Why I hate “nanobots”

by Eric Drexler on 2009/03/07

Honda “Asimo” anthropomorphic robot
Honda Robot

Yamaha robotic arm
Yamaha Robot

An elevator is now a “robot”, too
Otis Elevator Robot

For background:

  • I disliked the word “nanobot” the first time I heard it.
  • I like it less now.
  • I don’t know what it means.

Once, “robot” meant a machine that resembled a human being. In fiction, robots usually were intelligent and capable of plot-advancing independent action.

Later, “robot” also meant a machine with an arm that could do certain kinds of industrial work like a human being. Industrial robots are programmable, but most aren’t even responsive, much less intelligent.

Today it seems that a “robot” is any machine that has a moving part and is considered cool for one reason or another. An elevator mechanism is called a robot, provided that it’s for a space elevator, or a model of one.

By this standard, I suppose anything that is both “nano” and “a machine” can be called a robot, and a tiny robot is widely viewed as a magical thing, likely to be intelligent, fertile, hungry, and grouchy (or something). Never mind that researchers haven’t gotten large robots — real, mobile devices with sensors and grippers and stuff — to exhibit any impressive degree of intelligence, and never mind that making machines smaller doesn’t tend to make them smarter. Cool machines are robots, nanomachines are cool, “nanobot” sounds twice as cool, and there you are: a source of endless and destructive confusion.


It was once alleged (by a senior scientist who should have known better) that manufacturing systems that process molecular building blocks must inevitably be tiny, swarming, intelligent, socializing, conniving things that would be very, very bad if they weren’t impossible.

Real concepts for nanoscale manufacturing systems (of the advanced sort, very different from the accessible and attractive near-term objectives) have much in common with conventional automated manufacturing systems.
In recent posts, I’ve discussed how automated, high-throughput manufacturing system works today, and it doesn’t use robots to make and assemble small parts. Each post includes videos that show where common things in our lives come from, and the several show simple machines that make and assemble things with blurring speed:

And finally, if you haven’t seen it already, you might enjoy the nanofactory animation video.

Scott Jensen March 7, 2009 at 8:32 pm UTC

How do you like the word “nanite” then?

Eric Drexler March 8, 2009 at 9:42 pm UTC

@ Scott Jensen — The connotations of “nanite” are even worse, if the term is thought to describe “something about nanotechnology”. The term usually seems mean tiny, smart, swarming, proliferating, imaginary nanobugs. For comparison, imagine that people weren’t familiar with the technologies we use today, and had a vague idea that washing machines, cars, can openers, and laptops would be like science-fiction style robots — smart things that run around, do work, and are apt to run amuck. And that all this was inherent in the very nature of machinery and electronics. A pretty crazy misconception.

Will Ware March 9, 2009 at 8:16 pm UTC

How sadly true! During my years as an electrical engineer, the question of how to design electronics that would reliably run dangerously amuck was a subject of much discussion and many conference papers. Toys like Chuckie don’t simply happen. There is a lot of hard work behind those “Evil Inside” stickers.

nanotürkiye March 10, 2009 at 4:41 am UTC

Nanobots are solutions to all problems in the world! I think this analogy problem arosed, because people thought that they can explain nanotechnology better to people with this concept.

Scott Jensen March 10, 2009 at 5:01 pm UTC

Then, Eric, what would you call them?

Eric Drexler March 10, 2009 at 7:58 pm UTC

@ Will — Those would be conferences run by IEEE’s Evil Systems special interest group? ;^)

BTW, I’ve given some thought to why people imagine that super-sophisticated bad things might spring into existence accidentally, as if by magic.

One reason is that many accidents are bad, and many bad results are accidental, making it easy for people to wrongly equate “bad” with “accidental”.

Another is that anything active in the world, if not understood, tends to arouse a sort of primitive, animistic interpretation. People once attributed intelligence to wind, water, and trees, and it’s easy to see how, in confused minds, small machines could become capricious spirits.

The most basic reason, of course, is that it’s easy to attribute absurd properties to things that don’t yet exist.

Eric Drexler March 10, 2009 at 7:59 pm UTC

@ nanotürkiye — Advanced nanotechnologies can solve may problems and will create others, but what most people imagine when they hear the word “nanobot” is a vague fantasy that has interfered with discussing real engineering prospects. Practically speaking, one might as well promise to conjure “invisible spirits”. Enthusiasm based on this is, unfortunately, worse than useless.

Eric Drexler March 10, 2009 at 8:07 pm UTC

@ Scott Jensen — Before asking what to call them, we need to ask “What are we trying to name?” For an analogous question, first think of all of the refineries, smelters, chemical plants, steel mills, plastic molding machines, milling machines, stamps, presses, factory systems, automobiles, cell phones, air conditioners, solar arrays, and so on, then ask “What would you call them?” Modern technology, I suppose. Certainly not “macrobots”.

What to call a broad set of technologies based on new means of production is a difficult question. The best answer I know of is “advanced nanotechnology”, noting that this will include many advanced nanotechnologies and their products, such as [insert long list here]. So far as I can tell, the term “advanced nanotechnology” has never lost this meaning, and hasn’t been confused with any of the foundational nanotechnologies now being developed.

I’m sure that this technology base will include some very useful specialized devices that are small, programmable, and perform specific tasks. But not accidentally-or-magically intelligent, fertile, or whatever, any more than a car is.

As for what I’d call the popular conception of “nanobots”, though, the word “silly” comes to mind.

nanotürkiye March 11, 2009 at 8:01 am UTC

I think I could not express myself very well in previous comment.

