Nanotechnology Now

NN: A significant number of today's "Best and Brightest" postulate the eventual development of self-replicating "nanites" or "nanobots." For the sake of argument, let's say that they are right. Please describe for our readers your concept of nanites, how current scientific understanding may lead us to them, and their up- and downsides.

JRM: The general concept of a nanite or nanobot is a very small machine capable of running a program, moving around on its own, and modifying physical materials at the molecular level. Modifying means building something (assembly) or taking something apart (disassembly)-one atom at a time. In many if not most instances, a nanite will do both-disassembling one object and then using the disassembled atoms to assemble a different object. It will be possible, for instance, to remove the 300 billion tons of carbon we've pumped into the atmosphere (pollution and greenhouse effect reversal), and then use the carbon atoms to build enough homes to house every man, woman, and child on earth-and to put a nice, fat diamond ring (diamonds being made of carbon) on every finger. I believe Eric Drexler was the first to suggest (in chapter eight of Engines of Creation: The Coming Era of Nanotechnology link) pollution reversal through nanotechnology, and that's a considerable upside.

Another tremendous upside is this: once you can alter materials at the molecular level-one atom at a time-you can repair damaged cells, and you can restore the ability of maxed-out cells which have ceased to divide and reproduce to, well, divide and reproduce. No one questions this. Now, humans age and die because their cells become damaged and stop dividing and reproducing. We also die when our cells become traumatized in some other way-heart attack, stroke, AIDS, bullet wound. If we can repair and reverse that damage very quickly-which nanites will give us the capability to do-then we can become physically immortal because any damage (or very nearly any damage) can be repaired. For most people, that's one hell of an upside. I believe Drexler was the first to suggest this (in chapter seven of Engines of Creation) as well.

Downsides? Well, a single out-of-control nanite-much, much smaller than the head of a pin-could destroy the planet in a matter of days. These and a host of other up- and downsides are explored in the Nano novel.

The scientific understanding necessary to set us on the path to nanites already exists. Zyvex (corporate slogan: "assembling tomorrow…") is currently working to create them-and the U.S Army, Air Force, and Navy as well as DARPA and the same national labs which created the atomic and hydrogen bombs are all working to develop nanotech for military uses. From a military standpoint nothing could possibly be more useful than free-roving nanites which seek out and destroy the enemy by disassembling weapons, soldiers, vehicles, transport, supplies and infrastructure in a matter of seconds to minutes? The military has to be working on nanites, whether they want to or not, because other nations certainly will.

As to the specific development path, we seem to be farther down the mechanical (as opposed to the chemical or biological) pathway than any other-which is what Feynman anticipated in the beginning (link).


NN: The Foresight Institute labels these nanites as "artificial replicators" and has created a set of development principles and design guidelines that are "intended to provide a basis for responsible development of molecular nanotechnology." (link & link). If you would please, talk about these principles and guidelines, where you think they are right, and where they fall short.

JRM: I think anyone would have to agree that the Foresight guideline/principles document is a noble effort, and should evolve as we progress. For some reason, the version I'm finding online is nearly four years old. The any-replication-error-randomizes-blueprint touch is really quite brilliant, if workable. Hell, it's a brilliant idea even if it doesn't work. There are, however, a couple of problems which neither Foresight nor anyone else is in much of a position to resolve.

First, there's the issue of enforcement: who's in charge here-and who will watch the watchers? Who's going to tell the United States: "Hey-you can't do that…?" Self-regulation may be a bit optimistic when dealing with something which can destroy the planet if mishandled. Active shields might be one approach; the superswarm option is another.

A second potential problem is the prohibition on independently-functioning subassemblies, which are crucial to superswarm implementation.

Lastly, there are the proposed prohibition on self-replication in open environments, the proposed restriction on self-evolution, and the requirement that replicating nanites be dependent upon one of three things: a) an artificial energy source; b) an artificial vitamin, or; c) a broadcast transmission. All of these seem quite rational at first glance-though such restrictions would make superswarm implementation (as currently envisioned) impossible.

The problem is that wars do not take place in sealed laboratories, and no military establishment is going to pay much attention to these guidelines because following them renders nanoweapons useless. If nanites cannot replicate on the battlefield, they will be less effective than those which can, and become vulnerable to destruction; if they rely upon an artificial vitamin or energy source, their battlefield usefulness is compromised or destroyed, and they will be inferior to those operating with no such hindrance; if they depend upon a broadcast signal, that signal can be duplicated or jammed. Further, the development of such safeguards, even if desired, would slow deployment-for which reason they're not likely to be implemented.

