Sunday, October 2, 2016

SF Setting - Nanotechnology

Yes, it’s time to talk about one of the most troublesome technologies in science fiction. As with artificial intelligence, the full promise of nanotechnology is so powerful that it’s hard to see how to write a story in a setting where it has been realized. It takes some serious thought to even begin to get a grasp on what kinds of things it can and can’t do, and the post-scarcity society that it logically leads to is more or less incomprehensible.

As a result, mainstream SF generally doesn’t try. In most stories nanotech doesn’t even exist. When it does it’s usually just a thinly veiled justification for nonsensical space magic, and its more plausible applications are ignored. Outside of a few singularity stories, hardly anyone makes a serious attempt to grapple with the full set of capabilities that it implies and how they would affect society.

Fortunately, we don’t have to go all the way with it. Drexler’s work focused mainly on the ultimate physical limits of manufacturing technology, not the practical problems involved in reaching those limits. Those details are mostly hand-waved by invoking powerful AI engineers running on supercomputers to take care of all the messy details. But we’ve already seen that this setting doesn’t have any super-AIs to conveniently do all the hard work for us.

So what if we suppose that technology has simply continued to advance step by step for a few hundred years? With no magic AI wand to wave engineers still have to grapple with technical limitations and practical complexities the hard way. The ability to move individual atoms around solves a lot of problems, of course. But the mind-boggling complexity of the machines nanotech can build creates a whole new level of challenges to replace them.

The history of technology tells us that these challenges will eventually be solved. But doing so with nothing but human ingenuity means that you get a long process of gradual refinement, instead of a sudden leap to virtual godhood. By setting a story somewhere in the middle of this period of refinement we can have nanotechnology, but also have a recognizable economy instead of some kind of post-scarcity wonderland. Sure, the nanotech fabricators can make anything, but someone has to mine elements and process them into feedstock materials for them first. Someone has to run the fabricators, and deal with all the flaws and limitations of an imperfect manufacturing capacity. Someone has to design all those amazing (and amazingly complex) devices the nanotech can fabricate, and market them, and deliver them to the customer.

So let’s take a look at how this partially-developed nanotech economy works, in a universe without godlike AIs.

In order to build anything you need a supply of the correct atoms. This is a bit harder than it sounds, since advanced technology tends to use a lot of the more exotic elements as well as the common stuff like iron and carbon.

So any colony with a significant amount of industry needs to mine a lot of different sources to get all the elements it needs. Asteroid mining is obviously going to be a major activity, since it will easily provide essentially unlimited amounts of CHON and nickel-iron along with many of the less common elements. Depending on local geography small moons or even planets may also be economical sources for some elements.

This leads to a vision of giant mining ships carving up asteroids to feed them into huge ore processing units, while smaller ships prospect for deposits of rare elements that are only found in limited quantities. Any rare element that is used in a disproportionately large quantity will tend to be a bottleneck in production, which could lead to trade in raw materials between systems with different abundances of elements.

Some specialization in the design of the ore processing systems also seems likely. Realistic nanotech devices will have to be designed with a fairly specific chemical environment in mind, and bulk processing will tend to be faster than sorting a load of ore atom by atom. So ore processing is a multi-step process where raw materials are partially refined using the same kinds of methods we have today, and only the final step of purification involves nanotech. The whole process is likely different depending on the expected input as well. Refining a load of nickel-iron with trace amounts of gold and platinum is going to call for a completely different setup than refining a load of icy water-methane slush, or a mass of rocky sulfur compounds.

Of course, even the limited level of AI available can make these activities fairly automated. With robot prospecting drones, mining bots, self-piloting shuttles and other such innovations the price of raw materials is generally ten to a hundred times lower than in real life.

Limits of Fabrication
In theory nanotechnology can be used to manufacture anything, perfectly placing every atom exactly where it needs to be to assemble any structure that’s allowed by the laws of physics. Unfortunately, practical devices are a lot more limited. To understand why, let’s look at how a nanotech assembler might work.

A typical industrial fabricator for personal goods might have a flat assembly plate, covered on one side with atomic-scale manipulators that position atoms being fed to them through billions of tiny channels running through the plate. On the other side is a set of feedstock reservoirs filled with various elements it might need, with each atom attached to a molecule that acts as a handle to allow the whole system to easily manipulate it. The control computer has to feed exactly the right feedstock molecules through the correct channels in the order needed by the manipulator arms, which put the payload atoms where they’re supposed to go and then strip off the handle molecules and feed them into a disposal system.

