Future Current

In the 1980′s while a researcher at Rutgers doing artificial intelligence and computer architecture, J. Storrs Hall learned about Eric Drexler‘s ideas and founded the sci.nanotech Usenet newsgroup, which he then moderated for over a decade. He is the inventor of various nanotech concepts, ranging from utility fog to space launch towers. The founding chief scientist of Nanorex Inc., he is a member of the Foresight Batelle productive nanosystems roadmap working group. He has also published the book Nanofuture: What’s Next for Nanotechnology, which won the Foresight Institute’s communication prize in 2005. His latest book, which came out this summer, is Beyond AI: Creating the Conscience of the Machine. At the 2007 CRN conference entitled “The Future of Nano & Bio Technologies,” he went down the list from A to Z of things you could make with something called a “nanofactory.”

The following transcript of J. Storrs Hall’s CRN presentation “What Could A Nanofactory Make?” has not been approved by the author.

What Could a Nanofactory Make?

If you started talking in 1950 about software for taking statistics from website customers or virus detection or something like that, people would think you were totally crazy. Basically at that point computers were for solving equations, solving equations and solving equations.

According to Merkle’s Law, every talk should have one slide…

In computers we have this fairly remarkable Moore’s law timeline.

In the 1960′s we had these centralized megabuck machines. In the ’70s they came down to the 100k level. I was at a very small private university and we had an IBM 1130 that cost about 100k. Also in the ’70s hobbyist machines came out. I managed to build in my dorm room a computer for about $5000 that was the equivalent of a $100,000 computer they had in the basement of the university, though it didn’t have the printer and the card-reader.

In the 1980′s the PC really came of age. For basically the same amount of money you were able to get a machine that could be used by someone that wasn’t a computer hacker hobbyist. It had a killer app that began to make them useful enough that people felt they actually had to have one. In the 1990′s you got to the point where something that was previously considered a fairly expensive workstation got down to the level where you were able to buy it for the same general price as a PC. Desktop publishing took off, requiring a lot more compute power. Now in the 21st century we have computers that are the equivalent of professional movie studios for doing film editing. Again, they require several orders of magnitude more computing power, but it’s there.

One of the things about nanotechnology that we expect is when it gets going, it’s going to follow the same kind of timeline and price improvement track that computers did, only for physical machinery. In the 1990′s if you were lucky, and at the right place, you could simply draw something on your workstation and email it over to the machine shop, where there were big rooms full of fancy machines, and they would make it and send it back for you. There are companies that will do this for you as well, nowadays. Companies can get these rapid prototyping, the machines exist in various forms. Most of them simply make stuff out of plastic, but there are actually prototype, prototyping machines that will make stuff that has circuits built into it. There are prototype machines that make stuff out of metal, ceramics, and quite a number of different materials, although many of these are still experimental and very expensive.

On the other hand, just like the computer hobbyists of my college days, there are beginning to be people making hobby rapid prototyping machines – fabricators.

Off the top of my head I can think of the RepRap machine in England and the Fab @ Home in Cornell, which is actually not very far from where I live. These machines are coming in at the same kind of price range as the early hobby computers, because that’s the kind of price range that hobbyists can afford.

They’re now getting to the point where you can actually make stuff with them, but they still require a lot of work. You have to be into it and nobody is going to feel that they have to have one. I think that if you project out the same sort of timeline that you saw from computers, and I don’t see any reason to imagine that it’s going to be going any significantly slower, you’re going to wind up with killer apps for fab machines cheap enough to be owned by small businesses and universities. Then, finally, people will have them in their houses and be able to do fancy stuff that we today would consider to be so crazy, as 30 years ago we would have considered having a full-fledged movie studio in your home to be crazy.

