Big things, faraway things, crazy or obscure or commercially unpromisingthings that you need a PhD to even think about - that’s what people do at 200 Technology Square. A drab concrete office block in Cambridge, Massachusetts, Tech Square has been one of the centers of the computer universe for more than 40 years. Energy-aware lossless data compression? Amorphous and cellular computing? Biomechatronics? If any of that would keep you up until 3:30 in the morning, eyes wired to a monitor while shoveling down yesterday’s cold Chinese food, step right in.
Tech Square, formally the Massachusetts Institute of Technology’s Building NE43, is home to CSAIL, the Computer Science and Artificial Intelligence Lab. It’s a fluorescent-lit playpen stuffed with 600 in-progress or full PhDs and a thousand quirks. The pretty face of MIT - the iconic dome and quad on the bank of the Charles River, facing Boston - is a 10-minute walk away. This place is MIT’s scuzzy port on the postindustrial motherboard of eastern Cambridge. But inside are the tribal artifacts of intelligence at work: nine floors of networking cable and computers piled in sedimentary layers, couches bowed and stained by napping geeks, hallway whiteboards covered in arcana, and vintage machines chugging away in the homegrown programming favorite Lisp.
Stroll the fifth floor - quietly, please - and you may see Robert Morris, who released the first Internet worm (and is really, really sorry). On four is Richard Stallman, elder god of free software, who may or may not have another real-world address. On three is Tim Berners-Lee, who, when most of the world had still not heard of email, sat down and wrote the software for the World Wide Web.
Before he left to start supercomputer company Thinking Machines, Danny Hillis secretly wired the old building’s elevators to the Arpanet. The first true videogame, Spacewar,incubated here. So did Emacs. RSA encryption. Akamai. Tech Square is a Darwinian niche where human beings, bathed in the CRT glow of SubZero-sized minicomputers, evolved into hackers.
One of them is Gerald Sussman, who showed up here in 1964 at the age of 17 and never left. Coauthor of a classic text on computer programming, Sussman studies everything from orbital mechanics to automated chip design and shepherds a squad of eager students. Today he’s ignoring 6-foot-high stacks of orange plastic moving crates piled in the hallway.
Though Sussman and the rest of Tech Square might wish otherwise, moving day is coming. A hundred yards across Main and Vassar streets is CSAIL’s astonishing new headquarters: a gleaming 440,000-square-foot foundry for genius created by the world’s most famous designer of buildings. The Ray and Maria Stata Center for Computer, Information, and Intelligence Sciences is the latest CAD-spun, holy-shit wonder from Frank Gehry.
So all systems go? Ready for liftoff? Not exactly. "You can quote me," Sussman says, smiling a thin how-clueless-are-you smile, "I didn’t ask for it."
Love it or not, the two-tower, 2.8-acre Stata Center symbolizes everything the institute wants to be. Named for the cofounder of semiconductor maker Analog Devices (MIT class of 1957) and his wife, it may be more than a bit Bilbao for some of MIT’s engineers, but the interior is hackable, rack-mounted, and user-friendly. It has space for everyone from hardcore theoreticians and linguists to robot-builders. There’s even a special little street for MIT’s infamous food trucks to trundle up and shovel out Chinese and Middle Eastern delights by the pound. As the computer scientists say, this is nontrivial.
The Stata Center is the linchpin of a $1.4 billion bet that space and place actually matter in the production of esoteric knowledge. It’s MIT’s $280 million ante in support of the idea that the boundaries dividing science into warring tribes are literally antique, and that the mystery of how humans think can be cracked by putting 1,000 hackers and other assorted "intelligence scientists" under one roof (all right, lots of roofs). The notion that the best science is interdisciplinary and serendipitous has been percolating for decades; Stata is supposed to prove it. MIT’s mandarins bravely hope that whimsy and amenities might also soften the school’s boot camp atmosphere.
All fine stuff, but there’s still plenty about the building that puts geek teeth on edge - "silly" tilted walls, "wasted" space. It’s a bonfire of rectilinearity. Humor at CSAIL: "No worry at Stata about earthquakes - it already had one."
Doors officially open in May, then the real experiment begins: What happens when you take a vacuum bottle like Tech Square and open it to light and air? Can pocket-protected life survive on Planet Gehry? "Maybe it will destroy us. Who knows?" says the lab’s director, ever-playful Aussie roboticist Rodney Brooks. "I prefer to be optimistic."
And he has every reason to be. The Stata Center is, well, a quantum step up. Tech Square, vintage 1959, was the kind of place only a troglodyte could love. Besides perpetually hissing air vents and windows that didn’t open, its cramped floor plan pigeonholed researchers already wary of venturing beyond their own little domains. Sussman’s team dubbed their fourth-floor haunt Switzerland ("neutral, but heavily armed"), a reference to decades-old feuding between the lab’s computer science and artificial intelligence wings. In data networking lingo, Tech Square was stovepiped.
