In the past few years I've done a lot of dabbling with 3d computer graphics. I've spent a lot of time tweaking the vertices of geometric shapes in drawing programs the hopes of creating something that looks alive and organic and maybe interesting. In an attempt to get better at modeling creatures, particularly humans, I spend a certain amount of time staring at people in public (I try to be surreptitious about this and not embarrass or frighten any one, honestly). After while, I've started to realize that there are patterns in human facial and bodily features which start to organize in a number of permutations. For instance, a young person's eyes are of a particular shape, and an older person's of another shape. A slim person has hips that curve out so much, and a skinny person's so much less. I find that I start thinking of people in terms of angles and sets of simple parameters.
Reducing people to a handful of numbers is probably my programmer world view creeping in. (Or my world view as a creep, perhaps). How do I reduce a problem to the smallest number of quantifiable variables? If I could create a template for a human which captured all cases, and had only a few parameters, properly defined, by changing a very few variables, I could reproduce all human forms with ease. That's the dream, anyhow.
Something else occurs to me: sexual attraction is based on those very same parameters. In other words, most people are attracted to one body type versus another. What is a body type? A series of angles and ratios. Someone might be be attracted to a 32 degree angle of attachment between two body parts and repulsed by any angle greater than 40 degrees. For example, say you were to show pictures of women's breasts where the slope of the breast as it fell from the collarbone varied over a number of degrees to 10,000 straight men and asked them to evaluate each on a scale from 0-10. Chances are, you would find a distribution of answers where some set of values would be typical, and therefore "ideal". You could do the same, say with a man's proportion of hips to belly. Some ratios will be considered attractive statistically, others not.
Sure, you could call this reductionist. It's certainly not new -- look at Leonardo's drawings of the geometrically perfect human (male). You might say, "Human attraction is far more complex than that. What about personality? What about chemistry?" Fair enough. There are other variables. I'm not saying it's all about one or two angles, I'm just saying that it's about a number of parameters. A bunch of numbers representing angles. In sum, a bunch of geometry.
There's even actual science attached to this: amount of abdominal fat translates to health risk. Absolute waist sizes are being considered by health officials as an accurate measure of risk for serious physical conditions. Don't evolutionary biologists suggest that sexual attraction is linked with an appearance of health?
I conclude: People are sexually attracted to geometry. Some shapes are sexy and some aren't. Some people want to mate with triangles, others with squares.