November/December 2008

Open Up and Say Eureka

Innovation can come from anywhere when hardware designs are freely shared.

By Kate Greene --- MIT NEWS, Technologyreview.com

Limor Fried '03, MEng '05, was on the train from New York City to Boston recently when the woman next to her called a friend on her cell phone and began chatting about uncomfortably intimate matters. They were traveling in a designated quiet car, where phone calls and loud conversations are verboten. But even though the woman hung up when the conductor asked her to, she soon placed another call, again regaling fellow passengers with the gory details of her personal life. Fried, who studied computing culture at the Media Lab, decided to take action. She surreptitiously pressed a button on a pocket-sized gadget she'd designed as an MIT grad student. Fried's so-called Wave Bubble jams cell-phone reception by creating destructive radio frequency interference. When the chatty woman's call suddenly dropped--and her repeated attempts to reconnect mysteriously failed--she finally opened a book and began reading, silently.

Alumna Limor Fried sells kits for open-source gadgets like this device for playing the Game of Life, for which she retooled a kit by the MIT engineers at Dropout Designs.  Credit: Jordan Hollender

Fried's frustration at having to listen to other people's phone calls led her to create the Wave Bubble as part of her master's thesis. But she's no lone vigilante. In fact, she freely shares the design of her secret weapon with anyone else who wants to rid public spaces of intrusive cell-phone conversations. She posted instructions and a parts list for her gadget online, enabling anyone to make a cell-phone jammer--with the caveat that it's illegal to use one. And by sharing the design on the Web, she neatly sidestepped the U.S. Federal Communications Commission's prohibition on distributing the devices themselves.

Today, Fried remains committed to the idea of sharing her designs--and her engineering expertise. She runs her own company, Adafruit Industries of New York City, through which she sells kits that allow people to undertake less subversive projects, such as making an iPod battery-pack charger and transforming a bike wheel into a customized LED display. All her kits are based on designs that can be modified or improved by her customers. She also finds time to host an Internet video program, called Citizen Engineer. In a recent episode, she and cohost Phil Torrone explained how cell-phone SIM cards (the chips that identify a user's account and network) work and how to hack them to, say, recover deleted text messages.

"I'm helping people learn electronics," Fried says, "and also ensuring that the information will always be available." And when users adapt and enhance her designs, she says, it helps make her products better.

Fried isn't the only one freely sharing her projects and knowledge. In fact, open-source hardware, as it's called, is increasingly popular among all types of engineers, from students and hobbyists to entrepreneurs to engineers at large companies like Sun Microsystems and Nokia. The basic idea is that if the parts and designs for devices--and electronics are only one example--are available to the public, then more people can modify the designs to fit their specific needs. By posting designs online, hardware engineers have found a huge community of consumers and fellow professionals who are willing and able to provide feedback, and even to extend the designs in ways their creators hadn't thought of. In effect, the open approach to hardware design can serve as a massive, distributed research-and-development effort, and even a quality assurance program on the cheap.

Sloan School management professor Eric von Hippel, SM '68, describes these ideas in his book Democratizing Innovation--which anyone can download free from his website. One of the world's leading proponents of open technology, von Hippel, who is also a professor in the Engineering Systems Division, maintains that products are best designed and modified by the people who actually use them. While it is prohibitively expensive for a company to design, prototype, and test-market every design tweak that customers could want, motivated customers will take on those tasks themselves to get exactly what they want without having to wait for it.

The concept of open technology isn't new, of course. For more than a decade, technically inclined people have been modifying the underlying code of open-source software such as Linux. Once an obscure operating system for geeks, Linux is increasingly going mainstream; the numbers are hard to track because users aren't required to register their systems, but Dell currently ships five models of computers with the Linux-based operating system Ubuntu preinstalled. And IBM uses Linux with more than 15,000 business customers. The Firefox Web browser, which has been downloaded onto more than 180 million desktops, has become another poster child for open-source software. More than 800 programmers supply bits of code to the Mozilla Foundation, the nonprofit organization that manages Firefox. Mozilla's small team of engineers then cobbles all those bits together and releases prototypes of new browsers to thousands of testers, including the programmers themselves, who provide feedback and more code to make the browser run well and operate securely.

Engineers like Fried believe that open-source hardware may be on a similar path. "Open-source software took many years to create something 'useful,'" she says, "but [the fact] that it did became the inspiration for duplicating that model [in hardware]."

Von Hippel observes that open-source hardware actually predates open-source software by centuries: people have always shared blueprints and sketches for such things as furniture and machinery. But the visibility of the open-source-software community "has created a new awareness of what has long been the historical practice in hardware," he says.

What's different about today's open hardware is that the Web and new types of design software are making it easier to build, share, distribute, and modify hardware designs. "Most products are designed in software first," says von Hippel. "So you're designing and simulating on the computer, and in the last step you turn it into hardware. If you think of open-source software as an information good, then open-source hardware is also an information good until the very last stage." Hardware designs can be shared and improved and reshared as easily as software designs.

Open-source hardware has a loud and passionate following in the hobbyist community. In 2005, O'Reilly Media began publishing Make magazine, a quarterly how-to guide for all sorts of engineering and science projects. Make now has more than 100,000 subscribers and has spawned events known as Maker Faires, which are a cross between souped-up science fairs and high-tech craft shows. Last spring, 65,000 professionals and amateurs flocked to the San Francisco Bay Area Maker Faire to demonstrate projects that ranged from arts and crafts to engineering and science--and many that blurred the boundaries. And as they showed off their creations, attendees also shared ideas and met potential collaborators.

