In Praise of Ambiguity

We game designers, raised in the culture of programming, tend to be an explicit lot. Good programming is mostly a matter of telling the computer exactly what you want it to do; most neophyte programmers’ errors arise from ambiguity. Thus, we despise ambiguity, we shun it, we try in every way possible to stamp it out.

This perfectly appropriate response to ambiguity in programming tends to splash over into other areas of our work, most notably our designs. The most extreme expression of our disdain for ambiguity lies in our endless search for ever more explicit cosmetics. Can we make our graphics more detailed, more colorful? Can we get more pixels onto the screen? How many channels of sound can we squeeze in? The more explicit detail we can get into our cosmetics, the better our product will be right?

In this essay I want to sing the praises of ambiguity. I believe ambiguity to be essential to effective artistic expression.

I shall begin with most titillating example I can think of: pornography. I shall concentrate my attentions on graphic pornography, that is, photographs of females. My impression is that there are three broad classes of such pornography. First, there’s the swimsuit pinup, a photo of a woman in a revealing but still legal bathing suit. We often see such pinups pinned up on the walls of male workplaces (but the times, they are a-changing). Next come the R-rated photos; these are characterized by bare breasts and possibly some visible pubic hair. Lastly, there are the hardcore shots, mostly closeups of genitalia.

Note that the sequence I have described constitutes a sequence of explicit display of female genitalia. The most legal, proper stuff, stuff that you wouldn’t be ashamed to put on the wall of your mechanic’s garage, shows scantily clad women. Their genitalia and bodily form are partly revealed, but critical components of the visual experience are withheld: nipples, pubic hair, and so forth. The next class of pornography reveals more, but not all female genitalia. This R-rated class is less publicly acceptable than the first class; the proper gentleperson of the 90s would confine its display to his Harley mechanics’ shop. The third class, hardcore pornography, is simply not permissible for mature individuals at least, not openly.

Now, male sexual fantasies tend towards the explicit, so one would expect that the popularity of pornography would be proportional to its explicitness. Yet this does not seem to be the case. At this point, I must confess to a virginal lack of data on my part: I have not carried out extensive statistical studies of the relative popularity of various kinds of pornography. Moreover, the data is not readily available because most males are understandably reluctant to reveal the extent of their pornographic consumption. Me, I don’t even own a pornograph.

Nonetheless, I do have some impressions or hunches. It seems to me that the hardcore stuff is not the most popular; I’d guess that the R-rated stuff is more popular. I don’t really know if swimsuit pix are more popular than bare-breast pix, but I strongly suspect that both are more popular among men than straight X-rated shots.

There’s another factor: the stuff I’ve seen seems to thrive on ambiguity. If explicitness were the sole determinant of quality, then all pornographic photos would present women standing facing the camera with legs spread for easy inspection. But in fact, I don’t think I’ve ever seen anything like that. Almost all pornographic photos involve some deliberate use of ambiguity to enhance the experience. This girl coyly glances over her shoulder at the camera, a bit of nipple just visible. That girl casually drapes her arm just over her breasts, so that you can see most but not all. Another one swims toward the camera, her breasts just below the water so that they’re visible but not perfectly clear.

Why the ambiguity? There’s no legal basis for it; all the legal definitions of pornography that I have seen make no distinction between exposed breasts and exposed vulvae. In the eyes of the law, showing half a nipple is the same as showing both. So why do males seem to prefer half a loaf?

I suggest that the crucial factor at work is the role that ambiguity plays on the imagination. The entertainment value of pornography does not lie in its merit as an anatomy lesson; most men have figured out the basics of female anatomy. Certain aspects of the female image seem to trigger deep-seated associations in the male brain. Think of it as a kind of pattern-matching process. Somehow the various curves he sees link up to push a button labelled "SEXY!" and the experience is pleasurable. The key idea here is that the pattern-matching is not based on any explicit pattern. It’s not as if you have a detailed mental image of the Sexiest Female in the World, and you compare the image you see with the mental image, pixel by pixel, and if you get a match, it’s pleasurable. It’s much messier than that. There are a variety of vague visual concepts that are impossible to articulate. Take "hippiness", for example. The shape of a woman’s hips can be more or less visually stimulating. But what, precisely, makes for this attractiveness? Curvature? Width? Boniness versus fattiness? I don’t know but I do know that there are definite factors at work here. I can look at one female’s hips and say "Ugh!" and another female’s hips and say, "Yowzah! Woof! Bark!" although, being a gentleman and a scholar, I say it in Latin ("Josa! Oof! Barcus!") or in Greek ("ouwa! ouf! barkos!"). Moreover, the factors at work are not arbitrary; I suppose that if you showed me 50 photos of female hips, I would be able to sort them by attractiveness, and you would probably be able to detect some regularities in my sorting process.

