Inside the MIT Media Lab

Inside the Lab

Sensory overload:

You step out of an elevator on the third floor of a university building. You're in a beige carpeted corridor, with whitewashed walls on one side. On the other side, a glass wall separates you from the Disruptive Technology Laboratory. Open the door and look inside. The lab is about thirty feet square, with small offices off to either side. It's cluttered with open plan desks; at one corner a cluster of black sofas sit in a circle around a big television set with a stack of video equipment and an Xbox. At the opposite corner, there's a bench with oscilloscopes, soldering irons, and the other detritus of electronic prototyping.

In one corner, a bunch of students are trying to reinvent the wheel — specifically, the car steering wheel, which they're trying to add intelligence to. (Imagine you're driving a car, approaching a busy roundabout. Your hands tense and your pulse rate soars — it's a bad time for your cellphone to ring, isn't it? That's why the intelligent wheel knows enough about your emotional state to switch the incoming call to voice mail until you're calm enough to talk, and turn down the radio volume while it's about the job.)

You look around the back of the room. There's a sign identifying an experiment in Borgables. Under it you see the waistcoat that ate Silicon Valley, an unlikely offspring of a mating between a sewing machine and a laptop computer, bristling with memory, sensors, and i/o devices. (It runs Linux, of course. Linux is a favourite tool of hackers because it's so reconfigurable. These guys love it: they love it so much they want to wear it.)

Take the elevator down a floor. Walk through another glass door beneath a sign proclaiming the Opera of the Future, and you find yourself in a room full of brightly coloured balloon-shaped musical instruments plugged into a rackful of experimental electronics. The instruments are actually the user interface to a whole bunch of computers that comprise the Children's Symphony. The object of the project isn't to reinvent the Stradivarius but to change the way toddlers learn to make music — by giving them brightly coloured toys that provide immediate feedback, letting them explore the shape of sounds for themselves rather than struggling for years to master the piano keyboard or the guitar fretboard. Against one wall there's a table covered in gadgets that look like oversized computer mice. They're percussion instruments — toys that you can teach a beat to. Then you can beam the sound you just invented to your neighbour's beat bug, until you've taught the entire band a new rhythm.

Take the down elevator again and you're in the quantum computing lab, next to a two-metre high dewar flask full of liquid nitrogen. This is where they're trying to build a quantum computer — exploiting the eldritch physical phenomenon of quantum decoherence to solve complex iterative problems in linear time. (It's a bit of a culture shock after the children's symphony and the sympathetic steering wheel, but you're beginning to get a feel for how off-balance a tour of this building can make you — if you expect a random surprise around every corner you won't go wrong.)

Walking down a corridor you pass a room where a bunch of postgraduate students are in a seminar with two professors — one of computer science, and one who is an expert in the history of cookery. They're inventing the kitchen of tomorrow, explains your guide.

Round the bend you come into a huge open-plan room where Seymour Papert (inventor of Logo and pioneer of computer education for pre-teens) and his research students came up with Lego MindStorms. At the other side of the lab there's a comfy sofa in front of a webcam and a video projection screen — one of the ongoing six-way teleconferences that knit the Media Lab campus together. You just plonk yourself down, pick up the trackball to select a window, and wave hello to someone, somewhere — a far cry from the stiff formality of a pre-arranged video conference. On a bench at one side of the room there's a stack of Lego bricks and some microcontrollers. At the other side of the room the students are working on personal media — the convergence between weblogs and video, or collaborative tools designed to let classrooms of children build their own newspapers: there's another group building a software environment that lets you compose music by painting in broad swatches of colour (and turns the resultant picture into conventional musical notation, as well as playing it).

Fleeing the open plan environment full of brightly coloured Lego parts and video cameras, you find yourself in a machine shop full of robots, laser cutters, and prototype inkjet printers that print integrated circuits instead of pictures. There's a hard engineering back-end behind the brightly coloured toys — you're slowly realising that most of the experimental gadgets surrounding you were built right here in this building by research students and engineers.

The atmosphere in the media lab is elusive and very unlike a traditional university department, but one thing you can't help noticing is the tremendous buzz of enthusiasm and creativity that hums through it.

The Media Lab is a cross-disciplinary collision between the arts and the sciences. Computer science, as Edsger Djikstra observed, is about computers the way astronomy is about telescopes. Most computer science departments in academia emerged as spin-offs of the mathematics department, and indeed the study of algorithms is the major preoccupation of most academic computer science departments. The Media Lab, by this yardstick, is not a computer science faculty. It's something else — brash, loud, obsessed with human-machine interaction, always open to new ideas about how to apply design and architecture to new ways of living.