What are these machines actually going to make stuff out of? What you see on the left are a couple of items made from some of the standard fab machines. There is a metal piece and a plastic piece there. On the right you see something that was actually made from the fab @ home machine, which will make stuff out of peanut butter, chocolate, and icing. Many of these things, because they are prototyping machines, they make stuff for example out of wax and then you are supposed to take the wax and cast stuff into metal. The machines of today are not necessarily going to produce the finished product, but they can get a jump on producing the finished product.

In the long run, people won’t buy something that they have to do those extra things to. The machines will ultimately come to simply produce whatever it is you want right out of the top. What they are going to produce are raw materials. This is part of an DARPA-sponsored project at the University of Massachusetts to produce a factory that was built of Legos. The cool thing about Legos is that they are sort of digital. If you have a Lego, you know exactly where it’s going to fit because it has these bumps that match up to the holes in the bottom and so forth. So you can make something that is as remarkably precise.

This should look slightly familiar to some people in the audience. We can do the same sort of trick at the molecular scale. The key issue here is that if you can have these chemically synthesized nanoblocks, you can actually make molecularly precise objects using the same tricks with mechanisms that themselves don’t have to be molecularly precise. Even if the gadget that you build out of these nanoblocks is not diamond-stiff, you still have a hope of building a machine that is precise enough to build more of the same machine. In the long run, your looking at building stuff out of actual elements and doing all of the low-level reactions.

For very good elements to start with we colloquially refer to as CHON: carbon, hydrogen, oxygen, nitrogen. This is almost all of the human body and an enormously large part of most materials that we are used to working with, including food, wood, fiber, plastics and so forth. A huge amount of the stuff that we use is primarily made of these four elements with trace amounts of other stuff. Basically you are looking at a technology that ultimately will, because of the range of things that it can make out of the few most useful elements like carbon, won’t have to use too many other elements, except for the specific case of food, where you’re trying to make something you need to include all the trace elements for the human metabolism. Most of the ordinary everyday objects that we make can be made almost entirely out of CHON.

How fast will this be able to work? There is always going to be a difference between a machine that is simple and does a single kind of operation in a predetermined mechanical way and a machine made of the same kinds of parts and with the same ability to control that acts more like a robot arm and makes choices. There will be several phases. In the nanofactory movie, you actually see both these kinds of operations going on. You see in the early stages these factory-like assembly lines of molecules that similar to the one that you see here that do the same thing over and over and over. That’s probably going to be quite fast. If you build a machine that is complex enough, because it still is extremely tiny, it will be able to rip through stuff at a remarkably high rate. Even though the following machines that do things somewhat less directly and systematically will still be moving very fast by our standards. We can assume that the parts of the machine that do very stylized, direct molecular manipulation are going to be fast enough basically that you can just pour stuff in and it comes out the bottom transformed in whatever chemical way that you want.

This is an early version of a nanofactory concept that is Ralph’s, showing how you go through these early stages where you are creating reagents and slapping them together in a factory-like mechanistic way. In the later stages you are taking robot arms and putting blocks together in a programmed fashion. As Ralph pointed out, the figure of merit for a machine like this is how long it takes to produce a copy of itself. Or, less specifically, how long it takes to produce some mass in products.

It turns out that I was one of the reviewers of a book by Ralph and Rob Freitas. One of the things that got mentioned about this convergent assembly and that bugged me was that the third power law for replication time of nanomachines sounded right, but in life, which is our existing model of self-replication machinery, there is actually a well known and very well documented fourth power law. It takes, rather than the cube root of the size of the machine, the fourth root of the size of the machin. That’s the form of the function that tells you how long it takes to replicate itself. I won’t actually put up equations the way Ralph did, because I’ve already given my obscure slide, but I proved a theorem that showed that if a machine were actually fractal in the precise sense, where it was completely self-similar at all scales, that yes, in fact it was true, you had to have a third power law.

But you could build a machine that was slightly un-fractal, that was double fractal, that had a self-similarity but two different scaling rules. One of which is the sizes, and the other is the ratio of the size of this pipe the output comes out to the size of the cube. The pipe keeps getting bigger in relation to the size of the cube as you build the thing. In fact, given a machine like that, you could have a fourth power law.