Unfortunately, if we do the math we discover that this marvel of engineering is going to take several hours to assemble a layer of finished product the thickness of a sheet of paper. At that rate it’s going to take weeks to make something like a hair dryer, let alone furniture or vehicles.

The process will also release enough waste heat to melt the whole machine several times over, so it needs a substantial flow of coolant and a giant heatsink somewhere. This is complicated by the fact that the assembly arms need a hard vacuum to work in, to ensure that there are no unwanted chemical reactions taking place on the surface of the work piece. Oh, but that means it can only build objects that can withstand exposure to vacuum. Flexible objects are also problematic, since even a tiny amount of flexing would ruin the accuracy of the build, and don’t even think about assembling materials that would chemically react with the assembly arms.

Yeah, this whole business isn’t as easy as it sounds.

The usual way to get around the speed problem is to work at a larger scale. Instead of building the final product atom by atom in one big assembly area, you have thousands of tiny fabricators building components the size of a dust mote. Then your main fabricator assembles components instead of individual atoms, which is a much faster process. For larger products you might go through several stages of putting together progressively larger subassemblies in order to get the job done in a reasonable time frame.

Unfortunately this also makes the whole process a lot more complicated, and adds a lot of new constraints. You can’t get every atom in the final product exactly where you want it, because all those subassemblies have to fit together somehow. They have to be stable enough to survive storage and handling, and you can’t necessarily fit them together with sub-nanometer precision like you could individual atoms.

The other problems are addressed by using more specialized fabricator designs, which introduces further limitations. If you want to manufacture liquids or gasses you need a fabricator designed for that. If you want to work with molten lead or cryogenic nitrogen you need a special extreme environment fabricator. If you want to make food or medical compounds you need a fabricator designed to work with floppy hyper-complex biological molecules. If you want to make living tissue, well, you’re going to need a very complicated system indeed, and probably a team of professionals to run it.

Despite their limitations, fabricators are still far superior to conventional assembly lines. Large industrial fabricators can produce manufactured goods with very little sentient supervision, and can easily switch from one product to another without any retooling. High-precision fabricators can cheaply produce microscopic computers, sensors, medical implants and microbots. Low-precision devices can assemble prefabricated building block molecules into bulk goods for hardly more than the cost of the raw materials. Hybrid systems can produce bots, vehicles, homes and other large products that combine near-atomic precision for parts that need it with lower precision for parts that don’t. Taking into account the low cost of raw materials, an efficient factory can easily produce manufactured goods at a cost a thousand times lower than what we’re used to.

Of course, fabricators are too useful to be confined to factories. Every spaceship or isolated facility will have at least one fabricator on hand to manufacture replacement parts. Every home will have fabricators that can make clothing, furniture and other simple items. Many retail outlets will have fabricators on site to build products to order, instead of stocking merchandise. These ad-hoc production methods will be slower than a finely tuned factory mass-production operation, which will make them more expensive. But in many cases the flexibility of getting exactly what you want on demand will be more important than the price difference, especially when costs are so low to begin with.

So does this mean all physical goods are ultra-cheap? Well, not necessarily. Products like spaceships, sentient androids and shapechanging smart mater clothing are going to be incredibly complex, which means someone has to invest massive amounts of engineering effort in designing them. They’re going to want to get their investment back somehow. But how?

Copy Protection
Unfortunately, one of the things that nanotechnology allows you to do much better than conventional engineering is install tamper-proofing measures in your products. A genuine GalTech laser rifle might use all sorts of interesting micron-scale machinery to optimise its performance, but it’s also protected by a specialized AI designed to prevent anyone from taking it apart to see how it works. Devoting just a few percent of the weapon’s mass to defensive measures gives it sophisticated sensors, reserves of combat nanites, a radioactive decay battery good for decades of monitoring, and a self-destruct system for its critical components.

Obviously no defense is perfect, but this sort of hardware protection can be much harder to beat than software copy protection. Add in the fact that special fabrication devices may be needed to produce advanced tech, and a new product can easily be on the market for years before anyone manages to crack the protection and make a knock-off version. The knock-offs probably aren’t going to be free, either, because anyone who invests hundreds of man-years in cracking a product’s protection and reverse-engineering it is going to want some return on that investment.

All of this means that the best modern goods are going to command premium prices. If a cheap, generic car would cost five credits to build at the local fabrication shop, this year’s luxury sedan probably sells for a few hundred credits. The same goes for bots, androids, personal equipment and just about anything else with real complexity to hide.

Which is still a heck of an improvement over paying a hundred grand for a new BMW.

