The following is a guest post by Bob Mesibov. Nico Franz and Beckett Sterner created a stir last year with a preprint in bioRxiv about expert validation (or the lack of it) in the "backbone" classifications used by aggregators.