EE Times: The subtitle to your new book, Fab, is The Coming Revolution on Your Desktop — from Personal Computers to Personal Fabrication. What is personal fabrication, and when will we see it?

Neil Gershenfeld: Let me step back to answer that. We have had a digital revolution in communications and a digital revolution in computation. Your magazine and the careers of your readers are largely based around that, but what does it really mean to be digital?

The answer dates back first to [the demonstration by Claud] Shannon in the '40s — we are used to the idea now, but it was shocking and revolutionary at the moment — that if you add information to a signal, you can remove it later without error. In the '50s, [John] von Neumann did the same thing for the computer, showing that you could build a reliable system from imperfect parts. What that brought was the idea of scale, so that we now have a globe-spanning Internet and single-chip processors that are approaching a billion transistors.

EET: And advanced fabs for building them.

Gershenfeld: The $10 billion chip fab. It's the highest of high-tech. But they still use a process that artisans have been using for millennia: You spread some materials around and bake them.

Compare that with how your body works. There is a protein that builds your body: the ribosome. It's a molecular machine that decodes messenger RNA and assembles amino acids to build peptides. And it does it in a fundamentally digital way, in exactly the sense of Shannon and von Neumann. It reads a redundant code, it assembles from a discrete set of components and it can correct errors.

The point of all this is that we have digital communications and digital computation, but we still live in an era of analog fabrication.

EET: Are we about to see digital fabrication systems?

Gershenfeld: At MIT, I direct a program called the Center for Bits and Atoms, where we are working at the boundary between physical science and computer science. One of the big things we are working on is the idea of digital fabrication: additive assembly of computers that don't just control tools but that are tools. Ultimately, what that promises once again is scale: building perfect macroscopic things out of imperfect microscopic parts. That's the deep idea: digital fabrication.

EET: Will these experiments eventually scale up to the $10 billion digital fab?

Gershenfeld: The deep idea of digital computation appeared as mainframes, passed through minicomputers and reached everyone in the form of personal computers. And the killer application was the personalization of computation — not inventory and payroll, but computers solving the needs of individuals, rather than just whole organizations. Likewise, digital fabrication will reach individuals in the form of personal fabrication, of machines that make machines.

EET: People are working on concepts of nanofabrication or molecular-scale factories, but isn't that way off in the future?

Gershenfeld: The real surprise is just how close we are to it. Even in labs at places like CBA [the Center for Bits and Atoms at MIT], we are working on universal molecular assemblers — but that is a 20-year road map, for that kind of science-fiction output device. I bought millions of dollars' worth of tools to do that research — focused-ion-beam writers, supersonic water-jet cutters, excimer lasers. These were just the tools to do the research.

Then I was spending a lot of time training students to use them all. So I started a class, "How to Make Almost Anything," which was just that — it was how to use the tools to make almost anything. But I was completely swamped with nontechnical students, who were desperate to take the class — not for research, not as a business model, but just because they had stuff they wanted to make.

EET: Do you think people in general will be willing to get involved in manufacturing and other very technical areas just to satisfy some whim?

Gershenfeld: The passionate response led me to wonder what would happen if the rest of the world gets access to this. So we started setting up, with National Science Foundation support, field Fab Labs, where the idea was to approximate both what was on campus at MIT and where we are going to be 20 years from now, but using tools available today. And [the program] exploded all around the world in both developed and, most interestingly, developing countries. We found the same response in the field as we found at MIT.

These labs were not meant to be all that useful — this was supposed to be a warmup experiment — but at this point we have labs in India, above the Arctic Circle in Norway, in Ghana, Costa Rica, inner-city Boston.

This summer, we are going to South Africa, and we have demands to take these all over the world — more than we can handle.

EET: When it comes to the social or economic implications of this, let's suppose that, like the personal computer, these labs just appear everywhere. Won't that be quite disruptive to the economy of the developed world?

Gershenfeld: Absolutely. Some aspects you can predict, and some you can't. There is a tremendous parallel between where we are now with personal fabrication and the PDP [minicomputer]-to-PC transition. Industry didn't take that transition seriously — [PCs] were considered just toys, not real things — until, generally, it was much too late. So in the same sense, I suspect that the extent to which you can create advanced technology with kids in an inner-city community center just sounds ridiculous to serious industry, and they would look at that as a nice toy but not a serious threat. But I really think it is.

EET: So what is the new business model for personal fabrication?

Gershenfeld: One is the world of open-source software, where ownership of code is no longer a protectable asset. You can build businesses, but the businesses are closer to a service industry [model], based on adding value in designs and solutions, not on owning a scarce asset. The second piece is microfinance, which [is already revolutionizing] global economics by making large numbers of tiny net loans, not large ones.

The thing I think will emerge that doesn't exist yet is micro VC. Personal fabrication leads to the opportunity for high-tech innovation, but on the scale of tens of thousands — not tens of millions — of dollars. You'll need some of the skills of a good venture capitalist, but with the fanout of a microfinance network.

EET: Will we still need the chips that are produced by the ever-increasing speed and circuit density of VLSI technology?

Gershenfeld: We are working with the idea of paintable computing. The idea was driven by Bill Butera, who used to be a chip designer and thought it was nuts to just go on building bigger and bigger wafers; he wanted to make the smallest affordable chips and then "pour out" computing by the pound or the square inch.

So, we have been working on that — and Bill is now at Intel — not just as a way to do a fab, but also, crucially, as a programming model.

EET: It's difficult to imagine how you could build a system with that approach.

Gershenfeld: An example application is a display. There are lots of fabs working on bigger and bigger displays. What we have been working on and have prototyped at the pc board level — and we are moving it rapidly to silicon — is a display that is statistical, where you just spread these nodes out, they communicate locally, there is no frame buffer and no I/O processor, but each one has an emissive element. And we did a very very simple subset of Postscript interpretation.

The idea is that you can send graphics into the medium and it gets turned into a display. But compared with a current display, it is statistical, meaning you add as much of this material as you need, and if you add more, the display gets bigger. Instead of using the display just as a display separate from the computing, the display itself becomes a computational medium.

EET: So how does this bulk circuit fabrication mesh with personal fabrication?

Gershenfeld: Ultimately, our interest is in actuation so these little nodes can move. And then once the nodes can change their physical configuration, these two threads merge.

EET: Are you talking about self-assembling systems?

Gershenfeld: The crucial distinction is not self-assembly but programmed assembly. You need embedded computing to program the assembly.

EET: Does this future technology mean that everyone will eventually become an engineer?

Gershenfeld: The thought I would like to leave people with is this notion of the complement set to the readers of EE Times — the rest of the world becoming electrical engineers, both consumers of knowledge and consumers of parts. [Consider] the applicability of that to the rest of the planet.

Neil Gershenfeld

Education
PhD, applied physics,
Cornell University

Current Position
Director of MIT's Center for Bits and Atoms, an interdisciplinary effort exploring how the content of information relates to its physical representation

Published Work
In addition to journal articles and patents, four books:

• Fab
• When Things Start To Think
• The Nature of Mathematical Modeling
• The Physics of Information Technology

Accomplishments

• Developed a form of molecular logic used to implement the first complete quantum computation
• Devised an analog circuit that efficiently performs optimal digital tasks
• Collaborated with nontechnical people to develop electronically functional systems, such as a computerized cello for Yo-Yo Ma
• Invented scaled-down fabs to provide functional systems at low cost for local needs around the world