So the military-ours as well those of other nations-is basically going to throw this guidebook out the window. Which is not to say it doesn't have its uses; it does. But the most likely source of a large-scale nanoevent is nanoweaponry-and the institutions developing it are precisely those which are least likely to concern themselves with cumbersome safeguards. They are also those most likely to be conducting research and development activities under the all-concealing cloak of national security.


NN: Please describe a "nanoevent" and some of the ways it may come to pass.

JRM: A nanoevent is an accidental or deliberate release of nonlimited, omnivorous nanites-meaning self-replicating, free-roving nanites with both molecular disassembly and assembly capabilities and without a meaningful limitation on the speed of replication or the number of replications. Basically a little nano-robot capable of tearing apart anything in its environment and using the pieces to make copies of itself, with no speed limit or stop sign. I suppose we could modify that to include heedless nanites with more specific appetites-ones that devour only people, for instance.

Such a nanoevent could happen in any one of several ways. It could occur as the result of an accident-a defectively-assembled nanite or nanite program which doesn't know when to stop. Given the vast number of the things likely to be created, some of them are going to contain errors. Most of those errors will be harmless, but not all. Even so, the likelihood of nanites employed for peaceful purposes accidentally turning out a microscopic menace is pretty slim. Probably.

Weaponized nanites are another story. Here you have something intentionally designed to: a) replicate as swiftly as possible, and; b) destroy everything, or nearly everything, in its path. So the basic malevolence is already contained in the programming code of each individual nanoweapon. It is a much smaller step from this to a programming error which leads to out-of-control replication and destructive swarming. You have a much higher likelihood of a doombot here. (Doombot-the kind of nanite involved in a nanoevent.) This could happen through defective replication, through an unforeseen turn in programmed adaptive battlefield behavior, or through some kind of unintended mutation caused by any number of factors-just as ultraviolet light causes mutations in our own self-replicating skin cells.

And then of course you have your various and sundry deranged individuals who wake up one day and decide, "Today, I'm going to destroy the world." In the past, it wasn't possible to actually do that. Soon, it will be. Terrorists may set out to create a nanoweapon capable of destroying a city or country. They may succeed in doing that-or they may succeed in creating something that doesn't stop at one city, or one country for that matter.

Which of these is most likely, I don't know. Will any of them actually happen? I don't know-nor does anyone else.


NN: Aside from the Superswarm (which we will get to next) what are some of the ways in which an accidental or deliberate release of nonlimited nanites might be dealt with?

JRM: That would depend upon the severity of the event. If you're talking about spilling a can of something with a low replication rate, Bob the Janitor has time to finish lunch and clean up the mess-and by "clean up" I mean destroy the problem nanites, or alter their programming. Program alteration is impractical for several reasons: you're always playing catch-up; you've probably got to immobilize the little buggers to change their programming, and; it's easier to incinerate or disassemble the damned things and crank out a new batch than it is to tinker with the old one.

For small-scale accidents, something similar to an in-place fire-suppression system which sprays antinanites (nanites programmed to seek out and destroy other nanites) into the room should be able to overcome even the most destructive of nanites in short order, because they will both outnumber and surround the bad guys within seconds of their being deployed. Such a system is employed in an early scene in the Nano novel.

The swifter the replication rate becomes, the fewer your options. At the other end of the scale you have nanoweapons-grade nasties, which devour their environment to create copies of themselves, which then do the same, until the environment-in this case the earth-is gone. Which, under ideal circumstances-for the nanites, not for us-could take place in a matter of days, starting with a bullet-sized mass of free-roving nanites. Obviously, Bob no longer has time for lunch; Bob is lunch. This is the climactic scenario in the Nano novel.

To deal successfully with a nanoevent of that severity, you must do two things: hit it fast and hit it hard. Fast because even a slight delay makes any response futile-the nanite "swarm" becomes too large to destroy by any means; hard because even a fast response must destroy the self-replicating capability of every single nanite within that swarm-or you've accomplished nothing, and the surviving nanite instantly replicates the problem you were trying to fix.

So, fast and hard. Fast means minutes, at most; hard means the most destructive capabilities we have. Nanites are machines; like other machines, they stop working when they get too hot. Unfortunately, you're dealing with something that: a) can take a lot more heat than we can, and; b) will devour a city in minutes. So, whatever happens-from the instant that nanoevent begins, the city is doomed; it cannot be saved. That's not a pretty thought, and it goes against our grain-but that's the way it is.

We've never faced or thought about anything like this before; if it's not stopped immediately, it won't be stopped, ever-and the damage won't be confined to a single vaporized city, as might be the case with an accident involving nuclear weaponry. Now, the damage might be confined to one city if, say, it's not an accident at all but a nanoattack targeting a specific city-but you won't know that unless you take a wait-and-see approach, which isn't going to cut it if what you wait to see is something that doesn't stop when the city is gone. So, again, the city is doomed.