Common Benefits
Aside from low manufacturing costs, one of the more universal benefits of nanotech is the ubiquitous use of wonder materials. Drexler is fond of pointing out that diamondoid materials (i.e. synthetic diamond) have a hundred times the strength to weight ratio of aircraft aluminum, and would be dirt cheap since they’re made entirely of carbon. Materials science is full of predictions about other materials that would have amazing properties, if only we could make them. Well, now we can. Perfect metallic crystals, exotic alloys and hard-to-create compounds, superconductors and superfluids - with four hundred years of advances in material science, and the cheap fine-scale manipulation that fabricators can do, whole libraries of wonder materials with extreme properties have become commonplace.

So everything is dramatically stronger, lighter, more durable and more capable than the modern equivalent. A typical car weighs a few hundred kilograms, can fly several thousand kilometers with a few tons of cargo before it needs a recharge, can drive itself, and could probably plow through a brick wall at a hundred kph without sustaining any real damage.

Another common feature is the use of smart matter. This is a generic term for any material that combines microscopic networks of computers and sensors with a power storage and distribution system, mobile microscopic fabricators, and internal piping to distribute feedstock materials and remove waste products. Smart matter materials are self-maintaining and self-healing, although the repair rate is generally a bit slow for military applications. They often include other complex features, such as smart matter clothing that can change shape and color while providing temperature control for its wearer. Unfortunately smart matter is also a lot more expensive than dumb materials, but it’s often worth paying five times as much for equipment that will never wear out.

With better materials, integrated electronics and arbitrarily small feature sizes, most types of equipment can also make use of extreme redundancy to be absurdly reliable. The climate control in your house uses thousands of tiny heat exchangers instead of one big one, and they’ll never all break down at once. The same goes for everything from electrical wiring to your car’s engine - with sufficient ingenuity most devices can be made highly parallel, and centuries of effort have long since found solutions for every common need.

This does imply that technology needs constant low-level maintenance to repair failed subsystems, but that job can largely be handled by self-repair systems, maintenance bots and the occasional android technician. The benefit is that the familiar modern experience of having a machine fail to work simply never happens. Instead most people can live out their whole lives without ever having their technology fail them.

Now that’s advanced technology.


  1. Very thought and planned. It takes into consideration that nanotech is alot more than magic robogoop. As well that it is not a refined process and that it is still a developing technology. It nearly four hundreds years of using guns to get where we are today, Same said for nano manufacturing. As well as that it is not dominating science and manufacturing, yes it has opened up new technologies. But most if not all were already theorised. As well as its place in large scale manufacturing as more a step or process in being better suited for small highly delicate or complicated parts or larger simple ones and that others may be made in old fashioned processes or newer exotic ones?? Who knows what weird breakthroughs we've made in four hundred years traversing space.

  2. It depends a lot on what the reader is going to get out of the knowledge you inpart on them with all the techno stuff. If it's running the storyline yeah go for it, but if its just a filler I'd go for the KISS system. Keeping the interest of the reader on the story. Me I'm all for a little understanding about the tech but if I get to bog down with to much information I just give it a miss and step to the next interesting part of the story. Especially if it doesn't do a great deal to the storyline.

    Is the techknowledge going to be a major part of this new story, or is it just a overview of the world the main character lives in. At the present time I'm all fullup on teenagers saving the universe so may be I'll wait until you've finished your Black series before going off on another Space Odyssey.

    1. I don't have a lot of infodumps in the story itself, aside from some scenes where the protagonist is learning about things she didn't know that will turn out to be relevant to the plot. But working out all the tech is vital to making the setting feel like a realistic possible future instead of a raygun fantasy story.

      For example, the combination of AI and nanotech outlined in these last two essays determine the general wealth level of the setting. If I hadn't figured all that stuff out in advance I might have assumed that buying a new space suit is a significant expense, or gone with the genre convention of spaceships having tiny little cabins and lousy food.

    2. One of the things i found interesting but didn't see addressed were the economic implications of the combination of bulletproof copyright protection and functionally everlasting products.

      If every item you buy will never out or fail within your lifetime that means that nobody will buy more than one of a particular type of product. In fact a lot of people will likely pass on big ticket items for generations.

      That means overall demand will be much, much lower than today, and across the board.

      Which in turn means that the rate of technological innovation is going to take a _serious_ hit as development costs for new products can only be passed on to a comparatively small number of potential buyers.

      New products will either be prohibitively expensive, or take a long time before enough units have been sold that it actually starts to turn a profit for the company who developed it.

