One nitpick: the speed of a mechanical computer is not really that limited, except by technology and cost. There are, of course, certain factors like the speed of light, but it’s theoretically possible to make extremely fast mechanical computers that can compete with electronic ones. One method would be to use massively parallel processing, combined with nanotech, so that you have billions of microscopic gates and storage switches - imagine a single molecule as a storage bit.
Of course, this is not plausible today, but there have already been demonstrations of components. Also, this would not be applicable to a desert island scenario (at least not in the present).
Good point: if we represent dit and da as digits, and intercharacter and interword spaces as digits, then there’s no need to define an intra-character digit.
dit
dah
intercharacter space
interword space
So:
11132223111411132223111
Would be unambiguously interpreted as “SOS SOS”.
Well waddya know, Wikipedia’s wrong! who woulda thunk it!?
NETA:Also, the longer silence actually could be just two shorter ones. But if you want to go that far, you might as well realize iit’s all really binary, since the states are either sound on or sound off. A dit is really 10, a dah is really 110, a letter space is 0, and a word space is 00. (Thus SOS is 101010011011011001010100)
Two impertinent remarques,
Before ENIAC, the first digital computer was probably invented by Atanasoff, who needed to set up integrals for his physics research, and decided to create a suitable machine, this machine was put aside when he was called into the war effort, and possibly stolen by one of his students.
Otherwise, Martin Gardner for the fun of it, described an elephant powered digital computer in his Scientific American column.
Someone already mentioned computers based on fluid flow. Don’t airplanes already have simple fluid-based switching circuits? (What’s the most switches of any such “computer” in existence?)
