 |
 |
Stuart, I don't think there's anything wrong with what you're saying, except that your scientific logic (as a naturalist and observer of ravens) relies on implicit assumptions about the likelihood of observing so many ravens without seeing a white one if white ravens do in fact exist. If you're willing to estimate, given your experience as a naturalist, scientist, etc., the likelihood of failing to see a single white raven in your sample, then of course it's completely reasonable to say that seeing all those black ravens is informative, and that each additional black raven you see is additionally a tiny bit informative. Whether or not there's a reasonable way of formalizing this intuition probably varies, and depends on things like whether or not your 1000 animals were sampled appropriately randomly. But clearly if your intuition were different -- i.e., that observing 1000 consecutive black ravens in a universe heavily populated with white ravens is quite likely -- it would never occur to you to draw the same inference. Of course, I don't know how well in general logicians are impressed with inferential statistics. Even if you collect data that would be only 0.00000001% likely to occur if white ravens exist, they would still be correct in saying that you haven't proven anything, even if you've checked all but one raven. Science isn't much about proof in that sense, and I think both scientists and philosophers can be comfortable with that. My guess is that the reason these white/black raven problems are counterintuitive is that for better or worse, people do form intuitions about the likelihood of seeing 1000 black ravens without a single white, even when you give them a completely artificial scenario and don't tell them anything about how the data were collected. It's not shocking that scientists would have weird intuitions about scenarios involving single exemplars in an infinite universe. In reality if you run a raven experiment, you design it such that you're very likely to observe a white one if white ones do exist. But logicians never set it up that way. Anyway, if you really wanted to generate an annoying debate, the trick is to post right *after* a conference. dan