For another thing, the whole approach to research and industrial partnerships and work is almost the antithesis of a regular college. Students are here to explore; if they go off-course from their original track, they're encouraged to pursue their new interests. They're not here to learn a trade and go forth to fit a clearly defined role in industry when they graduate; more often, corporate sponsors come here to figure out what they ought to be doing, then hire graduates to do it — or invest in their startups. Above all, there's a playful feel to the Media Lab. To someone used to the work ethic of a conventional university it feels almost sinful: these people have turned research from a vocation into something you do for fun, education into recreation, invention into art. And the results are startling, if seldom predictable.

Down in the basement, I spent some time chatting to graduate students working on research degrees, trying to get a handle on some of the developments the Media Lab is currently coming up with. Take, for example, Yale Maguire, who together with some of his colleagues invented ThinkCycle. 99% of research never gets published, Yale says. We got thinking about trying to create a systematic way of doing research. Most research work stays in the laboratory or worse, in the head of the experimenter. Meanwhile, vital problems go untackled because people with the knowledge to develop a solution for them never know they exist. We want to use the spare processor cycles of human brains, Yale says, pointing to SETI@home as an example of using spare computer processor cycles to solve problems.

ThinkCycle provides a central database of scientific challenges, and allows people to pick challenges off the database and submit answers, lessons learned, and their results. It's a weblog for basic applied science work, and it reaps dividends. Among the inventions ThinkCycle has already come up with are a low-cost water purification system for use in the developing world, a passive incubator for prematurely born infants, and new cholera treatments. It's an impressive example of what open collaboration can achieve — but it's only just starting. Why not get students working on spare unsolved problems rather than solving contrived problems in class for their coursework? asks Yale. Every year thousands of engineering students plod through the identical questions for their course credits. ThinkCycle might change that, harnessing some of that brainpower for productive tasks while providing their teachers with feedback on how well they've tackled the problem.

Then there's the Fab Lab.

Rehmi Post is just wrapping up his PhD. Among other things, he teaches a three month class to new students, titled how to fabricate (almost) anything. Students start with using CAD packages, then he takes them through fabrication of parts using machine tools, how to design circuit boards, and — literally — just about anything, up to and including MEMS, microelectromechanical machines etched out of silicon wafers using the same lithography techniques as microprocessors. One thing we've learned in the course of this study is that the fabrication tools currently available all suck, he says.

Which is why he and some other researchers are working on the Fab Lab. The goal is to build a toolkit that can be sold for under $10,000 (£6500) and that contains everything you need in order to make almost anything. We want to take arts and crafts to a level where people can do their own prototyping, build their own radios, oscilloscopes, or computers, and do it on the cheap with full support in tools and hardware. He's not kidding. The Fab Lab — personal fabrication — includes a CAD workstation, a modified vinyl cutter able to carve circuit boards, a computer-controlled milling machine, an FPGA programmer, and may eventually include a 3D printer and other machine tools. One important element they're working on is a library of electronic components, royalty-free, than the system can be used to handle various tasks. Using FPGA (field programmable gate array) chips means the system can contain sophisticated electronics — FPGAs are designed to be reconfigured at the hardware level to emulate arbitrary circuits, all the way up to an ARM processor. The Fab Lab team are trying to develop a system comprehensive enough that any one Fab Lab can be used to build copies of itself, and they're looking at a hardware design strategy akin to the GPL (GNU General Public License) — spin offs such as Project Pengachu give a feel for how they're thinking these tools can be used.

The Fab Lab is basically a prototyping toolkit — what happens when Media Lab technologies get rolled out on a broad scale? Internet Zero is another of the basement projects, worked on by Raffi Krikorian. Light switches are a dumb technology. What happens if you make a light switch smart? If you make every electrical outlet in your dwelling intelligent? Raffi has a demo he likes to show off, of a halogen lighting track with some lights on it, and a switch. You push the switch and the lights go on and off. But what's happening isn't a simple electrical connection. Raffi's light switch contains a microprocessor with a TCP/IP stack and a simple web client. When you push the button the client sends an HTTP request to the nearest light bulb-controlling computer, which just happens to be running a web server and handles the request to switch the bulb on.

Sounds silly? Imagine your light switch has an extra brain cell or two. It can poll the light bulb while it's on via SNMP and ask what its state is. When it's nearing its life expectancy it can ping your PDA and tell it add light bulb to shopping list. Or it can sense when a human enters a room and turn the light on automatically, adjusting to the occupant's expressed lighting preferences by polling a device somewhere on their person.

Internet Zero is what happens when everything is on the internet, from your light sockets to your stereo, the microwave oven, and your front doorbell. (Who rang it while you were out of the house? Now you can find out.) It hasn't happened yet, but it's going to. And when it does, there's one thing you can be sure of — the Media Lab will be where it happens first.

