The Psychology of Everyday Things

by Donald A. Norman

Fools Can Be So Ingenious

This doesn't sound like much on the face of it, and it's reasonable to ask what Cog. Sci. has achieved since, say, 1956 or 1957, especially since its sibling, Artificial Intelligence, has gone from one disappointment to another. If the accomplishments are substantial, they should be better known; if they are not, well, these are hard times, and perhaps we should think about letting some of the professors go, and teaching the (public domain!) works of William James and Ernst Mach.

Donald Norman was, when he wrote this book, a professor of cognitive science at UC San Diego (I believe he's since gone to work for Apple), and The Psychology of Everyday Things nicely does three things at once: introduce the new knowledge gained by his discipline, document our inability to make good gadgets, and show how the first can help fix the second. The combination is shrewd, because while much of cognitive science sounds trivial (so does all for sociology, for that matter), the horror stories make it clear that many of our daily frustrations stem from ignorance of these supposed trivialities. (As for sociology...) Norman shows quite convincingly that design in ignorance of these facts is very rarely good, and almost as convincingly that taking them into account will improve matters. (This point is important: sometimes knowledge just substitutes learned for ignorant error.)

There are innumerable books on design, most of them pedestrian or horrible or both: and almost without exception, even the very best of them are written from the point of view of designers. Norman consistently looks up from the receiving end of the drafting table, from the viewpoint of the user, surrounded by incomprehensible VCRs, intimidating remote controls, arcane microwaves, computers as recondite as Mayan glyphs or Linear A, thirty thousand identical buttons --- to say nothing of doors, faucets, and light-switches. (We'll get back to the light switches.) A Stoic might argue that daily frustrations are a good thing, or even say there'd be fewer if we lived simpler lives. Norman, on the other hand, sees no reason why everyday gadgets ought to befuddle everyday people --- much less professors with engineering degrees from MIT. In any case, sort of bad design which keeps us from programming our microwaves cannot be met with just a stiff upper lip when it shows up in the cockpit of an airliner, or the control room of a nuclear power plant. Designers, in his phrase, ``worship false images,'' the look of things, and it is no wonder that ``prize-winning'' is, from his pen, a term of abuse. Norman manages to show how the current sorry state of affairs may be improved, limn the outlines of his science, and give us new eyes for objects, all at once: and if he is sometimes meandering, if he sometimes sounds like a man holding forth over a beer or two, he is unfailingly clear, easy, and worth listening to. (I suspect he's a happy drunk.)

An example which may serve to illustrate the method. One of the major notions of cognitive science is that people make mental models: which is jargon for saying that we have ideas about how the world is put together; furthermore, these ideas guide our action. As I said, much of this sounds trivial. But consider: unless we have a fairly accurate model of an artifact, it is exceedingly unlikely that we will be able to use it effectively; the better our model, the better we'll know how to use it. Therefore it behooves designers to make sure we get a good model from the artifact itself (since they aren't around to explain it). Looking in front of me, I see a few things of which this can reasonably be said: a pan, a clipboard, sunglasses. (Observe the constraints which make sunglasses --- despite themselves, as it were --- transparent. You can't put them on backwards or upside-down.) It is significant that they are all anonymous, and rather old. Turning behind me I have my brother's boom-box (``editing/B.P.''? ``MIC/I''? ``NHIGH/II''? ``LINE IN''? ``SYNCHRO START''?), and, as luck would have it, an air conditioner. Now as it happens I formerly subscribed to what Norman calls the ``valve theory'' of thermoregulation, which is that the a.c. did more work the further the room was from the desired temperature, and used to dial up arctic temperatures in an attempt to get to a livable level of heat in a livable time. I have been chastened (pp. 38--39), and can report that the a.c. simply turns its motor off when it hits the programmed temperature --- or, if it is what the industry calls ``intelligent'', a bit before. Almost no one knows how air conditioners work, i.e. our models are inaccurate, consequently we don't use them properly. (Like Norman, I do mean how, not why. We can get away with a black-box understanding of many things, so long as it's the right black box. It would be nice if everyone understood the gas laws and thermodynamics, but that's not involved in using the a.c. correctly.) In the next room there is a computer and a VCR, and there, I think, I need say no more about the importance of understanding how they work, and the utter failure of the designers to get usable models into our heads. Norman shows computers programmers and manual writers no mercy, and, iconoclastically, attacks Emacs, though not by name. I feel this is a bit unfair, and that there are times when it is perfectly appropriate (as well as deeply satisfying) to yell ``RTFM!'' But a dishwasher or glass teat is not, pace certain multimedia flacks, Emacs, and ought not baffle someone with an engineering degree from MIT; or a physics degree from Berkeley; or a sociology degree from Kalamazoo State; or no degree from anywhere.

If I might call you back from the next room for a moment, consider the switches by the door: they toil not, neither do they spin, and Norman would (for the most part) pronounce anathema upon all three of them. They don't lose all points, because one of his principles is that constraints (physical, cultural, and what he calls ``logical'' and I'd call ``practical'') are a sort of ``knowledge in the world,'' which reduce the amount of ``knowledge in the head'' we need to carry around. The switches are pretty severely constrained by practical considerations, what Popper and the Marxists used to call ``the logic of the situation.'' You could, I suppose, bite them or try to use them to sharpen knives, but in sanity you can only flip them up or down. One of them is for the ceiling light; one for the ceiling fan; one of them the lights in the closet. The connections are totally arbitrary, and must be memorized. Norman would plump for a ``natural mapping,'' some arrangement of the controls which makes it obvious which is for what. (His own solution, when you have a lot of lights, is to make a map of the room, and arrange the switches so that each one is in the place corresponding to its light.) The movement of a mouse has a nice natural mapping to the cursor's motion on the screen. One can imagine other, far less natural ones --- lateral motion of the mouse changing the direction of the cursor's motion, and forward motion translating it along the current vector (like a Logo turtle). I'm afraid my ceiling fixtures would meet with more ire from Norman, since they have an additional pair of controls. The fan has four speeds --- two forward, one reverse, and stop, reached in that order by pulling a chain; similarly the four bulbs in the ceiling are in diametric pairs, and a second chain turns on one pair, or another, or both: and the chains are identical.

The last chapter is the one which comes closest to being addressed to designers, though even here he cannot resist his impulse (from lecturing?) to illustrative anecdotes and digressions, in the course of one of which he offers some sage observations on how writing implements change literary style and the difficulty of hypertext. Still, the main thrust of the conclusion is clear, and it is an elaboration on his seven-point program:

1. Use both knowledge in the world and knowledge in the head.
2. Simplify the structure of tasks.
3. Make things visible: bridge the Gulfs of Execution and Evaluation.
4. Get the mappings right.
5. Exploit the power of constraints, both natural and artificial.
6. Design for error.
7. When all else fails, standardize.

[pp. 188--189]