Smart Dust and Gray Goo

This month, I touch on some of those incredible opportunities and horrific dangers, particularly in the gray goo/smart dust realm, as seen by Bill Joy, Eric Drexler, John Seely Brown, Kris Pister, Michael Sailor, David Gelernter, Neal Stephenson, Bruce Sterling, and the dog with the chip in his neck.

Gray Goo

The term "gray goo problem" was already in common use among the nanotech cognoscenti in 1986 when K. Eric Drexler wrote the seminal book on nanotechnology, Engines of Creation. If we follow the nanotechnological imperative uncritically, Drexler warned:

'Plants' with 'leaves' no more efficient than today's solar cells could out-compete real plants, crowding the biosphere with inedible foliage. Tough, omnivorous 'bacteria' could out-compete real bacteria: They could spread like blowing pollen, replicate swiftly, and reduce the biosphere to dust in a matter of days.

Drexler's vision is as horrific as the dystopian science-fiction disaster scenarios of Kurt Vonnegut (Ice-9) or Frank Herbert (White Plague).

The gray goo problem involves the accidental release of voracious self-replicating nanomachines into the environment, a scenario that is easy to envision given the fact that exactly that kind of accident has occurred with genetically modified seeds, although not as yet with any globally disastrous effect. But within the past year, Christine Peterson, the president of Drexler's Nanotech think tank (the Foresight Institute), told the U.S. House of Representatives Committee on Science about three ways in which nanotechnology could be deliberately used for bad purposes:

• Very rapid construction of conventional weapons, making traditional arms control more difficult.

• Totalitarian control of civilian populations by surveillance using nanoscale sensors.

• New weapons made possible by the technology, which can be thought of as "smart" chemical weapons.

Peterson wasn't writing off the accidental gray goo problem, either. "Molecular machine systems able to build complex objects," she said, "could build copies of themselves, possibly overdoing this activity from a human point of view, as bacteria do." I imagine that reducing the biosphere to dust in a matter of days would be an example of this "overdoing."

Fears of Joy

But maybe we're the ones who are overdoing it if we worry about gray goo now. "[I]n fearing that nanodevices will run amok," say John Seely Brown and Paul Duguid in Peter Denning's The Invisible Future, "we are in danger of getting far ahead of ourselves."

Brown and Duguid point out that the molecular manufacturing plants envisioned by Drexler are still on the drawing board—if even that far along. Ralph Merkle's pioneering nanotech work may have shown that Drexler's dream is theoretically feasible, but theoretically feasible and practically feasible are two quite different things. As yet, nobody has shown how you get from here to there, and until that road gets built, there is no danger that anyone's going to drive it too fast or in the wrong direction because there is simply no road to drive on. They acknowledge that it's not too soon to think about the risks, but they reassure their readers that the Foresight Institute is working on that.

So what's Bill Joy worrying us for?

Around 1999, Bill Joy heard that nanoscale molecular electronics was now practical. That's true, although it's not the same thing as "true" nanotech—the molecular engineering of self-replicating devices. Still, it's another brick in the wall. The news sent him back to Engines of Creation, which he had read avidly years earlier.

"Rereading Drexler's work after more than 10 years, [I was] struck by how naive some of Drexler's safeguard proposals seem, and how much greater I judge the dangers to be now than even he seemed to then." Naive? Well, one of Drexler proposals was that we build a nanotechnological shield—a Star Wars for nanotech dangers. Maybe you'd agree that that sounds a little naive. Joy points out that such a shield would itself be a great danger—"nothing could prevent it from developing autoimmune problems and attacking the biosphere itself."

Okay, that's another view. Is Joy right, or are Brown and Duguid?

Ecophagy

If anyone has the answer, it's the president of Drexler's Foresight Institute, which is indeed studying the dangers of nanotech.

Peterson says that first you have to decide which nanotech you're talking about. Near-term nanotech, which means any technology involving artifacts smaller than a cubic micron, can indeed have environmental impact, but it's not going to lead to any gray goo scenario. Advanced nanotech, also called "molecular engineering," is a far more transforming technology, and it does raise the specter of molecular machine systems building copies of themselves, and "overdoing" it, with catastrophic consequences.

Robert A. Freitas, Jr. has run the numbers on various versions of the gray goo problem—which he calls "ecophagy"—such as gray plankton (oceanic goo), gray dust (airborne replicators), and gray lichens (they grow on rocks), and offers some reassurance that at least some versions of gray goo are thermodynamically implausible and that most of the threats could be dealt with given a vigilant Goo Patrol constantly watching for signs of the imminent destruction of all life on earth.

