A Quantum Diaries Survivor

I have long been of the opinion that writing about science for the public requires the writer to simplify things down to a level which is sometimes dangerously close to mislead the uninformed readers. I think is a small price to pay if you want to keep open the channel of communication with the general public, but it is indeed a narrow path the one you sometimes find yourself walking on, and fallacy is always a possible outcome.

As the well-informed readers will realize, I am hat-tipping Hank Campbell and the catchy title of his best-selling book "Science Left Behind" with the title of this post, for lack of more imagination. What I want to discuss is, however, something only partly in line with the interesting topics of Hank's book. It is something that I see happening around these days, and which I ache for: the dumbing down of our decision making in science.

Given the use that people do of Google searches nowadays, and the rather special nature of my usual readership, I feel I may need to first of all apologize for the deceiving title of this post to the 80 to 90% of the visitors, who came to this page by searching for ways to become a member of a selection committee of miss Universe.  Sorry, you had it wrong - we are going to discuss parametric models here, not top models. But if you are happy to hear about the issues of fitting data with different functional forms, you are welcome to read on.

[Update: I found the time to add a few links to the post below, which I had previously omitted for lack of time (hey I'm on vacation!), and I also updated it to add some commentary of Sabine Hossenfelder's latest post on "the end of particle physics".]In this age of short-term reward strategies (in politics, in society, and in individual behaviour) planning huge endeavours 20 years ahead is harder than it used to be. In the late eighties, when the Large Hadron Collider (LHC) was conceived and argued to be doable by a few visionaries, it immediately looked like a great idea to all. 

Supersymmetry (SUSY) is a possible extension of the Standard Model (SM), the currently accepted theory of subnuclear physics. SUSY has the potential to "explain away" some of the  problematic features of the SM, by introducing a new symmetry between fermions (the stuff that matter is made of) and bosons (the vectors of the forces that hold matter together). Introduced in the seventies, SUSY was tested with increasingly stringent tests in higher- and higher-energy collisions at particle accelerators, but all searches for its particles have returned empty-handed. In particular, many physicists thought that the turn-on of the Large Hadron Collider (LHC) eight years ago would result in heaps of new discoveries of SUSY particles, which unfortunately weren't. 

I don't remember who said it, but there's a quote I like a lot: "If you torture them long enough, the data will confess to anything". What the author meant is of course that the manipulation of experimental data and the a posteriori use of hand-picked methods, approximations, and other ad-hoc choices allows you to demonstrate anything with them, from one hypothesis to the opposite one. Statistics, in other words, is a subtle science, which must be handled with care. It is a powerful tool in the hands of people with an agenda. 

It's been a while since I last discussed something personal in this column. The reason is not that I changed my mind with respect to being open and freely share my ideas, experiences, and personal life things here - I have long argued that if a blog is not personal, it is not interesting, and I stand by that assessment. Rather, the reason of my not talking much about myself and my personal / work life is the good old one: lack of time. If I have time to write an article, I try to do it on a subject which I suppose will be more interesting to the readers of this site. Hence physics, rather than life and work, takes the precedence. But it needs not be so all the time, so today I will try to go in the other direction.

My CMS colleague Didar Dobur, who chairs the "Top Properties" working group in the experiment, presented today the first observation of the process whereby a top quark is produced in association to a Z boson. I could follow the presentation by videoconference, so I am blogging about this result in close to real time, for a change.

What machine will replace the Large Hadron Collider to further our knowledge of fundamental physics at the high-energy frontier, in the forthcoming decades? The question is not at all a far-fetched one: these large machines require two decades to be built - and can then typically be operated for two further decades, amassing collision data that slowly but steadily improve the precision of our estimates of the parameters of nature.

As I mentioned in yesterday's post, there is a workshop going on this week at Fermilab, where 110 attendees - mostly particle physicists, but some computer scientists are also present - discuss how to push for more effective use of machine learning tools in the extraction of information on particle collisions. Also one goal is to understand what new ideas from the world of machine learning could find ideal applications in the typical use cases of research in fundamental physics. Here I wish to mention a few interesting things that I heard at the workshop so far, in random order. I will rarely make direct reference to the talks, to encourage you to dig into the pdf files available here.

I flew to the US yesterday to get to Fermilab, where I am following a workshop titled Machine learning for jet physics". My goal of this post is to explain what this is about in general terms, such that if I have enough stamina I will give, in follow-ups to it, a few examples of the status of this interesting research activity, which encompasses particle physics and computer science and can provide spin-offs in a number of related areas of fundamental research.

What is Dark Matter (DM) and why should you care? I feel I should start this article by explaining these two things first, as we live in an age when nobody has time for long historical or context-setting introductions.