It was also a bad answer to some key questions facing research-oriented institutions like MIT: How do you make research pay for itself? How do you convince potential corporate funders that you’re really part of the 21st century? And how do you get tenured scientists to, er produce? In a perfect world, none of that would matter. But then, in a perfect world, MIT wouldn’t have to compete with the university up the street that shall remain nameless, not to mention Caltech and Stanford, for the best faculty, the best students, the biggest grants, and general bragging rights, otherwise known as Nobel Prizes.
The solution (sotto voce - we’re still in academia here) is to surround the creative types with other, equally creative people who will challenge their assumptions. Peter Galison, a physicist and historian at Harvard - there, we’ve said it - traces the idea to the 1930s, when European theoretical physicists fleeing Nazism made their way to the US and found themselves working with more pragmatic Americans. Then came World War II and the birth of Big Science. "The war changed all the rules," says Galison. "At places like Los Alamos and Oak Ridge and MIT’s RadLab, you had mathematicians and theorists literally sitting on the other side of the desk from engineers. It was transformative."
Spencer Reiss (spencer@upperroad.net) is a Wired contributing editor.
Galison calls these places trading zones, where different scientific traditions and disciplines intersect. Those first trading zones were accidental, including arguably the signature science of the 20th century, computing - forged from physics, math, and electrical engineering. Now, going into the 21st century, the most promising fields of science encompass even newer specialties. Networked pursuits like nanotechnology and genomics, Galison says, demand trading zones and researchers working together in trading zones, developing what he calls pidgins - the grammar of one field, the syntax of another, vocabulary from both.
MIT once operated an early trading zone, the legendary Building 20, a "temporary" structure built in 1943 to house 4,000 wartime radar researchers. It was here that Harold Edgerton devised his underwater cameras; Noam Chomsky more or less invented modern linguistics; Amar Bose is said to have spent several years surreptitiously testing loudspeakers in the anechoic chamber. MIT’s Jerome Lettvin and Harvard’s Timothy Leary painted psychedelic murals in C Wing, hoping to settle an argument about the artistic effects of LSD. And it was home to the famous Tech Model Railroad Club, whose networks of crisscrossing toy trains were a breeding ground for hackers. Building 20 stayed in use until 1998, when its rotting, sagging, asbestos-lined shell was demolished - with much hand-wringing - to make way for Stata.
When the idea of moving Tech Square’s geeks first came up in 1993, ambitions didn’t extend much beyond another faceless building. By the late 1990s, the project had evolved into the centerpiece of a campaign to remake MIT. It would confront the great intellectual challenge of understanding human intelligence. It would be a statement of IT’s indispensibility to science. It would be a (polite) kick in the pants to the geeks to get out in the world more. And it would be a gateway to the hot new biotech companies starting to wave research dollars at Tech Square.
The question was, Who could come up with a single structure to bear all that load?
Frank Gehry was not a shoo-in when he gave a tour of his European projects to members of Stata’s "client committee," charged with making the new building happen with the fewest possible blown fuses. Bilbao, with its titanium-skinned Guggenheim Museum, clinched it, at least for Chris Terman, the committee chair. "Arriving at the museum, our car bumped over the curb, drove across the pedestrian plaza, and stopped right at the entry stairs," says Terman, a compiler jock in real life. "The driver turned around and told us: ’Nothing’s too good for Mr. Gehry.’ It wasn’t just the architecture, it was the attitude. What he did changed that city forever."
The power to transform is why Gehry got the call from MIT’s president, Chuck Vest. Well, that, and maybe his stories about watching the sun rise over Los Angeles with that hippest of all physicists, Richard Feynman (MIT class of 1939). And maybe it also helped that MIT’s longtime dean of architecture, William Mitchell, was both an old friend and one of the academic world’s most energetic advocates of computer-aided design. What better than replacing Tech Square with a building by the main proponent of CAD?
But warning lights immediately started flashing. "MIT is a place where people have some limited understanding of what titanium costs," one researcher deadpans. Then Gehry and his crew wheeled out little models for the interior space. One was dubbed Orangutan Tree Village, with group activities on the lower floor and private offices above. An inverted version was Prairie Dog Town. There was Japanese House, with movable partitions. And there was Colonial Mansion, with a mixture of offices and open space. "We all knew how crazy they were," says Terman. "Some people really thought they might be serious." My God - they’re going to turn us into the Media Lab!
