There are no make-anything machines in “Flat Black”

There is a discussion going on in the Flat Black group page on Facebook that arises out of the discussion here about Development Levels. In that discussion the question was asked as to which in considered to be a higher tech level “a craftsman with a fabber/3Dprinter making the Gizmopie necessary for his village in his blacksmith/fabber shop, or a human/robot in big conveyor like assembly line who makes a specialized part which is assembled to make the Gizmopie in mass quantities for cheap”.

The question of tech level is moot, since Flat Black doesn’t deal in tech levels. What Flat Black does have is dev levels, and there the answer is that having the village craftsman make your gizmopie is a low-dev way of doing things even though his fabber/3D printer is higher tech level than a gizmopie. And in Flat Black everything is to an overwhelming extent done the high-dev way because it is much cheaper. Things that would not be affordable if they are made without specialisation and exchange are cheap and plentiful because of a mind-boggling degree of specialisation and exchange that is made possible by markets containing tens of billions of consumers and the exploitation of the economies of staggering scale.

A village craftsman in Flat Black probably doesn’t make a gizmopie at all, because people buy them imported for less than what he would have to pay for the materials. But if he had to, he would take some high-performance polymers imported from a low-dev world² and run them through his additive compositor to make the mechanical bits, install photonic quantum logic from Tau Ceti and a bioencabulator from Simanta, then flash the logic with firmware that was programmed on Old Earth and downloaded from an Imperial comms satellite, and charge you fifty simoleons. He doesn’t have a matter fabber that could make a photonic logic on the spot, because to scribe the features of a photonic logic into the substrate you need to focus gamma rays to a spot size of 2 nm, which requires grazing-incidence mirrors with a focal length of half a kilometre, in hard vacuum. One of Stephenson’s matter compilers could do it by atomic epitaxy. A Star Trek replicator could do it with TRANSPORTER TECHNOLOGY™.

But those things don’t work in Flat Black. Dr Doyle and Dr Watson would like to tell you why.

The account of Dr Sir Arthur Conan Doyle, KStJ, DL, MD

If Flat Black featured replicators such as Clarke described, transporter-based replicators, matter compilers like those in The Diamond Age, or GURPS-style nanofacs etc., then there would be no role for interstellar trade. So there would be little call for interstellar transport and no revenue for the Empire. More-or-less cosmopolitan PCs would have few causes and little opportunity to visit exotic worlds with strange societies on them. That would be a bad design choice.

It doesn’t matter whether you think that Clarkean replicators are likely, unlikely, or even implausible. Because Flat Black is not futurism.

The account of John H. Watson, MD

As it happens in Flat Black, Star Trek transporter “technology” simply doesn’t exist, there’s not even a physical principle that it might be based on.

3D printing by atomic epitaxy as in Stephenson’s matter compilers is useful for some limited applications, but only really when you care about shaping things with a precision comparable to the distances between atoms in typical solids. It is monstrously too slow (expensive) to consider using it to make things bigger than about 10^-10 moles, and it turns out to be not useful in assembling molecules atom by atom because of problems with thermochemistry and the stability of intermediate or incomplete molecules. Putting things together atom by atom isn’t practical, with very limited exceptions.

As for using anything like 3D printing, there is an entire hierarchy of problems.

  1. Sophisticated (“high tech”) devices require hundreds or thousands of different materials in their construction, for dozens of differently-doped varieties of silicon and germanium in semiconductor-based electronics, and dozens of differently-doped varieties of glass, silica, corundum etc. in the photonics, different metals and insulators in the the wiring, different structural polymers in the chassis and case — or titanium, steel, any of a bunch of different ceramics. Materials with different light-emitting and photoelectric properties, with different electrical properties, with different magnetic properties, with different mechanical properties. You aren’t going to be able to extrude all of those different things out of the same extrusion heads or whatever: you’re going to need hundreds of thousands of different, or hundreds of expensively adaptable, print capabilities in the same space. A make-anything 3D printer will have to consist of hundreds of different printers for different materials, all printing into the same space. But while you will need all those things to be able to print anything, you will only need a very small subset to print any particular thing. So if you want to print a samurai sword and a kevlar bustier that is going to tie up all the capabilities of the 3D printer, while producing a product that could have bee made by a far simpler and cheaper machine that is only capable of printing four materials (austenite, pearlite, kelvar, and nylon). A make-anything machine is going to have to have the special thinggumy to doping corundum with rubidium in it, even if most users will never make anything involving a ruby laser in their fabbers’ entire service life. So even if 3D printing were a practical and economic way to make things of all sorts of materials (which it isn’t), a single make-anything printer capable of working in any material would be vastly more costly than a range of separate fabbers combined with a number of assembly steps. And in Flat Black it worked out that way.

