Nineteen ninety-two and the unificafion of the European market will soon be here, bringing in a storm of problems whose solutions will require wetware, because they will involve semantics. The story of the computer program that translated: "The spirit is willing but the flesh is weak" into Russian and back to English, ending up with "The vodka is fine, but the meat is rotten," is probably apocryphal but the problem it exemplifies is real.
Some semantic confusions might be acceptable. Motorola chipset fans would probably not mind if a mention of the orthogonality of the 68OxO instruction set was rendered with the sense "Motorola has the right angle." But think how confusing technical documentation would become if "number" was translated as though it was the comparative form of "numb;" i.e., "senseless." Then every reference to number would become more senseless.
Sometimes dealing with numbers makes people more senseless. In a recent column, Jeff Duntemann told of a chain-smoking nuclear power opponent who wanted to see "the inventor of radiation" jailed. I'd like to offer a defense of the smoker, to make her position a little less -- or more -- senseless.
The idea is that there is more to risk assessment than comparing probabilities. There has been some research whose results suggest that when people assess risks, they consider not only the magnitude of the risk but also other factors affecting its acceptability. So even if the chain-smoker understood the relative safety records of the nuclear power and the tobacco industries (which I admit is questionable), she could well have regarded the smoking risk as more acceptable on the grounds that it is taken willingly, not imposed from without. Even that position can be questioned, but it shouldn't be dismissed as mere innumeracy, which I'm afraid too many supporters of nuclear power do. (I'm not talking about Jeff, though.)
And what is the probability of deliberate sabotage of a nuclear power plant? I submit that a sufficiently determined terrorist organization could locate key employees and put sufficient pressure on them to achieve any desired degree of damage. That fact doesn't mean a whole lot, given that terrorists could just as easily sabotage baby food, drinking water, airplanes, or anything else. But it does put the smug, so-precise nuclear safety record figures in a gritty real-world perspective. The risks of nuclear power would appear, unfortunately, to involve the psychology of the terrorist mind.
The point is: Numbers have a semantics as well as a syntax.
And they have a pragmatics. Often we want to use numbers for a purpose. Usually they can be made to cooperate. The latest IEEE salary survey figures showing San Francisco-area engineers making 25 percent more than Johnson City, Tennessee engineers, don't help a Silicon Valley engineer trying to get a raise unless he first helps the figures a little. But with just a little help, the numbers can be made to serve the engineer's purpose; for example, by expressing the salaries as percentages of the mean (not median) cost of a house. Of course, that might just convince him to move to Tennessee.
"Human interface design" is a cumbersome name for one of the most subtle and important issues in system development. My cousin Corbett has looked for years for a better term, but everything he came across seemed even more ungainly or ugly than human interface design (FrontEnding?). It was only, he says, when he read the "Real Time" column by My Pal Tyler in the June issue of ESP magazine ("For People Who Don't Need No Stinkin'Interface") that he finally found his neologisms.
Tyler was expanding on the metaphor of integrated circuits as jelly beans. Although (as Corbett pointed out to me) Tyler faded to mention that the metaphor logically ought to run in the other direction today (there are now more ICs than jelly beans in the world, so we should refer to jelly beans as ICs), he did announce the Era of the Jelly Bean Computer. Buy a few jelly beans and you can put together a jelly bean PC clone.
The term "jelly bean computer" combines two figures of speech: The obvious metaphor of the jelly bean for the IC and the metonymy of the part (the ICs) standing for the whole (the computer). The appeal of the term comes from the closeness of the two figures of speech to actual fact: The chips are (at least) as common as jelly beans, and the PC clone described is little more than a few ICs.
So what kind of human interface would a jelly bean computer have? Corbett asked himself. Applying metonymy, Corbett decided to represent the entire human interface by a keyboard, and what kind of keyboard goes with a cheap PC? A chiclet keyboard, of course. Hence Corbett's new term for human interface design: Gluing chiclets to the jelly beans.
He hopes it will catch on.
Copyright © 1989, Dr. Dobb's Journal