      What are the likely consequences of this? Well for one there will be very little variety in everything, and even less economic competition. The reduced markets will simply not support many small companies vying for the same number of customers.

      What you'll see is one, maybe two big companies who make make a particular type of product (say cars, or hand lasers), and whose catalog has a very limited selection that changes only very slowly if at all.

      Civilian R&D would probably become next to non-existant. Most new technologies would be first developed by the military and government sponsored university research labs as they could ignore the need for a ROI by socializing the costs.

      One alternative to this general contraction/slow down might be a broadening of the trend we have seen in the cell industry (and if Microsoft has its way soon IT) of turning consumer items into utilities.

      You don't actually own any of the items you wear, drive, live in or otherwise use, instead you have a contract GalCorp where for 49.99 Crd./month you can print out any item from their catalog that is included in your packet (Premium content costs extra naturally). If you switch to another provider, or fail to make the monthly payment, all the GalCorp stuff tied to your customer account automatically deactivates/self-destructs.

      This would give companies a constant stream of revenue with which pay for innovation, as well as a motivation to do so, and a mechanism to automatically distribute product-updates to their customer base with.

      Heh! Such contracts could/would probably even include the nano-printers themselves along with metered monthly raw material and/or power allotments similar, again, to the cellphone data plans we have today.

  3. My biggest problem with Drexler (which I read years and years ago) was he never considered binding energy and the massive energy balance problems involved in having nanaomanipulators grab an atom, move it, and then release it at the right time. I think, combined with Uncertainty physics, nanotechnology is going to be a whole lot harder than he realized.

    1. If you're interested in a detailed technical examination of the challenges and possibilities of realistic nanotech, I highly recommend the work of Robert A. Freitas. Unlike Drexler he dives deep into the implementation details, with a lot of detailed calculations.

      But yes, there's a reason why we didn't invent this stuff by 2010 the way Drexler originally predicted. Like most advanced technologies, it's a lot harder than it looks.

  4. Not to forget all of the wonderful medical applications of nanotech.
    Rapid healing, advanced strength, ability to modify metabolic rates to lower oxygen consumption during space travel, these are all fantastic advances that any space faring civilization would love too have.

    1. Make that rapider healing. Such processes would inevitably generate waste heat, and the human body has a fairly low hard ceiling on how high a temperature it can tolerate before protein denaturisation escalates and kills you.

      Which means while long convalescence periods could likely be shortened by quite a bit (days instead of weeks, weeks instead of months) the sort of cinematic healing where serious injuries disappear within in seconds (or minutes, or even hours) you often encounter in the superhero and space opera genres isn't going to happen because it'd cook the patient's flesh.

  5. Very creative and detailed imagining of future tech. What I am unable to understand is where the waste heat actually comes from? As in what part of the process actually generates heat and why the heat cannot be recycled to generate power instead of being a limitation on the tech.

    1. Welcome to the Laws of Thermodynamics. Any machine that does work by necessity produces waste heat, and waste heat has lower utility than the original energy source. Moving and positioning atoms is work.

  6. I'm excited about what this means for biotech integration. If we can work at a Nano scale then we can safely explore and analyze and maybe work on integrating into the Human Brain. Obviously we would have to be careful about waste heat but low impact low heat systems, while expensive and complex, could be made and installed to interface with a human brain.

  7. All those Nanos and AIs and stuff are all well and good, but I'm seriously jonesing for a hit of DB.

    1. You are not alone. We are LEGION.

    2. > You are not alone. We are LEGION.

      ... also GROOT.

  8. About the mining, the bottleneck depends on the amount of energy available. We can, at least in theory, change one element into another via forced fission or fusion. It is just incredible energy intensive (and at least in a somewhat realistic setting only possible in orbit above a gas mine for fusion reactors powering the factory or in close orbit of a star using solar panels). It is certainly not pretty but if you need unobtainium 368 for your war fleet and you can't find any, then make some.
    The asteroid mining itself can be accomplished by spanning a tough cable over the surface of an asteroid, cutting it in 2 halves and start spinning it, creating pseudo gravity to aid mining, and then using solar powered "Laser" (in reality simply focused sunlight by mirror) to vaporize the chunks in reaction chambers separating the vapors by atomic mass.
    About the Nanos, I think there are actually quite harder problems.
    Nanos are small. Extremely so. That creates some engineering problems.
    1. They need onboard computing to interprete the orders and react accordingly.
    2. They need some source of fuel to actually act.
    3. They need some way to actually know where they are.

