Bring leading computer scientists together with leading astrophysicists, and exciting stuff happens—complex computer simulations of galaxy formation, algorithms churning through terabytes of data collected by telescope arrays.
Same thing goes for biologists, as they work with programmers to bring order to the chaos of neurons firing by the millions.
But get everyone working together under the same roof with extensive time and funding, and unexpected work might take shape. New ideas could form as computer scientists and researchers from a variety of fields hold meetings, chat over lunch, or just run into each other in the hallways.
That’s the kind of research environment the Simons Foundation is trying to cultivate with its latest endeavor, and its most ambitious yet, the Flatiron Institute. What’s more, Jim and Marilyn Simons also decided they wanted that same roof to be their own. More specifically, the foundation made a significant expansion into the building right across the street from its Manhattan offices to accommodate a new research institute fully supported by the foundation.
The new campus, officially launched in November, will eventually employ hundreds of scientists and programmers from varying backgrounds, using big data and computational approaches to crack some of their respective fields’ toughest problems.
With assets of around $2 billion, the Simons Foundation is one of the country’s largest private funders of basic science, supporting a mix of collaborative research initiatives, individual investigator awards, and science education. It’s also supported some research in-house, but the Flatiron Institute is something very new, requiring huge growth for the foundation—they will ultimately lease the entire building across the street. Simons’ current foundation staff totals 160, and Flatiron already has a staff of 60, which they anticipate will grow to 250. Flatiron’s budget is expected to be about $80 million a year.
“Once we were determined to do it, it seemed like the best place to do it was right here,” says Jim Simons, who made his $18 billion fortune running Renaissance Technologies, a data-driven investment firm once called “the commercial version of the Manhattan Project.”
“I have some experience in managing people who are looking at data. Not that I intended to run it, but I felt I could oversee it reasonably,” Simons says in his characteristically breezy manner.
The Flatiron Institute joins the ranks of privately funded research centers like the Broad Institute, the Allen Institute for Brain Science, and HHMI’s Janelia Research Campus. What makes this effort unique, however, is that it’s bringing multiple disciplines into one shop, united by an emphasis on computational methods.
The new effort provides insights into some of the more exciting avenues of modern research, the increasingly collaborative nature of science, and how billionaire funders are deciding how and what they’re going to support.
From Idea to Reality
The seed of the concept behind Flatiron was planted back in 2012, at a two-day retreat at Buttermilk Falls Inn in Milton, New York. Now referred to as the Buttermilk Falls meeting, the event corralled about 20 respected scientists for a discussion of the Simons Foundation’s future programmatic approach.
Out of that retreat came one of the foundation’s core strategies of funding long-term, goal-driven, collaborative research projects, such as the Simons Collaboration on the Origins of Life. But Belgian physicist and mathematician Ingrid Daubechies also floated the idea of a project devoted to data analysis.
“She imagined it as something we would set up somewhere, but I thought, well, I actually made a pretty good living doing data analysis, so maybe we could not only do that, but do it in-house,” Jim Simons says.
Simons is a hedge fund guy, but he’s also a mathematician, and an award-winning one at that. His track record in academia, a stint as an NSA codebreaker, and an overall staggering career that eventually led him to found Renaissance, give Simons one of the more colorful backstories in philanthropy. It also means he loves numbers, and quantitative aspects of science are a thread that runs through much of the foundation’s giving.
So the idea resonated, and it first took shape in late 2013 as the Simons Center for Data Analysis (SCDA), an in-house research project led by renowned mathematician and computer scientist Leslie Greengard. With the success of SCDA’s first couple of years, it eventually morphed into the Flatiron Institute, with the addition of the Center for Computational Astrophysics, led by Princeton astrophysicist David Spergel. Most recently, Flatiron added the Center for Computational Quantum Physics, focusing on materials science and led by Antoine Georges, physicist at Collège de France, Ecole Polytechnique, and the University of Geneva. Greengard, also an MD, continues to run SCDA as a component of Flatiron, now called the Center for Computational Biology.
Buttermilk Falls is a fairly representative example of the Simons Foundation’s decision making in action. The foundation often seeks the counsel of scientists or other grantmakers from outside for advice, sometimes formally, sometimes informally. Its leaders regard collecting input from outside to be an important part of how the foundation works. At the same time, Jim Simons can be quite decisive, as demonstrated during the recruitment for Flatiron's leadership.
David Spergel recalls when he was invited to the foundation to lead a half-day meeting on computational astrophysics, then getting a job offer later that day, much to his surprise: “Afterwards, Jim Simons came up and said, 'This sounds good. Let’s do this,' and offered me the directorship,” says Spergel, who was a department chair at Princeton at the time. He thought it over and took the job.
Leslie Greengard has a similar story, in which he was invited to give a presentation to the foundation and had an offer a few days later to launch SCDA. “I got an email from Jim who asked if I was interested in setting it up, and I said yes,” he recalls.
The foundation doesn’t always move so quickly in its decision making, Communications Director Anastasia Greenebaum points out, noting that long periods of thought and planning often lead up to a project. But on the other hand, “Let's also say the foundation leadership knows a good thing when they see it!”
That brings Flatiron up to three research centers, plus one team solely devoted to computing infrastructure, and plans for one more center, with the discipline yet to be determined. In addition to in-house research goals, foundation leaders hope Flatiron will become a hub for computational science, interacting with nearby universities, and already holding regular meetings and talks that draw researchers from around the world. They even recently got their hands on a supercomputer. And yes, they do have lunch together.
“Having scientists here, it was a great cultural shift,” says Marilyn Simons, recalling what it’s been like building an institute within a foundation. “They were inspiring, and they added an enthusiasm and a vibrancy to our office.”
