hypothesis: bad characters make for good a.i.
There is a fallacy that plagues much discussion about AI. We assume that the Frankenstein monster will want to take over the earth, want a mate, want revenge... want *something,* at least. But desires, a purpose, do not come automatically with knowledge, with the ability to use or find or create knowledge.. whatever we think "intelligence" is. Intelligence and purpose may both be essential to our humanity; but otherwise, the two are disjoint.

And something like this very fallacy seems to plague many works of literature. Let me explain:

I frequently criticize books for having no sympathetic characters. And it's not just me - witness the famous Eight Deadly Words, "I don't care what happens to these people."

But what does a sympathetic character mean? It is not someone who is sufficiently good, or sufficiently like me. So is sympathizable-ness just a mystery, or can we say anything meaningful about it?

Of the two books currently on my mind - Cat's Cradle, which I just read, and Mother Night, another book by the same author that I actually liked - the main character of Cat's Cradle is surely less evil, more normal, and more sympathetic by any objective standard. Howard Campbell, of Mother Night broadcasts propaganda for the Nazis with the intent of passing on secret info to the Allies but the greater effect of strengthening many Nazi soldiers' commitment. The Narrator of Cat's Cradle merely researches for his book and tries to save the world from ice-nine. But I sympathized with Howard Campbell in his quest for - what? only some form of peace? whether he deserved it or not.

And I did not sympathize with the Narrator, because he did *not* seem to want anything, besides, perhaps, the beautiful Mona. If he cared only about his research, and not about saving the world, that would be one thing; but he did not seem to care about anything at all.

And I suspect this may be the difference - it is not the character that we sympathize with, but the aims, the desire, the purpose. The chains of reasoning - "if-this-then-that" - coupled with the profound desire to have *that* or *not-that,* either way, but it matters which. We are not made to sympathize with robots.

... And here I run off the rails again. How can I tell that a character has no desires, or has them supposedly, but they are only told and not shown? I could try to read lots of books with this question in mind, and collect data for it, but that would take much time - and the data would still be subjective. Perhaps the data-compilation is what an English Ph.D. is for. But there is no way to remove the subjectivity, no way to end up with anything besides a variation on "I know it when I see it." So is there a purpose? - which seems to be a relevant question in this essay, after all. Okay then, here is the difference - in when I see it, what I know.