Here is a cartoon: http://www.sciencecartoonsplus.com/images/miracle3.gif

Nanobots are that “some miracle” in the cartoon.

Whenever there is a problem, people tell that by using nanobots we will solve that. But nobody thinks, as you said, from an engineering perspective.

Benjamin Abbott March 11, 2009 at 2:16 pm UTC

Robert Freitas wrote a paper on how “nanorobots” could convert Earth’s biosphere into “lifeless chemical sludge” in a matter of minutes. J. Storrs Hall has detailed how tiny robots could fill the air, forming into objects at the controller’s whim and granting classically magical abilities. Invisible spirits indeed! With such ideas developed by prominent researchers, you can’t be surprised they’ve captured popular imagination. Practical near-term applications lack the whizbang to compete. I think that’s just the nature of spectacle and mass media.

Scott Jensen March 11, 2009 at 7:03 pm UTC

First, I have heard the term “mature nanotechnology” more than “advanced nanotechnology”. The “mature nanotechnology” being the essential peak of the technology. A point that people can say, “We’ve arrived.” I think “advanced nanotechnology” would always be what’s coming next.

Second, the public is going to name it something. They won’t name is “advanced nanotechnology”. It just is too long. The battle is really over whether “nanobots” or “nanites” wins. Then again, they could both win and simply be interchangeable. Nothing wrong with that as it will make reading less repetitive.

Then again, Eric, if you can come up with a short substitute, it could be a contender. But for it to be a real contender, I’m afraid you’d have to title a nanotechnology book with it.

Eric Drexler March 12, 2009 at 6:24 am UTC

I can see some advantages to the term “mature nanotechnology”, but “nanobot” and “nanite” aren’t substitute terms for anything, they’re just off-topic labels attached to a mass of pop-culture confusion.

Scott Jensen March 12, 2009 at 6:20 pm UTC

Actually, there is already a shorten version for nanotechnology and that’s “nanotech”. That might do for what you’re wanting, Eric. However, “nanobot” and “nanite” are here to stay for two reasons.

First, one of their attractions is that both enable the general public to think of the technology as a finite object. Sort of like how “factory” is something the public can visualize and not “assembly-line manufacturing”.

Second, what would you call the nanoscale devises that many talk about with the arrival of mature nanotechnology? For example, the anti-cancer nanoscale devises that float through the blood stream hunting down cancer cells. What would you name those? Or are you saying those will never come to pass?

Eric Drexler March 12, 2009 at 7:33 pm UTC

Short terms will always be in use, and “nanofactory” is a reasonable label for an important and fundamental class of anticipated technologies. Microscopic medical devices will of course be feasible; therapeutic nanoparticles of various sorts are under development, and nanoscale structures can be made progressively smarter on a continuum that leads to quite sophisticated — and precisely targeted — devices. The practice of lumping together a wide range of devices under a label that has come to mean “magic nanobugs” should, I think, be discouraged.

I’d like to see discussion focus on what physics, engineering analysis, and experimental results tell us about both longer-term objectives and next steps on pathways toward increasingly capable atomically precise fabrication. Chatter that revolves around vague words is useless, or worse — especially when the words have come to denote absurd ideas. These words and ideas won’t go away, but they can be recognized for what they are.

Eric Drexler August 9, 2009 at 12:36 pm UTC

I’ve moved this discussion from another post:


James 08.07.09 at 11:07 pm UTC
Eric, did you see the GI Joe movie? It features ‘grey goo’ they refer to as ‘nanomites.’ Made me think of you when I was watching it. :)


Eric Drexler 08.08.09 at 2:59 pm UTC
@ James — This mythology is one of several reasons why I’ve never liked “nanobots”.


James 08.08.09 at 9:53 pm UTC
I’ve always wondered about your stance on ‘nanobots’ – you seem to be saying (correct me if I’m wrong) that productive nanosystems will resemble manufacturing plants, not self-replicating robots. This is because self-replicating robots are hard, correct? But how are you going to build an entire manufacturing plant on a macroscale of product nanosystems without some form of self-replication? And eventually, you will need ‘nanobots’ for things like human body repair, and fight off hostile nanobots in the environment, and so on. These are all things from EoC, but now it seems your stance has changed but these problems will still exist. Whether we like ‘nanobots’ or not, we are going to need DNA repair and nano-policebots, and some way to make huge numbers of macro-scale productive nanosystems. Do you think that macro-scale productive nanosystems can be mass produced without self-replication? What about hostile nano-agents, and other things we need defense from.
I read EoC, and I believe you were correct with what you wrote then, I don’t see it as a mistake as you seem to see it now. (sorry this rambles off topic, is there a better forum for this discussion?)

Eric Drexler August 9, 2009 at 1:07 pm UTC

@ James — The post and the comments above outline my views on these topics. Trying to play out technology scenarios many moves and years into the future can be valuable (in the right context, and with due skepticism regarding any conclusions we can draw today), but the constellation of muddled ideas around words like ‘nanobot’ has been worse than useless.

On the particular topic you raise, the key point is that scalable atomically precise manufacturing has requirements much like those of more conventional automated manufacturing. Making products and building new production equipment requires machines, but factories don’t use machines that can build copies of themselves. This would be difficult and pointless, because specialized machines are so much simpler and more efficient. See my posts on how automated manufacturing works — without even ordinary factory-style robots, and for good reason:

High-Throughput Nanomanufacturing: Small Parts
High-Throughput Nanomanufacturing: Assembly
High-Throughput Nanomanufacturing: Assembling larger products

{ 7 trackbacks }

Previous post:

Next post: