I have a friend who's an engineering student (let's call him ES) at a prestigious university; and although he's not socially awkward in the nerdish sense, nor is he ill groomed or unattractive; he has had little success with the opposite sex.
...Actually he's had little failure either; simply because his interactions with them almost always leave him thinking "Either she's insane, or she lives in a universe other than the one I live in".
Such interactions leave little room for the development of personal relationships. He's looking for logic and consistency... as those of us with more experience know, there is little to be found within the gender.
Which leads me to this conversation:
[15:10] ES: Someone needs to write a manual about women. You show attention they get pissy. You don't show attention, they get pissy.
[15:11] CB: In order to write a manual there has to be a defined standard, and a consistent interface. Neither exist.
[15:12] ES: It's amazing, isn't it. We spend all this money on science, we can put men on the moon, uncover the genome, but ain't no way in hell one can predict women... Let me rephrase that, not predict, rather have any idea to what level of randomness their actions will go.
[15:13] CB: Assume maximum entropy. It's the safest course.
[15:14] ES: Unfortunately that's an ideal solution. We studied this in "thermodynamics of alloys"; and I think much like ferrite, women are not ideal.
[15:15] CB: Generally speaking one prefers to optimize for the general case rather than edge cases. Unfortunately, with women, the consequences of general case optimization (presuming a maximally entropic system) are catastrophic. Therefore the wisest course of action is to optimize for maximum entropy, and compensate as necessary.
[15:16] ES: You know I had a dream about this last year...