When things go wrong: correcting the scientific record