"Vampirical hypotheses"

[mmcirvin]

On Language Log, Mark Liberman assembles a fine cri de coeur about terrible reporting of probably-bogus (or just plain inconclusive) scientific results, built in part on Ben Goldacre's post about a failed replication of the Bem precognition paper, that apparently wasn't published because it was just a failed replication.

Maybe Cosma's neutral model of scientific inquiry needs to be extended to a neutral model of scientific journalism.