That’s what this picture means, in case anybody wondered what that was: the illustration for the fourth power law version of the self-replicating system. If you go and get hold of Ralph and Rob’s Kinematic Self-Replicating Machines book, which is a classic of its field, you’ll find that in there, along with the mathematical theorem.

So much for the preliminaries, now for the fun part. What could you actually make with a nanofactory?

You can make Apples. One of the things I want to do while I’m going through this is talk about the level of technology that you would need. We are in fact talking about several different generations. If I’m saying, “What kind of software could a computer run?” one of the things I want to toss in is which decade I’m talking about. It seems reasonably likely that gadgetry like this could be made from fab machines that we will be able to build in the next decade, not out of complete raw feedstock materials, but out of stuff that is cheap enough to use as feedstock, like reprogrammable chips and little chunks of light-emitting stuff.

You could feed them in in canisters and boxes. Even though it seems unreasonable to imagine synthesizing something like an iPhone from molecules in the next decade, maybe a couple decades later that will happen. In the next decade, there is enough reconfigurable low-level background stuff that you could use in machines like this that I would not be terribly surprised to find a home synthesizer, or at least a hobbyist or corner store synthesizer, that could build stuff like this sometime between 2010 and 2020.

Bicycles. Let me point out that one of the nice things that near-term nanotechnology can do for synthesizing machines is in fact to come up with new materials that have the nice properties to be used in these machines. Not just chocolate, plastic and wax, but stuff that is strong enough to be used for structural uses, such as crazy designs like this.

Cups of coffee. Including the cup. The cup you can do today, if you don’t mind it being yellow plastic. That’s a very simple thing for a fab machine to create. Food in general is something interesting, because we were talking about mostly machines that use simple molecules whose deposition reactions can be analyzed today with available software on PC’s. It doesn’t mean that is actually the way they have to work. One fairly obvious set of feedstock, especially for a machine that you are hoping to make food with is a liquid that includes, for example, specially prepared fat, carbohydrate and protein molecules. Perhaps the proteins would include some synthetic amino acids with cool stuff in them that’s not normally there. Virtually all the fats, carbohydrates and amino acids are pure CHON. There are two amino acids that the body uses that have sulfur, but that’s it.

The chemistry for unlinking and relinking amino acids into proteins is simpler than the stuff that you need for pure, complete from the ground up synthesis. I see no reason, if you are anywhere near close to that, why you can’t make with machines using appropriate feedstocks of previously synthesized proteins and so forth a wide range of foods that were guaranteed to be edible. You’re not trying to build these food molecules from scratch, but from sort of the same set of molecular building blocks that animal life does.

You can do a remarkable number of stuff with proteins in terms of what they look like, act like, and taste like. For example, at the molecular level, when you taste sweetness there is a sensor, a fairly complex molecular gadget that is part of your taste buds, and it sticks to the cell wall and has this molecular clamshell that reaches up. Inside the clamshell, which can basically open and close, there is a whole in the middle that is in the shape of the sugar molecule. So when a sugar molecule floats along, it will close and bind on that. The closing actually causes a conformational change that activates something in the cell that causes the cascade of activity that causes you to taste sweetness if there’s enough of it. It turns out there are some proteins that taste sweet, and yet they are shaped nothing like sugar. What happens is, it turns out that they they dock with the stem and cause the conformational change to happen, without the molecule being in the sensor. And I think Stevia is one of them. When you want to go and build some food that doesn’t make you fat but tastes sweet anyway, you’re all set.

Diamonds. This probably doesn’t need a lot of extra explanation.