Where the lab came from

Through the history of computing, a few institutions have carved out a special niche for themselves. First, at Bletchley Park, Alan Turing and Tommy Flowers assembled one of the first computers — the Colossus — to break the German Enigma codes during the second world war. Then, during the post-war eras, several academic (and commercially funded) research labs drove the field forward: Manchester University pioneered civil computing in the UK, Xerox PARC in California invented the laser printer, graphical user interface, and ethernet in just one five year period, and IBM and Bell Labs between them seem to have invented everything from the hard disk drive to UNIX (and a whole lot more).

The institutions all had certain common characteristics. Researchers were under time pressure or (indirect) financial pressure. There was a strong emphasis on applied research combined with academic excellence — not simple product design or ivory tower theoretical studies, but research into how best to apply cutting-edge theories to current problems. And by bringing together a critical mass of bright, highly motivated people and encouraging them to talk freely (in most cases — Bletchley Park was something of an exception!) they cross-fertilized.

And then there's the Media Lab at MIT (the Massachusetts Institute of Technology). The Media Lab is not your ordinary university computer science department, and not just because it was founded by an architect. In fact, trying to sum up what the Media Lab is about in less than a page or three is next to impossible — in 1997 Stewart Brand started to write an article about it and ended up with a book. So here are the bare facts:

Back in the late 1970's, Nicholas Negroponte was working at MIT, researching the interface between architectural design and information technology. He was already interested in the way buildings adapt to the uses to which their inhabitants put them; he was becoming interested in the idea of pervasive computing. The picture he and his fellow researchers were putting together was of the coming intersection between TV/video, ubiquitous computing interfaced to people by unobtrusive mechanisms (compared to which the traditional keyboards and screens would seem as clunky as a punched-card reader), and architecture and design. Much as architecture exists at the intersection between structural engineering and art, he realised that the computing revolution was going to require a new type of architectural discipline — one that combines all forms of media (narrative and story-telling, gesture, music, film, touch, and smell) with the new pervasive computing environments and new ways of allowing computers to interface with their surroundings.

Negroponte, with backing from Marvin Minsky and other MIT professors, put together the plans for the Media Lab in 1982. The Lab is a postgraduate research faculty; with 40 full-time staff and another 80-odd part time and visiting professors, and 150 postgraduate students, it's a large department. While it issues postgraduate degrees in Media Arts and Sciences, it also conducts original research in conjunction with the private sector. The Media Lab is one of the world's most successful examples of a working partnership between industry and academia.

Industrial Research — the Media Lab and Industry

American universities have several types of funding. The Media Lab is relatively new, and doesn't have the enormous endowments of faculties like Harvard. As a consequence, most of its work is funded by industry — in an innovative way that seems to generate new companies and successful commercial applications and spread the Lab's research results far and wide even as it raises money. One thing the Lab can't be called is under-funded — the new wing under construction is estimated to cost $200M. So how do they do it?

Everyone at the Lab works in one or more research groups. The list of research groups is more than a page long, from Aesthetics and Computation to Tangible Media by way of the Future of Learning group and Robotic Life. But each of these groups is affiliated to one or more Research Consortia — there are five in total — to which their work is relevant. The Consortia cover Digital Life (technology that spurs human expression and social activity), Digital Nations (addressing major social challenges — education, health care, community development), Information:Organized (trying to tame and control information overload), Things That Think (adding brains to everyday objects), and Changing Places (a joint Media Lab and Department of Architecture consortium on new design, construction, and infrastructure concepts for buildings).

Companies who want to sponsor the Media Lab basically buy a stake in one or more Research Consortia. By sponsoring a consortium, the company gets complete access to all the research being conducted by all the research groups in the consortium — not just a chance to read the papers or visit the lab, but opportunities for Media Lab researchers to take their work out to the sponsoring company's labs to demonstrate it, and an early opportunity to make licence-free and royalty-free use of techniques being developed by the lab. Often, students graduating from the Lab take their research projects with them to a sponsoring company, which sets up a division to turn it into a successful commercial product — it's a very real pay-off that expects to generate profitable results. Alternatively, students often find investors for their own startups via the corporate sponsors. There's a heavy emphasis on collaboration with sponsoring organisations; it's not like a company simply endowing a university with the money to pay for a chair of Computer Science.

One point that the Lab directors emphasize is that the Lab acts as a kind of commercial imagination department for the sponsors — the students are encouraged to think divergently and explore new ideas, in a way that employees of a company probably don't have time for. A lot of new ideas turn out to be dead ends, but sometimes they turn into big money spinners. Lego's Mindstorms revitalized the venerable educational toy and won it a whole new market, and was a direct result of Media Lab research into education and learning; the RFID chips being deployed in inventory control today are another Media Lab spin-off.



About this Archive

Find recent content on the main index or look in the archives to find all content.

Search this blog