Do we need N3LO parton distributions?

We discuss the uncertainty on processes computed using next-to-next-to leading order (NNLO) parton distributions (PDFs) due to the neglect of higher order perturbative corrections in the PDF determination, in the specific case of Higgs production in gluon fusion. By studying the behavior of the perturbative series for this process, we show that this uncertainty is negligible in comparison to the theoretical uncertainty on the matrix element. We then take this as a case study for the use of the Cacciari–Houdeau method for the estimate of theoretical uncertainties, and show that the method provides an effective way of treating theoretical uncertainties on the matrix element and the PDF on the same footing. We briefly discuss the possible generalization of these results to other processes, and in particular top production.