Eggs. Well, we’re back to food again. I think it’s very likely that we will be able to simulate eggs as a thing to eat, and even the lettuce underneath you see there. I do not think it is going to be possible to produce actual hatchable eggs, because there are a lot of these cute protein tricks that you can’t get around. I don’t look for being able to produce living creatures with synthesizers in the easily foreseeable future. I think that in the long, long term it is inevitable, but it’s longer than I care to think about. Anyway, eggs to eat, but not eggs to hatch.

Folding furniture. This is pretty darn straightforward, actually. It just points out that virtually any standard-sized objects that you can pick up and carry around could easily be produced by a fab machine. As you saw in the nanofactory film, it is almost going to be par for the course expected that the thing comes out the machine, especially in the early days, in some sort of folded up form, and you pull it out into the larger form. That is a more efficient way of using the expensive space in the machine to produce stuff.

Gadgets and gizmos galore. I will talk a little bit about electronics. I don’t know how clearly it came through from T‘s talks, but one of the things he shows is that you can build remarkably complicated circuitry right into the structure of your object with exactly the same techniques that you use to build a big, solid piece of stuff. All you have to do is get conducting blocks and semiconducting blocks in the right place and there you are. As a matter of fact, it’s the case now that there are fab machines that will put down enough different kinds of materials to draw the circuit board. You still have to put chips on them for the functions, but the circuit boards for these things are fabbable with current technology, and in fact stuff that a hobbyist could build if you were really serious about it.

Headphones. Of course, these are big, bulky things. The range of equipment for entertaining yourself or being part of a virtual environment is something that has a very strong economic driver behind it. It’s pretty clear that when people start trying to build things with these gadgets, this is the sort of thing that many of them are going to start looking at first.

Ice cream. In this particular case, we are talking about temperature. On the flip side of the cold ice cream you have the hot food. If you have a machine that is specialized for food, which in the long run you almost certainly would, rather than synthesize this slab of meat and stick it in the stove, you would synthesize it cooked. I do want to point out that I think that once the capabilities get to the point of being able to produce food as well as plastic gizmos, it seems very likely that there are going to be specialized machines, probably more specialized than our computers are today. There are going to be a wide variety of machines that handle different things and have different capabilities. You wouldn’t really care whether your desktop matter printer could produce a PDA that was hold or cold, but you would with the food. Chances are the food synthesizer would be able to do temperature differences but the desktop one wouldn’t.

Jackets. Besides having this optical stuff that would allow you to see 200 miles, be invisible, or, dare I point out, project a laser beam sufficient to crisp someone at many of these distances, your clothing producing synthesizer is not going to look like one of these desktop things. It’s going to look like a closet. We’re talking about a technology that can produce this absolutely enormous amount of stuff. You have to think what’s a house going to look like once this technology comes in? There is probably going to be a lot less furniture because whatever you need you can just grab. For furniture-size stuff, there is going to be one or more synthesizer that are actually room size.

So when you decide to get dressed in the morning, the door of the thing is probably a 3D video screen. You stand in front of it, and what you see is not exactly your reflection, but what would be your reflection if you were wearing the clothes that you are talking about. You tell it, I’d like to wear so and so, and it shows you wearing that. It goes through your list of what it has software patterns for. It’s not necessarily going to be this little box. The little box implies it’s easy to make a big box, which implies it’s easy to make a closet- sized box. No one is going to want to pull their clothes out of a counter-top unit.

Knives. This actually brings up an interesting point. All the knives I show here, and all the knives I expect to be coming out of your home synthesizer, are ceramic knives. It smacks a little of these Dungeons and Dragons sorcery stories where the magician whips up this wonderful world, except that if you touch it with cold steel it all disappears on you. Many of the materials, particularly metals, that we use today to make stuff out of, are not going to be the materials that we choose to use in synthesizers. We may well be moving into a post-metallic world here. As we very well know because of examples like this, it’s very possible to make better than steel knives out of ceramic materials. If knives you want, knives you’ll get. There is one asterisk to this slide that I won’t tell you about until later, so keep it in mind.

