Hype vs. Reality at the MIT Media Lab

It was a validating moment for Caleb Harper. Norman Foster, the renowned architect who designed Apple’s circular headquarters, had come to the Massachusetts Institute of Technology’s research facility in a Boston suburb last June for a tour. He was there to learn about Harper’s high-tech agriculture project and to “explore future links” between the Media Lab, where Harper works, and the architect’s foundation. Harper himself had been trained as an architect, but as he admitted later on Instagram, he almost didn’t graduate from architecture school. It was momentous, he wrote, “to have an absolute architectural icon tell me the exact same thing happened to him and not to worry, as ‘genius is often misunderstood.’”

During the tour, Foster was shown a series of “food computers,” whose futuristic feel and world-changing promise were turning Harper into a star. The devices are tricked-out containers with multicolored LEDs that grow plants hydroponically, meaning without soil. The machines can control their own climates, Harper has said, and collect data on how the plants are faring. Food computers can be large enough to hold trees and small enough to sit on a desk. They are presented as the cutting edge of modern agriculture, distilled into a box.

The food computer promises a revolution on two fronts. First, it could turn farming on its head by creating “recipes” for plants’ optimal growth conditions. “What if you could take this apple, digitize it somehow, send it through particles in the air and reconstitute it on the other side?” Harper asked in a 2015 TED Talk.

Second, food could be grown anywhere, no matter how hot the weather gets. People could grow food in cities or in deserts, reducing the likelihood of famine and the need to ship it around the world. The precise climate in Bordeaux could be recreated in a box in the Sahara — or on Mars.

That kind of potential is seductive to journalists, funders, and anyone with a concern about the warming planet. Harper’s TED Talk has been viewed more than 1.8 million times. Food computers were featured last year on 60 Minutes. The Wall Street Journal published an article in October describing Harper’s vision of the future of agriculture. The BBC recently included the devices in a segment on food technology, calling Harper “a powerful voice in the exploration of our future food systems.” His book, The Future of Food: How Digital Technology Is Changing the Way We Feed the World, will be published in January by Atria Books, an imprint of Simon & Schuster. Food computers have already been sent to schools in the Boston area to be used in teaching kids about agriculture, and to a refugee camp in Amman, Jordan.

On his tour, Foster was shown food computers filled with plants. But what he probably didn’t suspect was that the specimens hadn’t been grown in the machines. They had been ordered from another hydroponics system, according to a person with knowledge of the visit. They had been placed in the food computers, the person said, to make it look as if they’d been grown there all along.

In recent months, The Chronicle has spoken with five people connected with the lab who say the claims Harper has made in talks, to journalists, and to visitors of the Media Lab are exaggerated. They say he has led the public to believe that his grand idea for what the project might one day achieve has already come to fruition. Some would not speak on the record, fearing Harper’s ire and possible damage to their professional lives. But others, having lost patience with what they see as his distortions, have started talking to reporters, including one at Business Insider, which reported on some of the accusations on Saturday.

Harper did not respond to questions about specific allegations regarding the Open Agriculture Initiative’s work. But in a Chronicle interview, Harper said that his big-picture vision is much needed in the field of plant science, where progress can be slow and research is often narrowly focused. It’s a posture that is welcome at the Media Lab, an unconventional environment in which researchers are meant to break through traditional academic strictures as they innovate. The Media Lab even gives a $250,000 prize to people who “engage in responsible, ethical disobedience.” But some of the researchers who have worked for Harper say that often, rather than engaging in genuine scientific inquiry, he’s selling a fantasy.

One former researcher described buying lavender plants from a gardening store, dusting the dirt off the roots so it looked as if they’d been grown without soil, and placing them in the food computer ahead of a photo shoot. The resulting photos were sent to news media and put on the project’s website. The same employee said the food computers that had been given to schools either didn’t work or needed extensive support from Media Lab technicians. Former employees also said that when Harper has given presentations on his work at the Media Lab, he has described research projects that either they didn’t know about or believed to be exaggerated.

In recent weeks, the Media Lab has found itself enveloped in controversy after revelations that it had received significant donations from Jeffrey Epstein, a wealthy investor and convicted sex offender who committed suicide in jail in August. The Media Lab’s director, Joichi Ito, resigned on Saturday after The New Yorker revealed emails showing that he had tried to hide the full extent of the lab’s relationship with the disgraced financier.

