Two of the most widely reported stories of the year in particle physics,
both depend crucially on our understanding of the fine details of the proton, as established to high precision by the NNPDF collaboration itself. This large group of first-rate scientists starts with lots of data, collected over many years and in many experiments, which can give insight into the proton’s contents. Then, with a careful statistical analysis, they try to extract from the data a precision picture of the proton’s internal makeup (encoded in what is known as “Parton Distribution Functions” — that’s the PDF in NNPDF).
NNPDF are by no means the first group to do this; it’s been a scientific task for decades, and without it, data from proton colliders like the Large Hadron Collider couldn’t be interpreted. Crucially, the NNPDF group argues they have the best and most modern methods for the job — NN stands for “neural network”, so it has to be good, right? 😉 — and that they carry it out at higher precision than anyone has ever done before.
But what if they’re wrong? Or at least, what if the uncertainties on their picture of the proton are larger than they say? If the uncertainties were double what NNPDF believes they are, then the claim of excess charm quark/anti-quark pairs in the proton — just barely above detection at 3 standard deviations — would be nullified, at least for now. And even the claim of the W boson mass being different from the theoretical prediction, which was argued to be a 7 standard deviation detection, far above “discovery” level, is in some question. In that mass measurement, the largest single source of systematic uncertainty is from the parton distribution functions. A mere doubling of this uncertainty would reduce the discrepancy to 5 standard deviations, still quite large. But given the thorny difficulty of the W mass measurement, any backing off from the result would certainly make people more nervous about it… and they are already nervous as it stands. (Some related discussion of these worries appeared in print here, with an additional concern here.)
In short, a great deal, both current and future, rides on whether the NNPDF group’s uncertainties are as small as they think they are. How confident can we be?