Basically, the only way you're going to stop a nanoevent like that is by vaporizing the city with one or more nuclear weapons. Further, you don't have time to think about it. You may not even have time to do it.

Let me explain that. First, the event must be detected; second, it must be evaluated; third, that information must be relayed to someone with the authority to launch; fourth, that person must decide to launch; fifth, that person must approve and order the launch; sixth, the order must be transmitted to the launch facility; seventh, the order must be confirmed; eighth, the weapon must be targeted; ninth, the weapon must be launched. After that you have flight-time, which could be as much as thirty minutes. And, of course, if the response takes 30 minutes from detection to detonation-it's probably not going to work because the swarm will by that time be too large to destroy. Remember, if one single nanite avoids the heat and, say, rides the blast wave to a new location-the whole situation begins anew. If ten nanites are blown outward in different directions, you have ten times the problem you had half an hour before.

Now, that's a very simplified version of what could happen. Potential complications are many. Say, for example, that the nanoevent is first detected by the United States-but is taking place within the borders of another country. The President literally does not have the time to pick up the phone and try to reach that nation's government officials; to end the event, he must launch, now. The target country's government may prove less than understanding, to say the least. One nation might even covertly create a nanoevent within the borders of another in order to justify a nuclear strike by itself or cause such a strike by a third nation--knowing that any evidence will be vaporized. Why would a country which already had nanoweapons bother with such a ruse? Perhaps to set the other two nations at war. Or-try this-it's not a nation with nanoweapons but a terrorist group with neither nukes nor nanoweapons, and the "nanoevent" is not in fact real, but a computer simulation the monitoring nation's computers have been hacked into displaying as real. The decision-makers do not have time to confirm.

Even without these complications, response time makes success a very iffy proposition. Short of having a ready-to-blow nuke on every street corner-which presents its own problems-such a scheme is, sooner or later, bound to fail. Other response possibilities include the use of plasma or antimatter weapons, or nanite warheads, but no one sees either of the first two options materializing any time soon, and the delivery problem remains with all three--the nanites in particular, because once on-target they have the improbable task of catching and overtaking the bad swarm's replication rate--which isn't going to happen..


NN: Please describe for our readers your Superswarm option, why you devised it, and how it is better than other options?

JRM: The idea came to me after reading K. Eric Drexler's Engines of Creation. In chapter eleven, Engines of Destruction, he proposes a solution to weapons of mass destruction which he calls "active shields." One permutation of this would be orbital weapons platforms. In any event, the shields as proposed would be controlled by an independent, artificial intelligence. On no one's side, with weapons superior to all others (particle beams, for example), these shields would monitor the globe for signs of hostile activity (a missile launch, say) and respond by destroying both the weapon (missile in flight) and possibly the aggressor (the launch facility). My recollection is that it wasn't really proposed as a way to deal with nanoevents; the idea was to have something capable of imposing peace by force, with an AI in charge.

I don't know about you, but I'm not terribly comfortable with the idea of a nonhuman intelligence in sole control of the most fearsome weapons in existence, nor do I think it wise. I came to think of it as the Klaatu Solution-analogous to Gort, the alien Klaatu's invincible peacekeeping robot in The Day the Earth Stood Still (link). I didn't trust Gort, and I wouldn't trust this. I keep thinking of James Cameron's Terminator (link).

Discounting active shields didn't really get me anywhere, though, because the anger management problem they'd been proposed to address wasn't going away any time soon. I tried to think of some alternative-one that could also stop a nanoevent in its tracks, regardless of where or when it occurred. Something which could, in short, "maintain unending, omnipresent vigilance over the whole world," in Drexler's words.

The most obvious problem with a nanite swarm is that, once it reaches a critical threshold, it has the capability to expand faster than any possible countermeasure; once the swarm's expansion rate passes point x, it can't be stopped. Further, the likelihood of getting the countermeasure on-target before point x is reached is not good. It occurred to me that the only way to be absolutely certain the countermeasure reaches the target in time-is to make sure it's already there, before the event begins. And the only way to do that is to create a single swarm which is everywhere at once-a superswarm.

Now you've turned the tables on any possible nanoevent. Instead of playing catch-up with an expanding swarm and breaking your neck to deliver a nuke before the thing expands beyond control-in the process vaporizing what you're trying to save in the first place-you have the home-field advantage. Regardless of its point of origin, any malevolent swarm now begins in a hopeless position: surrounded on all sides by a vastly larger swarm hostile to its very existence and capable of instantaneously replicating on an almost infinitely larger scale and attacking while simultaneously consuming the environmental resources needed by the bad guys to expand their numbers. Under these conditions, the malevolent swarm cannot expand; it cannot even survive.

