A lesson from the debunking of Oliver Sacks (and other nonreplicable findings): "Trust literatures, not individual research studies," says Bryan Caplan, by way of Tyler Cowen. I've long advocated this not just for readers but science journalists: Stop getting your ideas for…