Around the time that Make was getting off the ground, Eric Wilhelm '99, SM '01, PhD '04, launched the Instructables website, which provides a template for step-by-step instructions that lets people document their engineering projects online. Since its users are allowed to comment on other people's projects, Instructables has created a vibrant community of technology enthusiasts who share information on building just about anything--including a computer mouse made from an actual dead mouse, an eight-foot-long match, and biodiesel fuel. (See Wilhelm's profile as a TR35 winner in the September/October issue of Technology Review.)

Andrew "Bunnie" Huang '97, MEng '97, PhD '02, another MIT entrepreneur, cofounded a company called Chumby to sell soft cubes, the size of a tissue box, equipped with a screen, a simple embedded computer, and a Wi-Fi connection. The Chumby, as it is called, can be used to display data from the Internet, such as Flickr photos or weather forecasts, or to tune in an Internet radio station. From the outset, Huang and his cofounders decided to make the Chumby open source so that customers could modify the basic unit to suit their own purposes. In essence, Huang has created a platform for building other devices. Huang himself has provided instructions for adding a higher-resolution screen, and more whimsically, he's posted instructions for turning two Chumbys into a remote-controlled car and hand-held controller. By making the Chumby hackable, Huang and his collaborators are creating a community of customers invested in helping with ongoing product development.

"Because of the Internet, and because of computing tools, the practice [of sharing hardware designs] has become much more powerful and practical," says von Hippel. In the past, he says, he would have had to mail out circuit diagrams or a drawing of a new robot arm to a couple of people and wait for their responses. Now, he can mock up an idea in a CAD program or one of several software systems that allow researchers to design microprocessors and radios, and blast the plans out to a user group, collaborators, or even visitors to his personal website. The feedback is almost immediate.

Bluespec, a software tool developed at MIT for designing and testing circuits, is one of the advances making life easier for open-source-hardware engineers. Jamey Hicks '87, SM '88, PhD '92, director of the Nokia Research Center in Cambridge, uses Bluespec to collaborate with MIT students to build designs for next-generation cell-phone radios and open-source video decoders--chips that decompress video for display on cell phones and other mobile devices. The beauty of Bluespec, says Hicks, is that you can essentially "program" a circuit design using sets of preëxisting instructions. "For educational purposes, it's useful to have preëxisting designs available just like those that are available in software," he says. "If you have a starting point to build on, you can get more done."

Nokia is interested in open-source hardware because it can move research forward faster, Hicks says. For instance, if video decoder chips are open and widely available, they can be updated more quickly as new video standards, such as high definition, emerge. "Some company might decide to pick up an open design and incorporate it into a [chip] that Nokia could use," says Hicks. He says that Nokia decided to work with MIT on open-source video decoders (among other open-source projects) because it realized that if all chip makers--even those that compete with Nokia's suppliers--had access to decoder designs, Nokia could reduce development time overall and reap the benefits of potential innovation.

It's not surprising that MIT is seeding open-source hardware projects and startups, given its long and storied tradition of hacking. And the Institute demonstrated another truth about the open-source approach when it launched OpenCourseWare in 2002, with the aim of putting nearly all of its class materials on the Web.

By 2007, MIT was offering online access to lecture notes, suggested reading, and in many cases video for more than 1,800 courses. While that may seem like giving away a product worth thousands of dollars, the fact is that having access to class materials can't duplicate the experience of learning on campus. In creating OpenCourseWare, MIT realized that its real value comes from its professors, its students, and its community--all of which are impossible to replicate.

Although technology companies have fundamentally different objectives from those of educational institutions, they, too, are recognizing that embracing openness doesn't amount to giving away the store. If they aim to provide a platform for other designers rather than trying to design the best product for everyone, says von Hippel, they are freed to focus on what large companies are uniquely equipped to do: namely, high-volume production, customer service, and brand marketing.

Of course, for some companies, unlocking the secrets of design just won't work. "A developer [may have] some intellectual property embedded in the design, and their business depends on keeping that closed," Hicks says.

Von Hippel agrees. "If you're going to be open source, you've got to profit from something other than your designs," he says. To shift to open source, some companies might need to change their business plans. That might take time, or it could be prohibitively expensive.

But it should be noted, says Hicks, that hardware combining open- and closed-source components can still benefit from improvements provided by a large community of tinkerers. Some parts of a design can be opened up, while others remain under wraps to maintain intellectual- property rights.

Von Hippel predicts a gradual shift in many industries. Eventually, he says, companies that make such things as high-tech devices and mountain bikes will no longer design their products; instead, the people who use the products will design their own. For example, companies that currently design machines for Boeing might still provide the materials and manufacture the machines, but engineers at Boeing will design the machines themselves, since they best understand how they'll need to use them. "I see open-source solutions increasing their hold in many more fields and being empowering for us all," Von Hippel says. As people who use the products, "we'll get more of what we want and be able to participate in design much more. Manufacturers will go to becoming foundries that produce what users design."

It's as if the rest of the world is awakening to the joy of hacking, he says. "You can say that MIT for many years had the comparative advantages that the rest of the world is currently getting: access to better tools, access to a community of experts. The same thing that inspires the tradition of hacking at MIT can spread, because now a hacker can join from anywhere."