Lift the stopper from one of the three antique-style colored glass bottles and hear a jazz ensemble riff spill out, as if the music had been freed, genie-style. Across the room, touch any of a dozen doughnut-sized plastic bubbles to see a ripple of light spread through the crowd like a rumor. Downstairs, stick your personal mug underneath the coffee machine spigot and hear your favorite station lock in on the attached radio.
Are we having fun yet? If not, there’s plenty more to amuse here at the MIT Media Lab, including festive holograms, dancing microrobots and computerized jean jackets. Even better, all of these gizmos demonstrate unique approaches to technology that, the lab’s faculty assert, will eventually be widely embedded in the world around us.
No wonder the Media Lab has long stood out as the high-tech research darling of the public; it’s just so damned, well, mediagenic. More important, it’s also the fair-haired child of industry, which sometimes literally lines up to witness the alleged future aborning and then hurls money at the lab-enough funds to allow the construction of a second building nearly twice the size of its current headquarters on the MIT campus, as well as a daughter facility in Ireland.
But while outwardly the lab seems to be sailing along as smoothly as ever in its 15th official year of existence, in fact it now faces a series of daunting challenges that could, by the faculty’s own reckoning, lead either to its attaining a vastly greater level of influence or to the enterprise’s dissolution. Among these challenges are a rapid expansion that threatens the lab’s renegade, collaborative culture; a management vacuum left by the shrinking role of founding director Nicholas Negroponte; the lack of a clear direction for its research; and a funding model that looks increasingly anemic and restrictive in an age in which almost anyone with a bright Internet-related idea can collect venture capital millions. “We’re obsolete,” says lab professor Neil Gershenfeld. “We seriously talked about shutting down. The feeling was we should either grow or shrink drastically.”
The choice, made after months of debate among Media Lab faculty and administrators and just now unfolding, has been to embark on a risky strategy for growth-not only by expanding the lab, but by seeking new relationships with industry that could make the enterprise look as much like a management consulting and VC firm as a university research facility. “We’ve changed before, but this is different,” says associate director Andrew Lippman. “In the past it was about new directions and details. Now we’re talking about a change in vision.”
A Bold Experiment Runs Its Course
Founded in 1985 by former MIT president Jerome Wiesner and then-architecture professor Negroponte, the Media Lab’s mission was ostensibly to explore the convergence of various media, a bold concept at the time. But that justification was, in a sense, a cover story: From the beginning, the lab was itself an experiment in a new way to run a university-based research organization. The key to understanding this approach, points out lab academic head Alex “Sandy” Pentland, is to recognize the tradition on which Negroponte and Wiesner were drawing-namely, that of the architect guild rather than that of a science department. Instead of learning primarily from lectures and books, students would learn more from doing; instead of paying or having to teach classes, students would be paid for producing original work; and as for faculty, instead of being judged for the papers they published, they would be valued according to the public impact of their groups’ projects. What’s more, the approach would be decidedly multidisciplinary. “Universities have iron walls around disciplines,” says Pentland. “In the guild, it was what you produced that mattered, not what went into it.”
The other part of the experiment was the Media Lab’s sponsorship model-really more a form of patronage. Negroponte made corporations this offer: Give his operation a chunk of money and gain access not merely to one research project, but to all research within its confines. Sponsors would be welcome to rub shoulders with faculty and students, ask questions, make suggestions and absorb the creative energy brimming inside. The downside was that sponsors wouldn’t get to specify or influence any of the research or obtain any unique rights to it, as they typically might in funding university research.
This might seem a pretty skimpy payback from the corporate point of view. But Negroponte proved remarkably adept at selling the proposition, bolstered by the growing perception that something magical was happening-the result of brilliant public relations and a flattering popular book by Stewart Brand called The Media Lab: Inventing the Future at MIT. “You could say we engineered the good press,” says Negroponte. “But the simple fact is that the media likes to write about the media.” In any case, money has not been a problem; the effort now counts some 175 sponsors who provide $33 million of the lab’s roughly $35 million annual budget (the rest comes from MIT and a few government projects). Individual sponsors typically ante up $200,000 a year, but some chip in more. Mastercard, Motorola and Lego, for example, have given $5 million each to the lab in exchange for closer-than-the-usual collaboration with certain research groups.
Of course, the lab has long been the subject of grumbling from rivals inside and outside of MIT. These critics charge that the place gets an awful lot of attention just for churning out cute, flashy toys that-unlike the outputs of, say, MIT’s esteemed but less glamorous Artificial Intelligence Laboratory and Laboratory for Computer Science-almost never break technological ground, prove significant scientific theories or end up as important products. “We all recognize and envy the high profile the Media Lab has,” says John Anderson, a professor at Carnegie Mellon’s Human-Computer Interaction Institute. “We also recognize that some of the work is kind of questionable.”