Productive Interactions
Private wealth founding a research institute isn’t a new thing. In fact, in the U.S., we can look all the way back to the creation of the Smithsonian Institution in 1846. Using high-level computing to analyze big data in research is also not new, although it’s advancing rapidly. It’s mainly the combined multidisciplinary nature and focus on computing that makes Flatiron unique.
In other words, while Broad looks mainly at biomedical and genomic research, and Allen looks at the brain and more recently, cells, Flatiron isn’t defined by discipline. The hallmark is that scientists from different fields have a lot of leeway in terms of time and funding to collaborate with some very good programmers.
Researchers could benefit from that kind of setup in a few ways.
One is that pulling apart the datasets researchers need is becoming increasingly challenging. In biology, for example, one Flatiron project involves analyzing neuronal signals collected via electrodes implanted in the brains of active animals, often rodents. To understand the brain’s circuitry, neuroscientists need to identify which neurons in the brain are firing at any given time. This used to involve just a few electrodes collecting signals for 20 minutes, maybe 200 MB of information. These days, the data might come from a thousand electrodes, collecting millions of signals that add up to terabytes of data, which require complex software to sort through.
“The neuroscientists, they’re supposed to be doing neurosciences. They can’t say, we’re going take three years off and see if we can build a better widget to solve this problem,” says Greengard, who oversees biology work at Flatiron. “And it’s not something that a math community would say, oh, that’s obviously an exciting math problem.”
On the astrophysics side, the problems involve terabytes of imaging and other information collected from increasingly sensitive telescopes, both on the ground and in space. Another focus at Flatiron is the development of computer simulations to better understand cosmological events like galaxy formation, according to David Spergel.
When specialized investigators are not themselves computer scientists, a common solution might be hiring a postdoc if they have the funds, or recruiting a grad student to hack away at a program to solve the problem at hand. But if that work is not the programmer's specialty or top priority, they’ll move on to something else. In the case of Flatiron, teams of programmers and researchers can collaborate over time.
There’s serious value in that kind of collaboration, and it’s not just about providing better tools, says Joshua Bloom, a UC Berkeley astronomy professor who specializes in computational and data-driven science, and is not affiliated with Flatiron or the Simons Foundation. Bringing together methodologists and researchers can overcome blind spots that occur in science.
“People don’t often have the capabilities, the wherewithal to even know how to start,” Bloom says. “And because of that, a lot of our colleagues will wind up basically working in a safer computational domain, which restricts them to certain sets of questions.”
On the computer science side, collaboration introduces experts to problems and data they might not otherwise consider as uses for their expertise. “One of the things that I absolutely love about this deeper connection between domains and these methodologists is that we’re exposing them to something other than ad clicks and Twitter data.”
That cross-pollination between computer scientists and researchers in other fields, but also between the fields themselves, particularly excites Jim Simons. “You get ideas,” he says. “I’m hopeful that the 200, 250 people who are going to be in that building are going to have a lot of productive interactions.”
Researchers see a lot of promise in multidisciplinary approaches to science, and it’s become a big priority in science philanthropy. It was also a major draw for David Spergel back when Simons offered him the job. “It’s an exchange of ideas. I think a lot of advances in science happen when people from different backgrounds talk to each other.”
Flatiron’s Legacy
Bell Labs provides one example that Jim Simons cites as a model for Flatiron, in some ways. It's one of the most productive and revered R&D campuses of the 20th century, known for its climate of creativity and exchange of big ideas.
Of course, while Bell Labs has conducted far-reaching basic and applied research, it’s chiefly an industry project, long the R&D arm of AT&T. Which raises the question: If AT&T drove Bell, what’s driving Flatiron at the end of the day? And how does a philanthropy-driven institute fit into the greater research landscape?
As billionaire donors play increasingly large and influential roles in America’s research agenda, it’s an important question—whether giving is in the service of national research priorities, a company’s profits, or a donor’s personal inclinations, for example.
As for the Simonses’ giving, the couple’s own interests influence their philanthropy on some level, including Jim Simons' interests in data and mathematics. He says that while ideas will often come from in-house scientist staff who oversee giving, or outside advisers, the foundation ultimately wouldn’t pursue something either he or Marilyn didn’t like.
Science philanthropists have a great deal of freedom when it comes to what they fund, which can make it very powerful. A funder like Simons can, if it so desires, pour resources into something like Flatiron, without asking for short-term results, project proposals or grant reports from its staff. But good philanthropy balances the freedom to mount bold projects with a responsibility to the greater scientific community. Flatiron, and the way Simons engages with experts, is a case study of the tightrope private donors walk.
“One thing that we try to tell people who are interested in funding science is that scientists are very willing to share their ideas and advice,” Marilyn Simons says. “They have an interest in advancing science in general.”
Something like Flatiron also has the potential to serve researchers beyond its walls. As UC Berkeley’s Joshua Bloom points out, this level of commitment can serve as a lighthouse for other foundations and federal agencies to identify new and important work to support. As new software is developed, Flatiron will make it available for other researchers to use.
But the execution is key. There’s a lot of risk here, and much money being spent. For all the talk of serendipitous discoveries between fields, who knows if it will happen? This kind of collaboration also requires identifying novel topics that will keep the multiple researchers involved sufficiently engaged.
That being said, the foundation isn’t leaving its other commitments behind, and Marilyn Simons describes Flatiron as one part of a diverse portfolio, like you’d want with investments. Jim Simons doesn't seem too worried about it.
“I think what we’re doing here is really unique in many respects, and we’ll see. I mean, It’s very early days, and I’m 78, almost 79, so I don’t know if I’ll see the full flowering. But I think I will.”