Lights. Producing power-handling material. I doubt they will be building incandescent lights, but we are beginning to understand quite a lot about the solid state physics for making lights out of things like LEDs. You are seeing more and more lights that are LEDs nowadays. They’re more efficient, cleaner, and last longer. My guess is that your synthesizer will be able to produce lights in vast profusion and pretty much whatever shape and color you want.

Money. Of course, this is all Antarctican money so no one would care if you made as much of it as you wanted. But this problem has sort of already showed up with the ability of people with color printers and xerox machines to reproduce paper money. Now, the fact is they can’t produce paper money that looks and feels like the real thing printed by the Mint, but a top-notch home synthesizer would. So, we’re kind of in a quandary here. How do you keep people from just going off and printing their own money? I don’t have a super solution right off the top of my head. I will point out that it is probably not as bad a problem as people necessarily think it might be when they first consider it. You’re going to have a lot more credit and everything you do is going to be connected to a central computer that knows who you are. It’s already the case that you can now get credit cards that you wave in front of an RFID gadget rather than go through the rigmarole of having to sign this stuff. That’s only going to become more and more common.

The fact that paper money is going to get easier and easier to copy is just one more factor in what has been an ongoing process. The counterfeiters on one hand and the Mints on the other have been going at each other for ages. That isn’t going to stop, but it isn’t going to collapse the economy, either.

And, of course, they will be able to produce other nanofactories. This is probably going to have more of a social impact than money. This is actually one of John Burch‘s earlier renditions of a convergent assembly version of this thing. The idea is that we are going to have a self-replicating technology here in the sense that it will be able to make copies either of itself or parts that could be assembled reasonably easily, like today’s RepRap machine. Being able to reproduce the capital, in the industrial sense, of the world for almost nothing is going to bring to physical objects the same kind of problems that we have already been seeing for the past few years for things like music.

There are social forces that make a lot of money out of the scarcity of physical manufactured goods. They are going to be all in a tizzy over people being able to make stuff for themselves. In particular, people being able to make nanofactories for other people to make things for themselves, because it’s just like the copying of digital music. Once you let one of them get out there, you can’t sell another one. There is going to be a very serious social question and I have no idea how it’s going to come out. I just want to point out what this self-replicating property of nanofactories is going to bring to the fore.

Office supplies. Well, books, pens and paper. Probably the office of the future is going to look more like a virtual reality. Your PDA is going to look more like a pair of glasses and a glove, or something. The gadget-type nanofactories are very likely to be able to produce that sort of stuff for you, so you won’t even have to have a home office. You get up in the morning and decide to work, you go to your closet and say, “Make me an office.”

If you want good old fashioned stuff, there’s no reason that your synthesizer, especially the closet-sized one can’t make you one, whether you like them in the newest styles or the good old fashioned ones. The only problem with them is that they actually use a lot of metal in those things. The Victorian Age was when the West discovered cast iron, so there was a lot of heavy metals in some of these old things. That is very unlikely to be what what your synthesizer produces. But perambulators, sure.

Queens. And other chess pieces. In fact, these are something you can build today with almost any fab machine. This is very straightforward. One of the really cool things about the fab revolution, if you read one of Neil Gershenfeld’s books, one of the really good properties of having these fabrication machines is that they open up people’s creativity. You can spend your time designing a new thing, playing with it and getting it right, and like people do with their websites, throw it out for other people to use. People have different skills and artistic talents, so when I design a new chess set or a whole new game, I can throw it out there and it costs me nothing. I’m just transmitting information. Anyone who wants to then have it would be able to build it in their own home. A whole culture of people using home synthesizers, nanofactories, fab machines, and so forth is going to be a much more creative and much less scarcity-limited culture. I think that is one of the major up-sides that we can’t lose sight of.

