king@Kestrel.ARPA (06/27/85)
I'm considering writing a letter to SA about their article in the June issue claiming that computation may be possible with arbitrarily low energy expenditure. By the way, is the author on this list? The following will be incomprehensible to anyone who hasn't read the article. The problem I have is the idea that the enzyme turing machine would be unreliable because of possible non-enzyme-mediated state changes, but that the macroscopic unit was somehow immune to this. I think that the macroscopic Brownian motion unit is vulnerable to tunneling. As the sizes of the machined parts gets larger, computation with a given amount of power gets slower, allowing more time for the rarer tunneling events to occur. Tunneling is not a strict analog of non-enzyme-mediated reactions, but it is tempting to think that there is a relation to a system's resistance to unusual events and the amount of force required to drive it. Before I do the math, does anyone have any comments?