Learning, and Learning From, C+cean
Posted under Miscellanea at .
You might think — given a world in which there are uncountable ‘computer languages’ — that computers had evolved in the familiar way, from primitive valvifera to increasingly inquisitive and playful large-brained transistaria, maybe living off a foraging/scavenging diet supplemented with occasional catches of wild mice and trackballs, through a process of increased capacity for group relationships and pack hunting among the tape herds, developing the ability to communicate complex plans to each other, finally becoming able to consciously invent new words and grammars, write poetry, and construct spamming schemes.
But they didn’t. (Yet.) That’s not what computer languages are, it’s a trick of the randomly imprecise natural language we’re using; they are ‘computer control languages’ (CCLs). That’s a whole different thing. They are actually human languages, but formal extensions of our usual grammars, used to control computers.
There’s a kind of implication here . . . more interesting random imprecision . . . could it be that normal human languages are actually ‘human-control languages’ (HCLs)? Sometimes? Maybe. It’s probably more accurate to think that human languages are general languages usable for a range of functions. One of these might be controlling humans, but we tend now to attempt to control each other through legal mechanisms which are themselves — ideally — controlled with semi-formal legal language. Or just with money and/or weapons. Adverts are another example of an attempt to control human behaviour with unnatural (often counterfactual) language. Meantime we have also developed CCLs because, so far, we have to use more formal languages to control dumb machines, though the day may come when we can make smart machines that really are fully programmable with natural languages. (I await a formal proof of this with interest but little hope.)