My last post explained how I came up with my atheist persona for the ITT. Now I want to talk about judging heuristics and how I tried to game some of them.
I think it pays off to plant cultural cues. For example, last year Chris did fairly well in part by referencing SMBC comics in his entry and, according to one comment, I might have profited from mentioning Terry Pratchet. As a judge I didn’t put much trust in that heuristic, because it is so easy to fake. But I think it is fairly interesting because it’s one of very few positive kinds of evidence a shammer can send.
A large part of winning is not loosing. What I mean by this is that judges are mostly relying on negative evidence. They look for evidence of fakery and if they don’t find it they pass the entry. That’s a reasonable judging strategy too, because there aren’t many reliable positive signals.
As a judging tool, consistency is overrated. Of course if people say obviously contradictory things they are likely playing dumb. But that didn’t really happen in this year’s test. To actually catch people the criterion must be widened to more subtle inconsistencies. For example, evangelicals and Catholics are quite different from each other and mixing their traits is a sign of being neither. But then we are no longer talking about strict inconsistency. There actually are real evangelical Catholics who do mix those traits. It’s just that I wouldn’t expect many of them to participate in the ITT. So consistency becomes more of of probabilistic thing than the binary question one might expect. And it fails if it’s pushed too far. For example, I thought Sweet Tea’s Christian entry was written by an atheist because I thought his personality didn’t fit with his way of explaining natural law. That was too clever by half.
On the side of the faker I think this is best taken care of by trying to create a fake persona one can respect.
It’s worthwhile to pay attention to what the contestants care about. So it can be revealing if contestants give the right answers to the wrong questions. In retrospect this is where my own atheist entry sucked. I lead in with a thinly veiled reference to the atheist canard of belief-in-belief. And then I end my answer to the second question with “Moral progress actually happened by stepping past unquestionable authorities.” It’s not that many Internet atheists would disagree with that; these two bullshit ideas are actually quite popular with them. But they are also quite extraneous to the main line of my atheist’s argument. If he was real, the atheist I portrayed would have had more important ideas to communicate on a budget of 600 words. But they are things that annoy me, so they are important features of my image of atheism. Effectively that made me sound more bigoted than the real atheists. This didn’t harm me on the main results (Though the reckoning may yet come on the so-far unrevealed attractiveness scores), but if people had thought of that heuristic it would have been fairly easy to figure me out.
The Christian side actually did use this heuristic to do in atheist pretenders who made two much ado about homosexuality. This actually is a theme rife with hard questions and it might be an interesting question for a future iteration of the ITT. But honestly the question was about things where the Church corrects us, and most straight conservative Christians really aren’t all that affected by homosexual actions being a no-no.
Finally, there is another heuristic that is more problematic: We have expectations about who participates in the ITT. Jacob speculates that a genuine fundamentalist would have had problems passing the Christian round and I think he has a point there. On the other side of the divide, last year an Objectivist participated and his atheist entry didn’t find much favor with his coirreligionists. So we flunk not only entries missing the type they try to portray but also types we don’t expect to see. This doesn’t quite fit with the idea of judging the quality of imitation, but on the other hand it would be ridiculous to ignore information we actually have.
Defending against this reflects back to the strategy level. Basically the character must be constructed to match a type expected by the audience. For example, I didn’t dare to present a pro-life atheist. They do actually exist, but I don’t think judges recruited through the atheist blogosphere would have stood for it.
Pingback: Gilbert’s Turing Strategy and Judging Heuristic
Yes, a genuine fundamentalist would have trouble proving himself a Christian. But not just in the Turing Test.