But he considers the threats so serious that there ought to be an immediate moratorium on certain kinds of nanotech work. Well, maybe. But Peterson argues that moratoriums won't work, giving the "if guns are outlawed, only outlaws will have guns" argument. I don't know about you, but I am not reassured.

The Diamond Age

Gray goo is a somewhat vague threat. Both the incredible opportunities and the horrific dangers of nanotechnology can be made clearer by sketching some scenarios. In his 1995 science-fiction novel The Diamond Age, Neal Stephenson did just that.

Stephenson gives us a world populated by people who are used to the idea of living in an environment "enriched" by clouds of active nanobots, doing their bottish duties. Bots in their air, their hair, their food, their bodies. Huge matter compilers, meanwhile, suck matter out of the air and water and compile it into consumer goods. There are still consumers in Stephenson's world, but I guess you can't change everything in one novel.

Since then, some folks have built some of the diamond age (and some other folks are on the way to making diamonds as cheap as glass, but that's another story). For example, the smart dust folks.

Smart Dust

Dust is tiny and pervasive. HP Labs researcher Philip Kuekes says that molecule-sized computers will one day be "as pervasive and powerful, and as cheap, as life itself."

Michael Sailor, a chemist at the University of California, San Diego, has developed what he considers an early version of these tiny computers. His smart dust devices are about the size of the diameter of a human hair and have limited sensing and signaling ability. They don't replicate themselves, but they can spontaneously assemble into a large, more visible structure.

Calling these slivers of silicon "computers" is a bit of an exaggeration, but "robots" works. They are specialized devices that can be used to identify individual drops of oil in water and signal their location, possibly even walling them off from the water. The same technology could, Sailor says, be used to detect toxins in the air or water or even locate and destroy tumor cells in the body.

The term "smart dust" was coined, though, by University of California, Berkeley, EE professor Kris Pister. Pister's work is on a similar track. He wants to develop networked airborne motes of silicon that can sense, measure, and transmit data about their environment.

Some smart dust products are in use today, although they aren't airborne and are nowhere near the size Pister has in mind: An Australian company uses smart dust to detect hot spots on train wheels and identify wheel bearings that may need replacing or repair. These devices are about a centimeter in diameter, but the technology is young. They'll get smaller, although Pister's ultimate target is devices of a decidedly nonnano cubic millimeter.

Part of what keeps Pister's motes so large is the need to communicate. Even tiny radios need nontrivial power to pump out their signals. But what if you reduce the transmission range, one of Pister's grad students asked, and increase the number of motes? If you can count on finding another mote within 30 feet, you can get the solar cell down to a few millimeters. At that scale, you can also pack in an optical receiver, an actively orientable optical transmitter, and the signal-processing and control circuitry.

But what if you go nuclear, Pister now wonders? Add a trace of a radioactive isotope and use that for power, and you can reduce the size further.

Dust to Dust

To me, the payoff in smart dust comes when it's 1. ubiquitous, and 2. networked. (They'll probably have to find space in those motes for authentication, spam blocking, and tempestizing technology, too.) I don't know what data rate can be achieved in such a network, but a dust-to-dust network implemented on a dust-thick universal blanket of motes might provide an alternative to the cable and satellite franchises, as one scenario. Another scenario is that it becomes intelligent. That would be interesting.

Sahara in Soho

A small point: It may be unimportant, but it seems to me that, as is the case with nanotechnology, different things are being called by the same name here. Sand is not dust. Actual dust can be carried great distances by the wind: Dust from the Sahara desert often makes its way to England. "Robots the size of a grain of sand" can be blown or bounced along the ground for hundreds of kilometers, but sand doesn't travel anything like as freely as a dust mote. Or a pollen grain.

The Dog with the Chip in his Neck

Assuming we've got these motes, or nanobots, or whatever they are, all communicating with one another. What do they have to say?

In his 1995 book, The Dog with the Chip in His Neck, a collection of his NPR "All Things Considered" essays and other writings, Andrei Codrescu says that, at first, he thought the dog with the chip in his neck was Cerberus, the hound who guards the gates of hell. "Imagine my surprise," he says, "when I found Hades vacant, its pits full of blah-blah, no one suffering."

The dog, actually named Zena, certainly wasn't suffering. The chip was injected by a vet and paid for by Sherri Wigdore, both of whom had Zena's best interest at heart. The chip was a short-range transmitter, which beamed the friendly message, "I'm Zena, I'm here, and I belong to Sherri," more or less, to anyone who would or could listen. Get a lost Zena near a dog-neck-chip-signal reader, and Zena is no longer lost. That's smart tagging.