So what we have going on here is a kind of pattern matching, but one that is based on hard-to-define attributes.There is also some variance in male tastes, but surprisingly little; while men might argue over whether Cindy Crawford looks better than Kim Basinger, I doubt that anybody would champion Phyllis Diller against either of the first two.

Now at last I can drive my point home: pattern matching against such murky specifications is enhanced by ambiguity. If you had an unambiguous specification of sexiness, then all pornographic images but one would fail to yield a sexiness match, and that one image would do for all occasions. But because your concept of sexiness is ambiguous, the best fit to it is achieved with an ambiguous source image. This is the central point: because basic human drives are murky and ambiguous, the best way to satisfy those drives is with ambiguous stimuli.

Of course, women have known this truth for eons. Why do you think they wear clothing that suggests without revealing? Why do you think that they engage in deliberately ambiguous sexual behavior? Men resentfully call this "teasing", but the hard fact is, it works. Why else would women use the technique? And what about the fact that most copulation takes place in the dark? If men were really so dominated by visual considerations, then wouldn’t they always prefer to make love in the bright sunlight?

And another thing... no, I’d better not -- I’m pushing the limits of good taste.

I suspect that, were I to base my case solely on pornography, some might object that my data is skewed. So let’s consider some other areas. How about literature? After a century of moviemaking, books have not been pushed aside as a form of entertainment. Movies have certainly carved out an impressive niche for themselves, but books just keep chugging along. Indeed, the overall trend of fiction purchases for the last few decades has been steadily upward. The fundamental difference between books and movies is not so much graphics versus text as explicitness versus ambiguity. A book leaves much to the imagination; a movie leaves less to the imagination. The ambiguity attending the written word can be exploited by the author to powerful effect. It just takes a certain amount of skill.

Even within the movies, we often see deliberate use of ambiguity to heighten the intensity of the experience. How many times does the director deliberately withhold visual information from the viewer? In the ever-so-famous shower scene in Hitchcock’s Psycho, the murderer is never clearly seen, and although the victim is stabbed to death while naked, we never get to see any breasts or other good stuff. A great deal of visual information is withheld from us, leaving ambiguous gaps that we must fill in with our imaginations. You want to see the same thing in the modern, action-packed directors? OK, here’s an example from George Lucas’ Star Wars: Darth Vader is preparing to torture Princess Leia. "And now we shall discuss the location of the secret rebel base", he announces, as a device loaded with needles and probes approaches her. The camera jerks away from the scene, leaving us to imagine what happens next. Or how about the monster in the trash masher? We never see more than a single tentacle doesn’t that make it all the creepier? You want Spielberg? OK, here’s an example from E.T.: the guy masterminding the pursuit of E.T. is never shown directly. All you see are his keys jangling on his belt. Making him uncertain, ambiguous heightens his sense of threat. Or how about Jurassic Park: for an agonizingly long time in the movie, the T-Rex is sensed indirectly. How about that scary moment when the injured mathematician sees the concussion circles in the puddle of water and realizes that the T-Rex is approaching? If one of our typical computer game designers had directed Jurassic Park, the movie would have started with the scene of the lawyer being eaten. Who needs artistry when you’ve got computer graphics?

I don’t want to come down too hard on the engineering types; there is a reason for their value system. In any form of engineering involving communications, the principle obstacle is always bandwidth. The engineer is always asked to squeeze more bits per second out of the system. In the case of a visual communication system, that means more pixels per second. Note, however, that once you reach a certain amount of bandwidth, the pressure for more bandwidth diminishes. We haven’t seen many improvements in film resolution since Technicolor in the 1950s. Nor is there much demand for improvements in audio fidelity since the compact disk (except, of course, for the audiophiles, who will never be satisfied).

Thus, the involuntary ambiguity that arises from poor technology is undesirable; we seek to eliminate such ambiguity. But deliberate use of ambiguity is another matter. If you think of it in engineering terms, then the designer is trying to push a button inside the user’s mind that will be triggered by some long string of binary digits. The matter is complicated by the fact that, although the more significant bits in the string are common to all people, the least significant bits in the string are different for different people. Thus, your job is to hit all the high-orders bits and cover up the low-order bits in smoke. Knowing which bits are high-order and which bits are low-order, that is the stuff of art.