Amid the clamor, less attention has been paid to research the Media Lab supports and the environment it fosters. For years the lab has drawn fire from critics who say it has privileged what The Baffler once called “techno-futurist propaganda” over rigorous scholarship. In the case of food computers, people connected to the project insist, image won out over substance.

There was another person touring the facility with Norman Foster and his wife, Elena Ochoa Foster: Nicholas Negroponte, an architect who has been a mentor to Harper. Negroponte founded the Media Lab in 1985 and was its director for two decades, intending it as a place that would bring together the burgeoning world of computing with the publishing and broadcast industries. It can claim successful inventions including the electronic “ink” used in Kindles and the video game “Guitar Hero.” Researchers there are now working on prosthetic limbs and on computers that read people’s minds. The lab has bolstered MIT’s reputation for technological ingenuity and entrepreneurial ambition.

The Media Lab’s funding structure is unusual. The lab’s operating budget is $75 million a year. “Member” companies, which tend to be megacorporations like Google and Nike, pay an annual fee of at least $250,000 in exchange for access to the technology that’s developed there. Representatives from the companies visit the lab twice a year to learn about the researchers’ progress.

Research groups at the Media Lab receive their money from a central pool, but there are also “initiatives,” like the Open Agriculture Initiative, Harper’s project. They get some funding to start out, but after two years they are expected to be self-sufficient. They raise money on their own, often from the member companies, but also from foundations and other sources. In an internal email in 2018, Ito specified that when an initiative is run by a research scientist it will be funded with “soft money,” meaning the position is linked to the availability of funds. Faculty members, he said, would review initiatives annually; those “that no longer align with the direction and culture of the Lab will be asked to spin out or wind down.”

The drive to self-sufficiency creates freedom for entrepreneurial project leaders. But it also breeds fierce competition for funds. Presentations to potential funders need to be captivating and impressive, and it helps if the work will clearly benefit the companies that give to the Media Lab.

The Media Lab’s lack of disciplinary boundaries also sets it apart. It calls itself “antidisciplinary,” broadcasting a disdain for the traditional walls around academic departments. The idea is to get computer scientists, designers, artists, and engineers working creatively together.

While Ito was the director, he told employees that he wouldn’t tell them what to do. In an internal email with the subject line “What is the Media Lab and what do I do here?” he described himself as more like the steward of a rainforest than a boss. He’d resolve conflicts and raise money, he said, but he was “not really in charge of the Lab.” His role was “to help collect the guano, make sure the water flows, do some composting and help make each of the things that are trying to happen, happen with taste and ease,” he said in the email. “More like Hagrid than Dumbledore.”

Publishing research in peer-reviewed journals is the coin of the realm across most of higher education. Not so at the Media Lab, where it’s seen as more important to develop a product that can be marketed — more like a Silicon Valley start-up. The Media Lab’s motto used to be “Deploy or Die,” an adaptation of the phrase “Publish or Perish.” But as Ito explained in 2018, President Barack Obama encouraged him to drop the “or die” part. Ito changed the motto to simply “Deploy.”

Tony Luong

For Caleb Harper, that Silicon Valley-esque ethos has seemed to be a natural fit. He grew up around the food business — his family owns and operates a 2,000-acre ranch in Kansas, and his father was a grocery store executive. But, as he tells it, his family warned him that their industry was the past and he should look to the future. So he decided to become an architect.

There was only one place Harper wanted to go to architecture school: MIT. He applied and was rejected, but that did not deter him. He went to see the dean of MIT’s architecture program with his acceptance letters from other schools and asked her to reconsider. Whatever he said must have worked: Harper enrolled at MIT in 2009.

A 2011 trip to Japan with Media Lab researchers to study the redevelopment of cities set him on his current trajectory. It was not long after the earthquake and tsunami that had caused a nuclear disaster in Fukushima. Harper says he met a farmer who told him his business was devastated because no one would buy rice from the region, afraid it was still contaminated by radiation. Harper also noticed that people in the United States were starting to become more suspicious of big agriculture, turning to local, organic, non-GMO produce. He decided to change his focus. “I moved all my work into the Media Lab,” he said.