And that was last decade (century, millennium). Today, millions of dogs, I suppose, have such chips in their necks. Now smart RFID tagging is poised to invade every retail outlet, populating shelves with merchandise that broadcasts, "I'm razor blade RB111914500, I'm here, and I'm as yet unpaid for."

If smart tags can give dogs and razor blades a voice and an identity, a simulacrum in cyberspace, a node in the noosphere, why not other things? Why not every thing?

I'm a Fork, I'm Here, and I'm Okay

Science-fiction writer Bruce Sterling has a word for what that might look like. He calls it "ubicomp," which is ubiquitous computing after it has sunk deeply into the structure of future daily life.

Computer chips inside everything. Smart forks. That's what we need, Sterling thinks, smart forks. Okay, I'll bite: Why does a fork need a computer in it? Sterling thinks that our terminology is confusing here. A fork doesn't need to be smart, really, but it could benefit from having just a little bit of intelligence inserted. Imagine if it could tell us just three forkish things: "I'm a fork, I'm right here, and I'm okay—not rusty, broken, or stolen." Or maybe "I'm dessert fork DF109223081, my purchase data is at this url, here's a link to a blog I like, I'm at these coordinates, and I'm okay."

On one level, Sterling is just talking about something like those "smart" RFID tags. On another level, he's imagining what it would mean to empower objects with their own identity. Turning objects into Objects. One radical consequence: Stealing becomes impossible if all property can broadcast its identity and location. This, in turn, has some implications. Can capitalism survive without theft? I mean that as a serious question, and one that we might see answered in our lifetimes.

A world of artifacts that have OOP knowledge of their identities, spatial coordinates, and physical or working state could simplify many of life's labors. Rather than the manufacturer and shipper having to set up elaborate package-tracking systems, following the movements of your purchase as it crosses the country could be a simple matter of Googling it. And you'd know immediately whether it was damaged when it fell off the loading dock in Denver, and could reject the shipment. (It would be nice if you could just message the product to go back home—or to notify its vendor and report to the nearest recycling center.) And exactly the same locating technique would find it for you a year later when you've misplaced it in that mess in the garage—except that you would no longer use the word "misplace," and you would think of the mess as an idiosyncratic arrangement expressive of your uniqueness.

New Paradigms for Reality

I started out talking about the gray goo and smart dust of nanotechnology and segued into the future of smart tags, but I do think that the futures of these technologies are, or could be, convergent. So does Sterling. He lists some of the names used to try to get a handle on the part of it he's interested in: pervasive computing, intelligent environment, ubiquitous computing, things that think, locator tags, JINI, personal area networking, etc. He thinks that these overlapping concepts are like the rumblings of a tectonic fault. "The signs are very good that something large, expensive, and important will tear loose there." When and if it does, it probably won't be exactly like anything anyone has imagined.

For example, many programmers have decided that object-oriented programming isn't everything it promised to be. They're looking to other paradigms that take the virtues of class-based programming to a new level of usefulness. So does it make sense to follow Sterling and tag physical objects so that they become Objects? Or nodes in some XML-connected hyperspace? Are there other programming models that might be more powerful enablers of objects? What about, to pick a paradigm at random, aspect-oriented programming as a plan for the augmentation of reality?

Hey, I was just asking.

Sources

Brown, John Seely and Paul Duguid. "Don't Count Society Out," in Peter Denning's The Invisible Future (McGraw-Hill, 2002; ISBN 0-07-138224-0).

Codrescu, Andrei. The Dog with the Chip in his Neck (St. Martin's Press, 1996; ISBN 0-312-14316-8).

Drexler, K. Eric. Engines of Creation (Anchor, 1986; ISBN 0-385-19973-2).

Freitas Jr., Robert A. "The Gray Goo Problem," KurzweilAI.net.

Ghirelli, Michael and Edward Davies. "The Last Word," New Scientist Magazine (September 27-October 3, 2003).

Joy, Bill. "Why the Future Doesn't Need Us" (Wired, April 2000; http://www.wired.com/wired/archive/8.04/joy.html).

Peterson, Christine. "Molecular Manufacturing: Societal Implications of Advanced Nanotechnology," U.S. House of Representatives Committee on Science (April 9, 2003; http://www.house.gov/ science/hearings/full03/apr09/peterson.htm).

Stephenson, Neal. The Diamond Age (Bantam, 1995; ISBN 0-553-57331-4).

Sterling, Bruce. "When Our Environments Become Really Smart," in Peter Denning's The Invisible Future (McGraw-Hill, 2002; ISBN 0-07-138224-0).

DDJ