But though Negroponte’s creation tends to be perceived as a practitioner of research as an extreme sport, it is neither the exotic, future-inventing wonderland that is packaged to the public nor the tired repository of third-rate science-cum-eye-candy pilloried by its detractors. In fact, hype and resentment aside, the lab has carved out a fertile, if not as of yet terribly impactful, research niche for itself at the interface of computer technology, soft science and the arts-and the results could have at least an indirect influence on generations of products to come. “Computer science departments have people who are strong at figuring out how to make things, but you don’t see people looking at what these things mean and how you apply them,” says Ken Perlin, director of New York University’s Media Research Laboratory. “The Media Lab has broadened the definition of science to put together the science of things with the science of what people want, and that’s gutsier.”
That the Media Lab continues to be other than your typical academic department becomes quickly apparent in a tour of the building, which itself has the feel of a very large, colorful, high-tech toy. Mannequins, musical instruments, rubber chickens and rivers of wires are strewn throughout the glass-walled, mutedly lit workspaces. In addition to these inanimate obstacles, one also has to squeeze by an apparently never-ending international stream of journalists, camera crews, sponsor reps and visiting academics-not to mention students sprawled on the floor soldering or huddled around a computer screen in an act of communal debugging.
Although the energy is evident, the nature of the domain into which it is being poured has become increasingly unclear. Ask a dozen different professors what business the Media Lab is in, and you get a dozen different answers. “It’s about ‘me, means and meaning,’” pronounces Gershenfeld, who says he’s referring to the study of people, of embedding technology in everyday objects and of the content associated with these efforts. Walter Bender, who runs two large publishing-related research groups, insists that the lab studies “learning through expression.” Still others suggest the endeavor explores ways to weave technology more seamlessly into people’s lives.
This lack of a clear theme results in an almost scattershot feel to the research that takes place inside-as if the place isn’t so much inventing the future as making a run at interesting slivers of it. Consider, for example, the colored, stoppered bottles that “contain” music. These come from Hiroshi Ishii’s “tangible media” group, which explores how information can be transmitted and controlled by tactile means. The bottles are typical of many projects in that, at first glance, they come off as engaging and trivial, but after a little explanation the ideas behind them seem plausibly significant. “Current technology impoverishes the human senses,” explains Ishii. “If you have to manipulate objects with your hands, then visual information might not be the best form of feedback.”
Rosalind Picard’s affective computing group, meanwhile, strives to get computer-controlled devices to recognize tip-offs to a user’s feelings; a wireless voodoo doll that a desktop computer user is encouraged to hurl against the wall, for example, can keep track of which computer annoyances preceded the launching so that they can be identified and fixed. Lippman’s “Digital Life” group has looked at real-time variable supermarket pricing, so that the price of a container of milk could drop as it approaches its shelf date and specials could be tailored to a shopper’s preferences. Tod Machover’s performing arts contingent produces instruments that allow people with no musical training to create rich sounds. And the lab recently snatched up young Stanford physicist Scott Manalis to work on ways of integrating semiconductors and biological materials, perhaps resulting in chips that can be implanted in the body to monitor health.
It’s not easy to come up with a conceptual umbrella that covers this far-ranging activity. When asked how Manalis’ work fits into the lab’s research theme, Lippman ponders a long moment, then shrugs, “I don’t care.” Says Pentland, who works on wearable computers, among other things: “Maybe there isn’t a tight definition of what we do here. But there isn’t a tight definition of physics, either.” Pattie Maes, an expert in intelligent software agents, suggests that the lack of any clear theme is in fact the lab’s theme. “Most labs are getting better at a narrower and narrower domain of expertise,” she says. “We try to be different.”
The question of theme might not matter if it weren’t for the radical upheaval that the lab is about to undergo as it doubles in size over the next two years and expands into an attached new building, an orgy of clear glass and metal rods scheduled for occupancy in 2003. “Our biggest problem is being fresh again,” explains Picard. “With a PC and the Internet, the average 16-year-old can be as innovative as we are.” She and others agree that the enterprise will require more than a bigger play space to fend off diminished relevance in an era in which significant technological change occurs almost monthly, and more and more people and institutions are treading on its turf. It will also need a different model for managing itself, since Negroponte’s family-style approach to making big decisions has already fallen victim to sheer mass. “Five years ago you could hold a meeting that included everybody,” says Machover. “Now I don’t even know everyone in the building. It’s too late to go back.”
The Media Laboratories
The solution, to be implemented by the time the lab takes over its new building, is to divide the facility into three distinct and largely independent “centers,” each of which will be a Media Lab in its own right; the endeavor as a whole will pluralize its name to become the MIT Media Laboratories. One center, under the direction of Mitchel Resnick and already funded with a $27 million grant from Sega Enterprises chairman Isao Okawa, will be dedicated to education and Third-World development. A second, run by Gershenfeld, will focus on what the lab cloyingly calls “bits and atoms”-meaning how physical objects can be imbued with electronic intelligence. The theme of the third will be performance and the arts, to be “conducted” by Machover.