Now Harper’s mission is to build a shared online platform for growing plants. That entails collecting billions of data points about how plants grow and creating a way for people to download what he calls “climate recipes” that they can use in their own food computers. One climate recipe might grow tomatoes just like the ones grown in Italy, for example, while another might grow particularly durable cotton. Everything is open source, he says, so anyone with a computer can access the data. On 60 Minutes, Harper described it as “Tuscany in a box, Napa in a box, Bordeaux in a box.”

The idea became its own initiative at the Media Lab in 2015; since then it has grown in scale and ambition. Harper hired researchers to conduct experiments using the food computers and sent the simpler versions of the machines to schools. Big companies like Target, Ferrero, and Welspun have partnered with the Open Agriculture Initiative to try to figure out better ways to grow basil, hazelnuts, and cotton. Harper also started a nonprofit, the Open Agriculture Foundation, that is meant to support the schools project. This summer there were food computers on display at the Barbican, in London, and the Cooper Hewitt, in New York. The devices are being billed not only as an answer to global food shortages but also as impressive feats of design.

Harper, meanwhile, was named a National Geographic Emerging Explorer and a member of the World Economic Forum’s New Vision for Agriculture. He has given talks in Brussels, Abu Dhabi, and at the United Nations World Food Program. He appeared on a panel at the White House alongside José Andrés, a well-known chef, and was paid a visit at the Media Lab by the actor and tech investor Ashton Kutcher. “By combining the things he loves in life — the technology of computing, architecture and engineering with his innate desire to grow things — Harper is not only changing the future of agriculture, but also changing lives in the process,” reads a description for a panel he appeared on at Harvard University.

The promise of solving big problems with a simple box attracted many of Harper’s former employees to the project. But some grew disillusioned, feeling that he repeatedly crossed the line from boldly optimistic to plainly untruthful.

Paula Cerqueira was hired by the Open Agriculture Initiative in 2017 to study how food computers could be used to teach children about how plants grow. Cerqueira, who has a master’s degree in nutrition education, coordinated a program that sent the smaller, tabletop versions of the food computers into classrooms, first to three schools, two in Massachusetts and one in New York, and then to a larger group of schools in the Boston area in 2018.

For the Boston project, the Open Agriculture Initiative had designed a version of the food computer that was supposed to be relatively easy for students to assemble. The new model had sleeker parts that slide and snap together. It came with a circuit board that read information collected by sensors to measure the temperature, humidity, and carbon-dioxide level. The circuit board could “create, edit or maintain the climate recipe running inside,” according to the initiative’s website. But few of the schools that Cerqueira worked with could get the food computers up and running, she says. Some did, but their seedlings had to be replaced repeatedly because the plants kept dying.

But in his 2015 TED Talk — delivered in Geneva a year before Cerqueira started working at the Media Lab — Harper told an audience that the students working with food computers “select recipes that have been created by other kids anywhere in the world. They select and activate that recipe. They plant a seedling. While it’s growing, they make changes.” He got a standing ovation.

For a Media Lab members’ event in the fall of 2017, Cerqueira says, she was sent to a grocery store to buy herbs. They were served to potential funders in a drink and in a dish, letting those in attendance assume that the greens had been grown on site.

At the event, Harper gave a presentation outlining the various research projects his initiative was running. For each project, he showed a slide that included headshots of the team members who were working on it. On one slide, titled “Open Ag Food Server,” Cerqueira saw her photo under images of the initiative’s shipping-container-size device. She was surprised: She hadn’t had anything to do with the development of the device.

Then she saw a slide for a project called “Open Phenome.” She says she hadn’t worked on that project, either. A few slides later, she saw one called “Farmers Eye,” which she didn’t know anything about but assumed was an effort to use machine learning to make predictions about plants. Her photo was included on that slide as well.

The members event wasn’t the only time Cerqueira was told to buy plants to bring to the lab, she says. Another time she bought lavender from a garden store in nearby Somerville, brought it to the Media Lab, brushed the dirt off the roots, and placed it in the food computer. That day a photo shoot was held at the lab. The images were sent to media outlets and appeared on the initiative’s website, Cerqueira says. “When I first came on board, I was under the impression that what they were posing as vision was actually reality,” she says. “Finding out that it wasn’t so was disappointing and made it incredibly difficult to do the work that I was hired to do.”

