Maybe they should use double blind revision to select their talks..