computerwoche.de

Archiv

Bits to atoms (and atoms to bits)

03.04.2006
Neil Gershenfeld says we're on the threshold of the third digital revolution, one in which matter and information merge. He has kicked off the revolution in a fabrication laboratory, dubbed the Fab Lab, at MIT, where he's the director of the Center for Bits and Atoms (CBA). In the Fab Lab, students design and manufacture their own products using inexpensive fabrication and electronics tools driven by open-source software and programs written by MIT researchers. Other "fab labs" based on these principles are spreading around the world, especially in less-developed countries. Gershenfeld, who has laid out the precepts of personal fabrication in his book Fab: The Coming Revolution on Your Desktop -- From Personal Computers to Personal Fabrication (The Perseus Books Group, 2005), recently explained the mission of the CBA to Computerworld's Gary Anthes.

What do you mean when you say we are entering the third digital revolution? Historically, we have had two very distinct and important phases: communications and computing. What I'm referring to is fabrication, which is still on the analog side. The real state of the art in fabrication is in the body, in the ribosome. It's essentially a molecular computer; it runs a program. It doesn't control the tool -- it is the tool. And the output isn't arranging bits -- it's arranging atoms. But it has all the properties that Claude Shannon and John von Neumann [defined] for communications and computation.

One of the CBA's "grand challenges" is to create "it from bit." What does that mean? The research we are doing is looking at how you go, quite literally, from bits to atoms and from atoms to bits. If you have a description, how do you turn it into a thing, and if you have a thing, how do you turn it into a description? What are emerging are principles for how to do exactly that.

Is this a new branch of computer science? In many ways, computer science is one of the worst things to happen to either computing or science. The canon of computer science that's currently taught prematurely froze a model of computation based on 1950s technology. Nature is a much more powerful computer than traditional models of computation consider. One of the dramatic examples is quantum computing, but there are many other ways nature can compute that are poorly captured in conventional models of computation.

Can you give an example? One of the first projects we've done is Internet 0. It lets you build a Web server for US$1 that can go into a light switch. It takes the original properties of the Internet -- internetworking and the end-to-end principle -- and extends them down to the physical device level. It will let you do IP to everything, at essentially the cost of an RFID tag. It's the first step in breaking computation out of the boxes you see today and integrating it with the physical world.

What's another example? We are developing fungible computation -- computation as a raw material that can be poured, sprayed or unrolled, that can be applied where you want it in the quantities you need. For example, you have a display and you need a little more screen space, or you have a server and you run out of resources. Today, you can add another display or another server, but that's about the granularity that's possible. So the research is looking at how you can make millimeter- or submillimeter-size [computers] and put them in various form factors, such as paint or wallpaper, and then build programming models so the little devices organize locally and globally. So that display becomes wallpaper you unroll, and if you want more display, you add more wallpaper. If your server needs more resources, you open the top and pour in more server. We are pushing the frontiers of fabrication, process integration, packaging, communications and, most importantly, programming models.