The food computers weren’t being used for the kind of scientific research that Harper was talking about, according to Babak Babakinejad, who was hired in May 2017 and promoted to research lead that October. Babakinejad had earned a Ph.D. in neuroscience and nanotechnology at Imperial College London. He wanted to use the food computers to test whether he could introduce certain microbes into plants.

But Babakinejad was never able to conduct those tests. He wasn’t given access to the smaller food computers, he says, although there were several in the lab. There was a large food computer at the project’s facility in Middleton, Mass., that he had access to, but he found that it wasn’t climate-controlled in the way Harper often described. The LED lights that were needed for photosynthesis gave off heat, causing temperature fluctuations, he says. The containers could grow plants, but he could not control their levels of carbon dioxide, oxygen, or humidity. He was not “coding the expression of that plant, the nutrition of that plant, the size of that plant, the shape and color,” as Harper had put it in his TED Talk. “He was projecting achievements that Open Agriculture didn’t make,” Babakinejad says.

As research lead, Babakinejad felt it was his responsibility to raise these issues with Harper and other members of the team, and he did so in an email. He also raised his concerns with Ito. In an email, he told the director that the Open Agriculture Initiative had not been able to create a controlled environment in the food computers, and that the devices had been sent to schools and a refugee camp without being tested to ensure that they worked. He worried that Harper was misleading funders. Ito responded by asking if he could raise these issues with Harper. Babakinejad agreed to let Ito share his general concerns. (Ito did not respond to a request for comment. A Media Lab spokeswoman declined to comment.)

For Babakinejad, Harper isn’t merely expressing a vision about the future of food. “It falls under the premise of using the brand of a prestigious institution to promote himself and to express fictitious ideas,” he said. “It’s not science. It’s more like science fiction.”

Later Babakinejad wrote a more forceful letter to Harper, dated April 2018, in response to a work-improvement plan: “We need to publish data that meets quality standards. In my professional opinion none of the data generated so far is worthy of publication in a reputable journal. We cannot publish false results and make untrue claims: No self-respecting scientists with integrity would do this.”

1/2

Tony Luong

The machinery required to operate the food computers in Middleton, Mass., includes circuit boards, cooling fans, and copper pipes. Researchers say it didn’t work as expected.

2/2

Tony Luong

Shelves of “food computers” in a project of the Media Lab at MIT.

Caleb Harper’s office at the Media Lab is tucked away behind glass walls on the first floor. During a nearly two-hour conversation there in August, he described his project, how he had developed it, and the criticism that there’s a mismatch between what he’s presenting to the public and what is actually going on in his lab.

“It’s vision versus reality, and both are necessary,” he said. His vision is to create a data platform that will revolutionize modern agriculture. He believes that if he doesn’t build that platform, the data will still exist, but it will remain in the hands of big companies who won’t share it with everyday farmers. “I have a pretty good handle on where this field is going,” Harper said. “So I talk about that. And because I’m so clear on that vision, I think people misinterpret that as reality.” He’s persuasive, quick to be self-deprecating, and eager to show photos of his work. He admits that he’s killed many plants over the years, and that his early food computers were works in progress.

Harper’s optimism helps raise money, and without money he won’t be able to see this dream of an international network of food computers come true. His critics, he said, “are basically jealous because I raise a lot of funding while giving away knowledge for free.” Harper also said that he doesn’t mislead the public. He’s explained his progress in great detail in a series of Medium posts, he said. Some may have misinterpreted his vision as current reality, he said, but if they listened closely they would not be mistaken. “Can you email a tomato to someone today? No,” he said. “Did I say that in my TED talk? Yes. Did I say it was today? No. I said, you will be able to email a tomato.”

It’s true that Harper didn’t quite say that food computers can email tomatoes or apples, though you could be forgiven for thinking exactly that. He frequently leaves the impression that the project has achieved, or is on the brink of achieving, an enormous breakthrough. It’s a style that has attracted the sort of high-profile attention, not to mention corporate funding, that fuels projects at the MIT Media Lab, and his willingness to showcase food computers beset with problems feels consistent with Ito’s “deploy or die” philosophy.

But somewhere there is a line that divides scientific reality from promotional hype. And it’s a line that, according to some who have worked closely with Harper, he has crossed again and again to make his project sound more successful and innovative than it really is. “It’s research. Sometimes things don’t work out,” says Cerqueira, one of Harper’s former employees. “But the thing is that Caleb was so misleading about that — to funders, journalists, and anybody that came by.”