Long before the public learned that the Large Hadron Collider had unearthed the Higgs boson, physicists like Matt Hollingsworth knew it was coming. They had seen hints in the data: First, a small, statistically dubious bump where the subatomic particle---the one that explains why everything in the universe has mass---should be. Then, the bump began to grow.
Every day, some employees would religiously check an internal website that charted the growth of the signal, the probability that it was real. Before breaking out the champagne, they needed to reach three-sigma---or 99.7 percent certainty. “Once we finally crossed over, there was a huge celebration,” says Hollingsworth. “It was about a year before the papers were published.” The day they finally came out, and the God particle---lo---appeared to the rest of Earth, Hollingsworth didn’t even know it until a friend told him.
His emotions---and those of many other physicists---were mixed. Particle physics currently centers on something called the Standard Model, a fundamental framework developed in the 1970s that describes particles and their interactions. In this model, the Higgs explains why elementary particles like quarks have mass, and why photons don’t. But the Standard Model doesn’t explain things like gravity and the nature of dark matter. Physicists hoped their particle accelerators would reveal new physics, leading them toward a more complete view of the universe. They wanted to be at least a little wrong. And what the LHC had shown them was exactly what they’d predicted, and nothing else.
“Most people in the field would consider that the worst outcome possible,” says Hollingsworth. Especially for him. Because after that least exciting, worst, really cool, long-awaited discovery, Hollingsworth found that there wasn’t much more he could do in fundamental physics.
While scientists were finding exactly what they were looking for, high-energy particle physics funding in the US was faltering. In 2013, the National Science Foundation cut particle physics money by around 10 percent, leading some to predict a drop in the number of postdoctoral positions that Hollingsworth was looking for.
He started looking for jobs about six months before finishing his PhD. “We were still in the recession,” he says. “Postdoc positions more or less just disappeared. The high-energy research side of the Stanford Linear Accelerator shut down. They turned off the Tevatron at Fermilab. All of the things I could have done were just evaporating.”
Compounding such shortfalls was the fact that so many young students had been recruited to work on the Large Hadron Collider’s Big Science. In 2012, CERN spat out 327 masters and PhD theses---more than twice what it had before the LHC started---but the year after, the field only had 124 postdoc positions for those students, according to a 2013 Science article.
Two tracks diverged in a wood. Hollingsworth could take the one that would mean cutthroat competition, constant uncertainty, and existential distress about The State of the Field. Or he could go write software for a startup. About horses.
It’s not like it was an easy choice. Hollingsworth had almost always wanted to be a physicist. In third grade, when everyone else dressed up like Power Rangers for a school costume day, he showed up as Albert Einstein. In high school, he set his sights specifically on CERN, the parent organization of the Large Hadron Collider. Not an acronym in the typical locker set lexicon.
In college, at the University of Tennessee-Knoxville, he needed to earn extra money. “But I didn’t want to work at Taco Bell, so I went to the assistant department head and asked if I could do something, anything.” He became a research assistant in the high-energy physics group, and the very next summer that work led him over to CERN. He worked on-site every college summer after that, and once for a nine-month stint. He continued on this track in graduate school at the same university.
With CERN, Hollingsworth worked to create the program that shuts things down if radiation levels get too high. Then on an instrument called the Pixel Luminosity Telescope, which measures the full extent of the collisions going on inside the accelerator. After that, he jumped to the CMS Pixel Group, where he wrote software to reconstruct the paths particles had taken, and then to analyze work on the decay of a specific particle. He was living up to his Einstein costume.
At least up to the point of his group’s greatest success. That’s when he started thinking different.
One day—after the bump, but before the bump was heard ‘round the world—Hollingsworth was having coffee with his CERN colleague Dave Stickland, whose wife was a dressage coach. “He explained one of the main problems in dressage,” says Hollingsworth. At equestrian events, seven judges evaluate riders on 20 different criteria, giving 140 scores per performance, with three performances per event. That’s a lot of numbers, and for the most part, competitors were only looking at their final score—not so much the individual ones that averaged into it. “There was this large amount of data no one was using to make decisions, even though it could really help them,” he says.
Stickland showed him a report he’d drawn up for an Olympian, and asked Hollingsworth if he could transform that into a web-based platform that other people could use to do performance analytics. He could. The two founded a company Global Dressage Analytics, turning those score sheets into predictive models that guided horse training. More lucratively, they modeled how consistently judges were judging, by comparing their evaluations of a given rider.
The transition sounds almost comically abrupt. But it makes a certain amount of sense. Coding, Hollingsworth says—along with lots of startup work—is so similar to high-energy physics that the two are almost indistinguishable. It’s just coding and data analysis.
And Hollingsworth is not alone in his pivot. He notes that his friend group could be biased, but “there are zero people I know who stayed in high-energy physics,” he says. “One of them hopped fields in physics. The rest are doing software engineering.”
Once he and Stickland saturated the fancy-horse market, Hollingsworth moved on. He did artificial intelligence research for a government contractor. Half writing grants, half coding and machine learning stuff that [redacted].
Still, though, the business world lacked a certain je ne sais quoi. Hollingsworth struggles to articulate the difference, then tries: “I haven’t been part of any group that had as many smart and devoted people who were excited to be doing their work as at CERN,” he says. “In Silicon Valley, people are really smart, but it feels different.”
The programming is about the same as at CERN, he says; the long hours are about the same. But, “over there, people loved what they were doing. Here, they want to win something.”
Hollingsworth has now perhaps found a way to mix the two worlds---the monied one of startups and the nobly motivated but perpetually underfunded one of science. He’s enrolled in Stanford University’s MBA program, with the goal of taking university-level technologies---next-generation stuff that often stays in academic silos---into the commercial space. At CERN, he had worked on a detector, a decade technologically ahead of its time, that would be useful for PET scanners. But scientists hadn’t taken to the wider world. “Nobody thinks about it because you don’t get gold stars for it,” he says. “You get gold stars for publishing, and that’s it.” That often keeps scientists playing it safe, sticking to topics that will reliably produce splashy results.
He imagines visiting labs, talking to researchers, convincing them their custom-built hardware is commercially viable. Maybe in the future, universities could take this kind of technology transfer more into account, rather than focusing so much on the publish-or-perish model of academia.
In some sense, his frustration with risk-averse academic science is the same as particle physicists’ frustrations with the LHC results: So far, they have spent their time and money finding out what they already kind of know. And how they will find out what they don’t know---that’s a path they’ll have to chart.