This concept can be expanded (so to speak) to include detection and elimination of nuclear and other threats as well. The superswarm's sole mission (as currently envisioned) is the detection and elimination of large-scale threats, though it could of course be taken down to the micromanagement (nanomanagement?) level of disassembling dirty magazines in the hands of the underaged.

It did not occur to me until a recent rereading (this morning, in fact) of the relevant portions of EOC that the superswarm might actually be considered a different type of active shield, without the centralized control structure. So I guess that brings me full-circle on the concept.


How and when should the Superswarm be deployed, and what are its limitations?

JRM: Obviously the superswarm can't be deployed until after we've developed a fairly comprehensive nanotechnology-and that's the primary limitation: even if we do choose to implement the superswarm, we need stopgap measures to get us from here to there. The other limitation is the fact that, once implemented (as presently conceived), there is no limitation placed upon the swarm; in order to perform its intended function, it must be autonomous. Because of that, if, in spite of our best safeguards, something, somewhere, somehow goes terribly wrong in a big way-this thing can't be stopped, and that will be (to steal a chapter title from Nano) the end of all things. The decentralized control structure I've proposed should prevent that from happening by enabling the swarm to attack and destroy any portion of itself which turns rogue. One of the reasons I've proposed the superswarm publicly is to get feedback from others who may come up with things I haven't thought of-and many thanks to you and to Nanotechnology Now for helping to spread the word.


NN: What are the disadvantages of the Superswarm option?

JRM: If the thing gets glitchy, we could all be gone in seconds-the biblical twinkling of an eye, as it were. I am by no means certain the superswarm option should be implemented; I simply present the concept for discussion. If that results in useful suggestions for improvement, which is my hope, I plan to post the best portions on the superswarm section of my website at www.johnrobertmarlow.com. (Comments can be sent via the NN Contact Form, and will be posted to this page. Private comments may also be sent to superswarm@johnrobertmarlow.com. See more comments at sci.nanotech and Responsible Nanotechnology)


NN: In your opinion, what is the most likely scenario that leads to the survival of humanity once all the nano-pieces are in place and artificial replicators are no longer techno-fiction?

JRM: That's an interesting question because "the most likely scenario" and the "scenario that leads to the survival of humanity" may well be different things. As I say in the author's afterword to Nano, the most likely scenario is probably a nanoarms race which leads to extinction. Consider: from the moment it was realized that nuclear weapons were possible, their creation became inevitable because no major player could risk entering the future without them. It's a situation that drives all to seek the creation of something which none really want.

Today, global economics is a war all its own, and when you add the incomparable commercial benefits of nanotech to the unassailable military superiority conveyed by nanoweaponry-the only possible result is a global race to the finish line. Which, unfortunately, may really be the finish-for all of us. (See "Marlow's Second Paradox" in the side-bar)

The scenario with the best chance of leading to our survival would be one in which we all behave rationally, peacefully, and cautiously. You can see why I hesitate to call this a likely scenario. Nonetheless, the very nature of this particular beast presents another, intriguing possibility, and one which the lead character in Nano pursues: regardless of the number of players at the nanotable, Mankind's future can be assured if just one player does this right-and that player wins. Again owing to the nature of this technology, that player need not be a nation or even a large organization; it could be an individual.

The founding fathers of the United States themselves believed that the best possible form of government is benevolent dictatorship; the problem with that is that you cannot guarantee an endless succession of benevolent dictators-which is why they developed the constitution. As explained in the Nano novel (and originally proposed by Hazel Henderson in Creating Alternative Futures: The End of Economics), complex technologies are inherently totalitarian because most people do not understand them and are therefore not competent to oversee their use. As technologies become more complex, the number of people who understand them fully-and who will therefore control them, and through them the societies which cannot survive without those technologies-diminishes. In this way, complex technologies drive us toward totalitarian technocracies.

If you extend that, the end result may well be, for better or for worse, a single individual with truly absolute power, and the ability to retain that power-forever. If that winds up being a good, fair, just, compassionate and freakishly intelligent person, then we'll have a scenario likely to lead to our survival. If not, we'll have Bill Joy's "further perfection of extreme evil, an evil which spreads well beyond that which weapons of mass destruction bequeathed to the nation-states, on to a surprising and terrible empowerment of extreme individuals." (link)

Can we survive in a world where a national government-any national government-is possessed of absolute power? Would we want to? Can we survive in a world of multiple nanopower-nations? I don't know. The first war could be the last. Which scenario will come to pass-if any?

We'd better start thinking about it.