Consider this an addendum to my recent post on an influential study of Alzheimer’s that looks to have featured manipulated data. Retraction Watch has been in business for quite some time now, and is likely to get busier because of the extra opportunities for dishonesty available through machine learning. This situation will continue to get worse until science — and academia more generally — begins to get serious about correcting its perverse incentives. Every scientist knows that certain kinds of results get (a) attention and (b) citations, resulting in (c) prestige for the researchers’ institutions and (d) promotions and raises and maybe better jobs elsewhere for the researchers. 

Again, this is a problem for all of academia: as I have written elsewhere, “the academic enterprise is not a Weberian ‘iron cage,’ it’s a cage made from a bundle of thin sticks of perverse incentives held together with a putty of bullshit.” But when the bullshit takes over the sciences, especially the health sciences, people die. The incentive structure has to change.