There was method at work: Gehry hoped to get Stata’s future residents thinking about space as more than something you order by the square foot. Some checked out of the process, resigned to their fate; others were impressed by the Gehry team’s collaborative approach, the great man surrounded by bright acolytes feeding the fires of creativity - a version of their own way of working. The new priorities: light, air, and interesting views, along with the mantra adaptability. "We were struggling to break through - they just wanted what they had," Gehry says. "They loved Building 20 because they could beat it up. So I said, ’How about a building where you feel comfortable banging the walls out, putting up stuff?’" Building 20.1! Well, sort of.
Gehry got his soaring glass and "articulated" exteriors. The geeks got a high-ceilinged warehouse space on the second and third floors, topped by a nonidentical twin pair of six-story towers (a vestige of the days when the CS and AI leaderships were on less than cordial terms) named for photo innovator and entrepreneur Alexander Dreyfoos Jr., MIT class of 1954, who ponied up $15 million, and Bill Gates, who kicked in $20 million. Many of the 19 research units have double-height open areas and spiral staircases. There are seminar rooms and lunchrooms, exhibition spaces and the first amphitheater on campus. Stata has a gym, a day care center, and a basement pub floored with timbers saved from Building 20. Beside the Gates Tower is a 350-seat auditorium, with remote-control cameras connected to classrooms in faraway Singapore and the other Cambridge. Down front is not just any blackboard - it’s a 50-foot-wide, triple-banked Sistine Chapel of the most classic chalkboards you’ve ever seen. Because there is just so far you can push the MIT faculty.
What Stata is not is "smart" - no arrays of sensors or touchscreens in the walls, no biometric infrastructure. "You don’t want to be maintaining a working prototype for a thousand people," Brooks says. Instead, there’s ubiquitous wireless and a fiber-optic switching fabric that puts a 10-Gbps network within easy reach of every floor. Its heart is a back-to-the-future machine room - the "CPU" in Gehry’s plans - stacked with Cisco 6500 series switches, Avaya Light Intercept Units, and Internet-addressed APC smart racks. Chilled water to keep it all from melting arrives through a 30-inch main from MIT’s central refrigeration plant.
So what’s a good geek to do with all that? What they’ve always done - take on really hard problems, tinker, write code, think. But also interact. Connect. Roughly 40 percent of the total floor area is devoted to collaborative space, an experiment in supply-side behavior modification. And Stata’s soaring main public passage, Student Street - an extension of MIT’s famed Infinite Corridor - is lined with glass-walled labs for crowd-pleasers like robots or a machine shop. "The hypothesis is that by connecting more people and opening up to the rest of the campus, you’ll get more interesting work," says Brooks.
A classic example of Stata-style research is Thomas Knight’s synthetic biology lab. A hero programmer from the 1960s, Knight now hopes to create organisms for use within microelectronics. His group ties high-powered computing into a microbiology lab, where they can try building, say, plants that tell you when they need water. But even Knight, who spent the past decade tooling up his skills in graduate biology courses, wishes he had less atrium and more exhaust vents. Then again, he does have Frank’s permission to start banging.
Gehry’s team wasn’t above trying to pander: Next to the machine room, a windowless space has been set aside and dubbed the Holodeck. Howard Shrobe, who heads CSAIL’s Intelligent Room Group, is still scratching his head about what to do with it: "Maybe we’ll see how much of Star Trek we can re-create."
Given Stata’s crew, it’s no surprise that a lot of the gripes focus on their caves - or lack thereof. Stata has 370 lockable offices for 1,000 people, and math is something this crowd is good at. When Microsoft comes to recruit, notes one doctoral candidate, "one of their big selling points is: ’Our offices have doors.’ With the kind of focused work people do here, spin-up and spin-down times are excessive." (Translation: Distractions are bad.) Gehry’s team came up with plywood partitions. To which the grad students answered: Dilbert.
The other great whine is wasted space. The four-story atriums at the bottom of each tower are explained in design specs as accommodating "experiments that require tall spaces, such as those involving remote-control helicopters." You don’t need a PhD to spot a rationalization. Says a senior programming researcher: "I’ve already got an architect looking at what we might do to fill ours in."
And then there’s the square-corner brigade. As recently as November, with only finish work remaining, one emeritus faculty member informed CSAIL director Brooks that "the slanted walls are unacceptable and must change." Go geeks.
Brooks himself had a characteristically wiggy but fun idea: Make the building expressive. "Not just whether the lights are on, but how many people are inside, what they’re doing, how many packets are moving on the backbone," he says. "In other words, give the place an active skin."
Gehry was game. So over the Gates Tower entry he added a 30-foot crystal ball to the design, onto which Brooks could project movies, data, colors. "The MIT guys got all excited," Gehry says. "Then Alex Dreyfoos wanted one, so we added one for his tower. But they’re $5 million apiece, and the money wasn’t there." So much for the crystal balls.
