I know quite a lot about digital systems and the history of digital electronics, but not so much about analogue, my only experience of analogue electronics is in the general fields of TV/Radio so thanks for the info.<quoted text>
Even bits in a computer are analogue, but the intermediate states are wasted so as to minimize the chances of error. BTW a lot of logic is tri-state, even in digital electronics. Early computers I worked with were decimal or bi-quinary. Digital computers could use other bases than binary.
A lot of communications technology uses 4, 8 or even larger number bases. That's how advances in modem technology were made. I remember the confusion over what could be the maximum channel capacity of a band limited phone line. Teletypes started off using 75 or 110 bits per second, and them modems going 300, 1200 or 2400 were used on phone lines. First it was said that there was an absolute limit of 2400 baud, and then modem manufactures started putting out modems with 4800, 9600 and higher bit rates. Well actually it WAS limited to 2400 baud, but one baud is a state change and for a while people were so caught up in the idea that binary=digital that they didn't take to heart the fact that a state change can contain more information than simply on or off. As soon as that unnecessary constraint was lifted communications started using non-binary states and modem speeds took off again.
There is some theory that estimates that the most economic base for a computer should be that integer number that is closest to the base of the natural logarithms which is around 2.7. In that case the most economical computers should be base 3.
Still I haven't seen any technological developments that indicate that analogue computing is headed for a comeback except in special cases.
Following my conversation with Kitten I spent some of yesterday (while I spent 10 hours travelling to and from a 25 minute meeting) looking at the analogue side of modern computing.
From what I now understand they are more simulation specific than calculation specific (and therefore less flexible) than their digital counterpart but considerably faster at performing their dedicated simulations.
Here’s a paper from columbia uni dated 2005 for an analogue computer idea that solves differential equations at the rate of 14Gflops
And here is one from Reading uni with a discussion of the differences in processing methods and ways of comparing speeds.