Obviously a huge source of ideas for the design and building of mechanical computers and a mech computer network is in the area of electronic digital computers and electric-mechanical ones (not to mention quantum mechanical ones). For instance, we can improve on Zuse's 1938/1945 mechanical memory (and Babbage's 19th-century 'store' or memory) by looking at electric and electromagnetic memory. Also, we can copy electronic chip logic gates and construct logic gates that use water instead of current, or use sunlight, or heat, or steam. It might be possible to construct a primitive type of microprocessor using fluidics or air currents in micrometre-scale tubes; or sunlight sent along micro-fiberoptic cables. The latter by the way would have the advantage of light-speed contact, as well as being a type of (semi-) 'solid state' device (few or no moving parts, which of course is one of the great strengths of the integrated circuit that contemporary electronic computers are based on).
But there also do need to be fast mechanical or non-electric analogues of relays and basic electric-mechanical motors, as well as chips.
But it would be a mistake to think of mechanical computation as simply a series of copies of electronic methods, and poor copies at that. Mechanical computation has its own forms governed by the nature of the forces involved in it. It is a plum, and digital electronic computing is an orange.
As a sort of heuristic device, we can assume that non-electric computation has the potential to be every bit as powerful as digital electronic computing -- it's just up to us to find out how. But we cannot forget that it will be a powerful plum, not an orange.
To add:
Give sources, links.
No comments:
Post a Comment