The money question has been lurking in the background since the day Mr. Bilbao signed on. "People thought I was going to spend everything on ’Frank Gehry’s building’ - on titanium or something," Gehry says. "But there’s no titanium on it. Well, maybe a little, over those doorways on the back."
The efficiencies of Gehry’s computer-powered building methods, say Mitchell and other CADvocates, mean that his design idiosyncrasies add only 10 percent to the total cost. In exchange, you get a vastly more interesting building, not to mention a lot more corner offices (plus a 700-car, two-level underground garage). MIT officials swear by $283.5 million as Stata’s final cost, roughly $650 per square foot. Meanwhile, the University of Washington’s new Paul G. Allen Center for Computer Science and Engineering has about a fifth of Stata’s academic space; UW officials crow about bringing it in for a mere $72 million - 30 percent more per square foot than Stata. Maybe Boston is a cheaper place to build. Or maybe geek suspicions of art run rampant are wrong.
You might even get what you pay for. Bill Gates’ tower in Cambridge is undeniably more provocative than his old friend Paul’s box in Seattle. "We think MIT buildings should be as intellectually interesting as the work that goes on inside them," says Mitchell. Not exactly surprising coming from an architect, but he goes on: "To be successful today, a lab needs a strong brand, especially in dealing with the corporate world." That’s an idea practically invented at the MIT Media Lab (where, as it happens, Mitchell recently became boss). So chalk another one up for discipline mixing: For a great 21st-century research institution, architecture turns out to be marketing.
MIT is, after all, a business. It has almost 10,000 employees and a $1.3 billion budget (plus a $35 million deficit this year). The $5.1 billion endowment kicks in a quarter of the necessary revenue; there’s also tuition and fees. But the biggest single revenue stream - nearly 40 percent of MIT’s annual income, and rising - is research funding: $470 million in 2003. Labs such as CSAIL raise their own research money - and give 57.5 percent right off the top to MIT, a cut known as "facilities and administration recovery." So Stata is an investment. Fall short on your grants and, as MIT executive vice president John Curry puts it, "the question becomes, Can the space be put to more productive use?" Brrr.
That’s why MIT’s current building program includes three multidisciplinary labs, costing a total of $600 million, and zero new housing for English (or physics or engineering, for that matter). Hot labs attract "entrepreneurial" faculty, who in turn attract - and sometimes spawn - companies. The dream for Stata and the neuroscience building under construction next door is that they’ll someday anchor a whole new high tech neighborhood north of the campus. Novartis, Biogen, and 75 other companies are already there. "The days of MIT living off federal agencies are over," says Curry. "Private industry is where the demand is, and what a potential sponsor wants to see is a facility that meets those needs."
None of this will come as news to the Stata Center’s researchers, who like to complain that they spend as much time writing proposals as doing actual research. But it does beg Rod Brooks’ worry: Do geeks die when exposed to light? Or will Gehry’s space help them reach unforeseen new heights?
Stata may be MIT’s most expensive experiment ever, but it has the cynical virtue of being what researchers call non-falsifiable - there will be no way to know what might have happened had Tech Square’s residents never heard of Frank Gehry. So why worry? Everything’s for the best in the best of all possible labs. "What we’re seeing is a reconfiguration of the ways we think about knowledge," Galison says. "The labs we’re building today are for people who say things like, ’I don’t know whether I’m mathematician or an engineer or a physicist.’" Big idea - but Stata’s a big building. Go nerds.
credit Photographs by Jonathan A. Manzo
The Stata Center for Computer, Information, and Intelligence Sciences at MIT.
credit Photographs by Jonathan A. Manzo
The pocket-protector crowd jokes about Gehryés million lab: "No worry about earthquakes-it already had one."
credit Photographs by Jonathan A. Manzo
credit Photographs by Jonathan A. Manzo
"Maybe it will destroy us. Who knows? I prefer to be optimistic."
Rodney Brooks, Director of the Computer Science and Artificial Intelligence Laboratory
credit Photographs by Jonathan A. Manzo
"Thereés no titanium on it. Well, maybe a little, over those doorways on the back."
Frank Gehry, Architect
credit Photographs by Jonathan A. Manzo
credit Photographs by Jonathan A. Manzo
credit Photographs by Jonathan A. Manzo
A lounge and a lecture hall (opposite), plus a view up through the four-story atrium, which shows off a few precious office windows and lots of empty space for ... well, no oneés really sure.
credit Photographs by Jonathan A. Manzo
"We think M.I.T. buildings should be as intellectually interesting as the work that goes on inside them."
William Mitchell, professor of Architecture and Media Arts and Sciences