    One of these is a challenge. All three and we talk about micros instead of nanos. Just to make it clear, a single transistor on a modern processor is between 10 and 20 nanometer on one side. Lets assume that they make it to the 1 nm scale, a single very dumb processor (early electronic calculator levels of dumb) takes a few thousand transistors (the Intel 4004 had 2.3k) without any memory.
    The same with fuel. Sure neither the Nanobot nor the atom mass all that much, but the distances they move are in relation to their size (and speed) enormous. A single mm in distance would represent the equivalent of 150km+ for a human (or around 100 miles or more). That takes enormous energy in relation to the complete mass.
    The navigation is IMO the killer, as anything that is intelligent enough to actually say where it is and from there finds out where it should go will need considerably more processor power than an Intel 4004. We are talking several million transistors intelligent here.

    1. Its a waste to have ever single nanobot be that intelligent. It doesnt need to know where it is and what the big picture is going to be. It just needs to follow directions. Welcome to Swarm Logic.

    2. Actually as far as propulsion goes you don't really need it as we understand it. You don't need wheels and an engine like a car, you don't need a pressurized thrust based engine like a water jet and you certainly don't need something like a propeller. You really only need some way to shift the nanobots electromagnetic field a little so it gets drawn towards things. The reason for that is that anything operating on the nano scale has so little mass that gravity becomes a none factor while the distances between particles are so small the electromagnetic attraction becomes a huge factor in how things move around. If anyone wants the math to prove that I can post it but it's a little tedious to explain.

    3. Synthesizing elements that occur in nature is never going to be worthwhile, because building a bunch of giant particle accelerators that produce microscopic amounts of material is far less efficient that just building a swarm of mining drones. Although there's certainly going to be an industry devoted to making useful isotopes that don't occur naturally, like short-halflife radioactive materials for use in radioactive decay batteries.

    4. @Unknown

      What I described is the bare minimum. There is absolutely no decision making involved, no intelligence, not even on a level that allows hive intelligence. To use a nanobot for manufacture (for example) you need to tell it to place an atom at x:y:z oriented at °x:°y:°z. To enable it to do that it needs to know its position x1:y1:z1 and orientation °x1:°y1:°z1. It will have to be able to evade other nanobots and already placed atoms, and for that it has to be able to "see" them. THAT is the level of intelligence I was talking about. Anything less and you can use maybe self aligning bots for a very special job but that is not what most people think when they hear nanotech.

      @daniel young
      Sure, tiny electromagnetic charges are all that is needed, but the available energy for them is tiny as well, and if you use them to build something the distances they have to move are gigantic in relation to their size. My point was not that it takes so much energy to move, but that there is nearly no energy storage available, and as such its range is miniscule.

      @E. William Brown

      I did not mean particle accelerators, but fusion and fission.
      Use the principle you use in your fusion reactors to fuse into elements higher than iron (it is possible, you just have to add energy instead of gaining anything). Or bombard heavy materials with neutrons and split them.
      Or compress waste material x into mononeutrons and let them decay to hydrogen, starting the fusion process.
      As I wrote, extremely energy intensive, but if you absolutely need that Yttrium or Indium or whatever to build your defensive fleet and you don't have any handy asteroids containing them it is a last ditch option.

    5. @MadMcAl

      You don't really need fusion to get higher elements as there are major mines set up in the Alpha layer of hyperspace previously mentioned for the express purpose of obtaining the heavier elements. It would be easy enough to obtain from a supplier or to even mine some of the elements yourself if you had a mining crew aboard. Not to mention most large scale ships that could even consider fusion would probably have a mining crew to gather their resources so as to reduce upkeep costs for the ship both cost wise and energy wise.

      Now the way I see it is the higher layers of hyperspace are perfect for mining rare materials not found in abundance on our universe and would be perfect for all militant forces to have a hand in. Now of course spending time to wait for the crew to mine an asteroid is too inefficient for the military so large ships would most likely rove the hyperspaces using tractor beams (now possible through the momentum exchange drivers) to pick up asteroids in a flyby to use for processing into its constituent elements.

  9. One of the stories to examine the problems (and opportunities) of a post-scarcity world is Walter Jon Williams' "Implied Spaces." Far from paradise, post-scarcity brings with it whole new problems, still rooted in the human condition. What does one do when relieving boredom becomes a prime motivator, and hostilities include flinging artificial anti-matter universes at each other? How does one exercise free will when freedom of thought and action can literally be edited out of your mind?

  10. @E.William Brown, I hope you do not forget about redundancy in this new series. I do not care how supposedly "functionally" everlasting (as Grimmelhausen put it) anything is, it can still be broken, destroyed.. Redundancy redundancy.