Nathan: A new theory that matches old data is likely to be more complex than an old theory that matches old data because the old theory is probably the first that someone thought of and bothered publishing. This justifies some conservatism, but not so much. The effort required to revise beliefs or to hold them as a Bayes net is a further justification, but still justifies much less conservatism than is seen.
OTOH, I disagree with your dark matter example. The case for dark matter over weird physics actually seems fairly compelling to me, though this probably isn’t the place to discuss it.
Nathan: A new theory that matches old data is likely to be more complex than an old theory that matches old data because the old theory is probably the first that someone thought of and bothered publishing. This justifies some conservatism, but not so much. The effort required to revise beliefs or to hold them as a Bayes net is a further justification, but still justifies much less conservatism than is seen.
OTOH, I disagree with your dark matter example. The case for dark matter over weird physics actually seems fairly compelling to me, though this probably isn’t the place to discuss it.