These choices have not been without controversy at the lab, especially among senior faculty. Bender, for example, argues for a simpler twofold structure based on the themes “Things that think” and “How do people use things that think?” Pentland and others lean toward adding a fourth center, though there’s no agreement as to what its focus should be. The real problem is that many faculty members-among them mainstays like Lippman and Bender-don’t see their work fitting comfortably into any of the three centers, even though the centers themselves are multidisciplinary. Apparently, some are even concerned that being shoehorned into one could mean being placed under a layer of management that won’t necessarily view their work as a high priority. (As this article was going to press, the lab was preparing for a summer retreat at which such issues were to be addressed.)
Gershenfeld somehow manages to sound both empathetic and cold-blooded about the queasiness with the tri-lab structure. “There’s a great deal of concern over the issue of people who feel they don’t fit in,” he says. “In some sense, it’s correct-they don’t, not anymore. There’s a certain amount of legacy here.”
Even some faculty members who aren’t fretting about fitting into a center say they’re worried about how the new structure will affect the lab’s unusually collaborative culture. Resnick, though he will be running one of the centers, shares that fear. “There’s a clear advantage to having clusters with intellectual coherence,” he says. “At the same time, doing anything that jeopardizes the cross-connections would be crazy. Finding that balance will be a problem.” So far, the only solutions proposed for keeping those connections alive are to maintain some common areas and to hold occasional shared events. Both Lippman and Bender say they fear that even if the centers succeed individually, they will leave the overall lab weakened.
Negroponte has by all accounts proven masterful at maintaining a sense of unity and mission among his charges-not to mention the lab’s sterling reputation among corporate leaders and the public worldwide. Although only 56, he plans to pull back from his management role, even as his brainchild enters what some feel is the most vulnerable period since its founding. Negroponte downplays the effect his reduced commitment will have, claiming he has long since faded from the front lines. “The lab is often identified as Nick’s Lab,” he says, “but these days I often sit in the back of the room. I’m more like a foreign minister, tooting the ML’s horn.” And it’s true that in an interview, he looks like a visitor in his own large, mostly bare corner office, which is routinely used as a faculty meeting room.
The lab isn’t discussing what it might do to replace Negroponte-nor would anyone indicate when any replacement might be made. But Gershenfeld says an executive board of “heavy hitters” will likely be formed to share decision-making powers with whoever becomes director. One person cited as a board candidate is John Seely Brown, just-retired director of Xerox’s Palo Alto Research Center. Brown won’t say whether he’d consider such an offer-but he sounds upbeat on the topic, noting that extensive conversations with key players have convinced him great things are afoot. “I’ve become an extremely strong supporter of this style of research,” he says. “I had lots of questions before, but now I’m more and more impressed with the direction of the lab.” Another heavyweight already in the lab’s camp is 3Com co-founder and TR board member Robert Metcalfe, who recently joined the Irish venture’s board.
However things are run, management will almost certainly face a rethinking of the venerable sponsorship model, which has been a source of the lab’s strengths as well as of many problems it now faces. Sponsorships have in effect served as collective MacArthur grants that have allowed research groups to follow their intellectual fancy. Because faculty members generally don’t have to impress peer-review grant panels, there is reduced pressure to publish in academic journals, though many do. (Tenure, says Negroponte, is awarded on the basis of “becoming famous in your field”-which can be less daunting than it sounds, since lab faculty members typically operate in tiny fields they founded themselves.)
In return for such unconditional support, faculty give backers personal attention-usually via students, who average two or three meetings a week with sponsors and ultimately become polished demo-givers. But because sponsors can only ask questions, and not pressure researchers to follow up on them, rarely does anything tangible seem to come out of the interchanges. Staffers claim that products do sometimes emerge from their efforts, but everyone offers the same two examples-an NEC airbag system that prevents deployment if the body-position sensors pioneered in the lab detect the presence of small children, and the Lego Mindstorm programmable “brick,” a direct offshoot of the computerized toys developed by Resnick’s group. Indeed, most faculty members make it clear that they don’t want to be working on projects that will directly lead to products. Says Picard forthrightly: “If we’re turning out products, we’re not innovating.”
