tracks are about to disappear in a puff of paradigm reformulation. In which case all bets are off. My reply to Mark Plus's original question was not intended to deny the possibility of real AI or MNT. It was to discount Vinge & co.'s faith-based reasoning about same. The rational answer to Mark is, no one knows. Whether or not the next 30 years produces a technological singularity, it is extremely likely to produce new technology whose implications for good and ill we cannot imagine today. Cryonicists, as neophiles, should look on this uncertainty with pleasure. Peter Merel. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=25569