Robots. After I designed the flying car for NASA a few years back, I looked at the stuff that I had come up with and realized that I could use it to build a robot, except that the robot, which would be of human size and strength, would actually have a mass of a few grams and could fold up into the size of a ballpoint pen. So you don’t actually need a closet-sized synthesizer to build a big robot. You go to your desktop one and it produces a pen that you put in your pocket, but when you need the robot you take it out and it inflates itself to be just as big and strong as a human being. That’s sort of the up-side potential of using diamond as a structural material.

I don’t know what I should say about this. The one thing I probably should point out is that one of the major economic drivers for the internet was pornography. Nanotechnology of basically the same era will be able to produce synthetic human beings virtually indistinguishable from real human beings. Maybe everybody has guessed more about this than I need to say.

Tennis racquets. The good old fashioned wood kind, I think you could take a later version synthesizer and build something that was a very close analog to wood that would be very difficult to tell apart. People who like natural wood stuff would be able to have it. It would also be a nice increase in quality from the plastic stuff that everything gets made out of today.

Utility fog. The idea here, what if you took little robots that were able to extend and retract their arms so that the entire structure could change shape? You would get something that acted like a completely programmable material. It could even act like a gas, because though it was heavier than air, when you moved your hand it would actively get out of the way. You would only feel the same pressure against your hand as if you were moving it through the air. If you were to take your entire house and fill it with this material, you would essentially be in the position of a magician, because when you wanted something to appear you would simply request that it stop simulating air and start simulating the object. This is the answer to the people who want their offices to appear instantly in the morning. Not only the office but anything could appear instantly in the morning.

This is a much longer term invention. This is not going to happen in the next couple of decades, even though the actual robot you could build today. You just couldn’t build them small enough. You have to first get to the point where molecular manufacturing is really, really cheap, and you can build literally trillions of robots. Then you have to solve some really enormous software problems to control them. So this is blue sky, latter half of the 21st century stuff. But it nicely occupies the position in the alphabet.

Volantors. This is the word that Moller has coined for a flying car. I did this design for NASA about eight years ago using nanotech as the basic technology for it. It gets around the noisy part by having sails that consist of tiny little fans. It’s build of diamond balloons like the robot I was talking about. That’s a pointer to the sort of stuff a mature nanotechnology could build. If you’re talking about the late end of the scale, you’re getting to the point where not only your closet but your garage is a fabricator. So when you go out, you get into whatever kind of machine you want today.

Watches. It’s almost certain that we won’t need watches of this kind. We will have contact lenses that are full-function displays and let us see whatever it is that we want to be able to see. If we want to have the time displayed in part of our visual field, then that will happen.

You knew this was coming, right? People are going to be inventing new kinds of music and of expressing themselves, in terms of their costume or other modifications to their person. The fabricator revolution is only going to help them do this.

If you want to go camping, you carry this thing. You can almost do this with current-day materials. You could carry a fair-size tent as a backpack. With nanotech it would be able to be lit, air-conditioned, and have all the other comforts of home. You might just go trekking out in the wilderness with a fabricator on your back, making whatever you need at the moment, including a particular food. I imagine that for outdoors hiking, you would probably want one that used cellulose as a feedstock, so you could shove dead wood into it.

But not zircons! Why? A zircon is currently a fairly cheap costume jewelry type of gemstone, but it uses an element, zirconium, which is not really used for anything else. I can’t really think of any good reason that anybody’s home synthesis machine would use zirconium as a feedstock or would have somebody go through all the trouble it takes to work out the zirconium deposition reactions and so forth. At least that was the case when I wrote this list in my book a couple of years back.

It turns out actually that those ceramic knives that I showed, zirconium oxide is the ceramic they use for those, so that was my asterisk. But by and large, diamond knives work just as well, so it’s very unlikely that people would actually build your standard home synthesizers to use zirconium. Who knows, maybe you could use zircons as money?

This entry was posted on Saturday, December 8th, 2007 at 12:19 am and is filed under CRN Bio & Nano, J. Storrs Hall, Nanotechnology. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.