The result of all this freedom from orthodoxy is that the lab tends to turn out work that, by the faculty’s own admission, doesn’t follow either conventional scholarly or commercial standards. Members stress that their mission is to do work that is keenly original. But as a recent immigrant from a hard-science lab, Manalis can’t help noticing that there can be a fine line between the original and the goofy. “In a traditional lab you have to dig really deep, but that’s not always encouraged here,” he says. “It’s easy to ride high and stay on the surface, and the work becomes faddish.” Agrees Judith Donath, who runs the sociable media group, “Some of the work is too flaky.” Another criticism that regularly pops up-typically from observers at other universities who speak only on condition of anonymity-is that the cozy relationship between the Media Lab and its sponsors gives faculty such freedom from normal pressures and oversight that it breeds a certain amount of arrogance.
But the compulsion to re-examine the sponsorship model doesn’t come from these sorts of criticisms, which the lab has been shrugging off successfully for 15 years. Nor does it come from corporate reluctance to keep the money flowing. Rather, it stems from the fact that over the past five years an even more generous provider of funding has emerged: U.S. capital markets. Suddenly, the Media Lab has found itself competing with billion-dollar startups on its own turf. The VC-backed startup strategy is “a higher-octane model than ours,” concedes Pentland.
Media Lab students have gone with the flow: Most used to take jobs with sponsors after graduating, but now the majority end up in a startup, often one they’ve founded around their work at the facility-and sometimes before finishing their degrees. Lab faculty are finding that in this new, more fluid environment, the Media Lab sponsorship arrangement actually entails more constraints than researchers at conventional institutions face. That’s because sponsors have the rights to everything developed in-house-and faculty aren’t supposed to take their work into their own startups or make private deals to sell it to anyone else. Students, though, aren’t constrained in these ways-and faculty members sometimes sit on the boards of their students’ companies. Conceivably, a professor could form a company and have it become a sponsor in order to get rights to his or her own work, but the lab made this approach less appealing by limiting faculty ownership of sponsor companies to 5 percent.
Some professors have managed to start companies related to their work anyway-most prominently Maes with Firefly, a Web comparison-shopping concern recently purchased by Microsoft, and Joe Jacobson with E Ink, which is developing new types of electronic paper that act like displays. But these spinoffs have reportedly engendered resentment within the lab and among sponsors. (Maes says she’ll be leaving at the end of the coming academic year, but insists her departure is amicable and motivated only by her desire to spend more time with her family and work with nonprofit organizations.)
To remain competitive in a VC-mad world, the lab appears set to reshape or augment the sponsorship model. Negroponte has held high-level meetings with MIT’s provost and engineering dean (see “Rough Sailing at the Media Lab,” TR March/April 2000). The aim, says Gershenfeld, is to win university approval for a scheme that would enable the facility to partner with other investors in startups based on its technology. Faculty wouldn’t directly own equity in the resulting firms, but some of the profits from the lab’s stake could flow to them-a plan he acknowledges could have dramatic effects. “The new revenue-sharing and IP [intellectual property] policies will make us look more like a company than a research lab,” crows Gershenfeld.
Pentland envisions an even more radical metamorphosis. He suggests the lab might evolve into a “meta-incubator,” essentially working with large companies to help them establish startups around Media Lab and other technologies, with venture capitalists and business-school partners coming aboard. “To many companies we’d look like a consultancy,” he says. “That ought to be worth a couple of hundred thousand [to a corporate sponsor].” Pentland says he has had informal talks about the plan with faculty at MIT’s Sloan School of Management, but adds that other universities also might be involved.
Still Taking Risks
Some insiders express fear that once the faculty starts thinking so much about the money that could come out of their ideas, they will hesitate to share their best thinking. “The change in trust and openness could rip us apart,” says Negroponte. “As soon as one person at the table isn’t open, it becomes an epidemic. We keep looking for ways to fight it.” Worries Pentland, “How do you manage that process without going down the slippery slope to hell?”
And what about the original mission to promulgate a novel style of academic environment? Largely ignored for years both inside and outside the lab, it is now moving to the forefront, as more and more senior faculty members start to question whether their enterprise has achieved the sort of pervasive, world-changing impact they once envisioned. The result is a plan to spread the model farther and wider by founding spinoff facilities. “This is a methodology,” says Bender. “We should be able to replicate it.”
The first such venture will be opening next year inside a former Guinness Brewery in the heart of Dublin. In affiliation with several as-yet-unnamed European universities-in Ireland and on the continent-the new MediaLabEurope will focus on the performing arts and media content. Latin America and Asia are being considered for a second spinoff. Besides extending the Media Lab’s reach, one benefit to these offspring is they can serve as relief valves for intellectual property conflicts at the mother ship, since their sponsorship models allow for easier migration of technology to startups.
It may seem a little foolhardy to tackle these sorts of spinoffs just as the lab finds itself grappling with its ambitious headquarters expansion and associated internal conflicts. On the other hand, says Negroponte, “We’ve always had a license to take risks.” So why put off safely for tomorrow what can be done dangerously today?
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Driving companywide efficiencies with AI
Advanced AI and ML capabilities revolutionize how administrative and operations tasks are done.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.