I, too, used to dismiss witness testimony as a useless form of data. I've formed a different opinion over time, and I think there are statistically significant anomalies when you can gather a large number of witnesses, even if there is no proposed mechanical explanation in our current paltry understanding of science. This case is pretty unreproducible so I doubt there will be reliable sample sizes in the near future, but nonetheless it's interesting.
And why do people jump to the least likely explanation for things?
reply