> First, Dyson doesn't explain what he means by analog computing, other than a basic "operates on real numbers and continuous functions." I think he means that

> First, Dyson doesn't explain what he means by analog computing, other than a basic "operates on real numbers and continuous functions." I think he means that neural networks are not binary, but use analog communication between the neurons. We simulate that digitally, but maybe a system that uses analog by nature is better for artificial neural networks. No idea if this is actually the case, or if this is what he meant ;).