George Gilder’s Microcosm. The copyright date is 1989. I checked it out of the library shortly after it was published, and I have been meaning to buy a copy for a while.
When I first read it, I was struck by his emphasis on human ingenuity as more important than physical resources. That has stuck with me ever since. It influenced my decision in April of 1994 to quit my job to start an Internet-based business. It influenced my economic views, where I emphasize intangible assets.
I was also struck by his echoing Carver Mead that in the future analog computing would experience growth. It seems to me that this did did not happen, although one can argue that computing devices are increasingly linked to analog sensors.
He also did not mention the Internet. But he did stress the potential of computers as communication tools, and he understood that the cost of switches was falling relative to the cost of wires, which is the technological reason that the Internet took over the communication process. And of course, by 1994 he was writing long articles for Wired extolling the Internet.
He prophesied a decline of centralized power. On p. 361, he wrote,
The military threats of the future come not from mass mobilizations for territorial expansion but from nihilist forces of terrorism and reaction. . .all democracies will face the challenge of using new methods of electronic surveillance, security, and control, together with new non-lethal weapons, without seriously infringing on the rights of law-abiding citizens.
It all seems obvious now. It didn’t in 1989.
Deep Learning is, practically speaking, analog computing on a digital substrate. It certainly can’t be called algorithmic in the usual sense of the word “algorithm” – a well-defined intelligible procedure executed step by step (it is immaterial whether steps can be executed in parallel), modeled and named after Euclid’s greatest common divisor algorithm.
“‘The military threats of the future come not from mass mobilizations for territorial expansion but from nihilist forces of terrorism and reaction’….It all seems obvious now”
Actually, it seemed obvious about 15 years ago, but not so much now. China in Taiwan and the South China Sea and Russia in Eastern Europe and the Ukraine seem like mass mobilization for territorial expansion and reflect long-term, geopolitical rivalry rather than nihilist terrorism and reaction.
The West seems stymied about how to approach these challenges, or even to admit that such challenges exist. The main stumbling block seems to be a postmodern twist on the notion that one should avoid repeating mistakes. The twist seems to be an unwritten rule that the West must never under any circumstances repeat its victory in the Cold War. Winning the first Cold War was hard enough. Requiring that we re-win it with a completely different approach is proving to be even more challenging.
Nothing analog about Deep Learning, just maths.
That said, Mead was a major proponent of using analog neuromorphic computers. This is a rival model to DNN.
https://web.stanford.edu/group/brainsinsilicon/about.html