  2. In fact, a lot of materials you can’t make right by amassing extruded drops and threads anyway, In some materials successive deposits don’t adhere properly, because of drying, cooling, or polymerisation. In some don’t flow enough to fill in defects, and for some purposes you can’t afford the defects. For some uses (polaroid, mylar, high-strength fibres) you need the molecules to be lined up (as by drawing or stretching). Some materials have to be annealed, others quenched, others surface treated after forming (chemically strengthened glass), some treated with intense light, some kept in absolute darkness, some formed as droplets in free fall. Some things require of may require really decent-sized components to be made all out of a single crystal, and you can’t squeeze out a crystal through multiple passes of an extruding head. You can’t make everything — in particular you can’t make high-performance high-tech gee-whizz stuff — entirely out of thermoplastic resin and sintered metal.

  3. In most high-tech products there is an intimate intermingling of high-spec components that require specialist materials and ten-nanometre tolerances, all the way to plastic cases in which a millimetre here or there is neither here nor there. To build the whole thing to 10 nm tolerances involves 10^15 times and many injection operations as it necessary, and has to work out thousands of times slower even with billions of times as many print heads billionths of times the size.

  4. Speaking of 10-nanometre print heads, friction, surface tension, and conduction of heat are complete cows at tiny scales.

The best way to look at it is that there is a make-anything machine in Flat Black. It’s 300 light-years across, cost something like ¤100 000 000 000 000 000 to build, has half a billion people as moving parts, and is called “the interstellar economy”. Even that degree of miniaturisation and portability it achieved only by specialising ruthless to make only the things that people actually do want at the price, not everything that they might want.

Brett Evill's afterword

I am well aware that one of my failings in this sort of discussion is that I laid out a Watsonian rationalisation for a Doylist setting-design decision, and then when other participants discuss the details of the Watsonising (whether its propositions are true, universal, have alternatives) I tend to misconstrue those discussion as challenges to my statements about what is true in my setting, as challenges to its Watsonian plausibility, and as attempts to force me to change the decisions I made under Dr Doyle’s charming hat³. I tend to react to those as attempts to force me to spoil my setting.

Do not be afraid! This time I am fore-warned. I welcome a general discussion of replicators and matter integrators, and will strive to refrain from treating is as a discussion of such things in Flat Black, where there aren’t any. Whenever I am about to relapse, use some word that teems with hidden meaning like “Basingstoke”. It might recall me to my saner self.

¹ JBS Haldane was probably the first person to discuss these, because that’s what JBS Haldane was like. The earliest discussion that I am aware of is by Arthur C. Clarke, in an essay titled “Aladdin’s Lamp” that was published about 1960 (and collected in Profiles of the Future in 1962). Clarke called them “replicators”, and described variations that either required a wide variety of feedstocks or that make what they need by transmuting elements on the fly. They are perhaps better known as “replicators” in Star Trek (which make no pretence of working on any physical principle) or as “matter compilers” in Neal Stephenson’s Diamond Age, which do make such a pretence. I reckon that a lot of us here will know them best as ”nanofacs”, having encountered them in GURPS Ultra-Tech. TL;DR version of this post: there are no such make-anything machines in Flat Black.

² Epoxy bushes and kelvarbast grow just as well on the cheap land on low-dev worlds as they do on the expensive land in high-dev worlds.

³ Which is a viking helmet adorned with the fangs and claws of a Thylacoleo.

Well, we’re in the Flat Black section and I feel I’ve been derailing quite enough already.

Also I generally agree with your Watsonian justifications. (Having been operating a 3D printer at home for a couple of years now I’ve become acutely aware of its limitations and how to work round them, and I see no reason to assume future technology won’t have equivalent limitations.) In Wives and Sweethearts many warships carry small nanofactories, which can put atoms where they’re needed; but carrying a big one would cut into other capabilities, and carrying the raw materials for it even more so. Planets have nanofactories, and generally some asteroid-mining infrastructure to feed them, because why wouldn’t you, but they still mostly print parts rather than assemblies or finished products, and things made by conventional means can be stronger, lighter, etc. The main advantage of the nanofactory is that when you suddenly need something you don’t need to wait months for it to arrive by freighter.

I have been thinking about the things in my house that would be very challenging to make by additive manufacturing alone. Examples include:

  • a bottle of tonic water
  • a compact fluoro light bulb
  • a well-tempered steel blade
  • a toughened glass drinking tumbler
  • a chiller bag made of aluminised mylar
  • Polaroid sunglasses
  • an iPhone
  • magnetic door fittings that hold the doors open
  • a bottle of hydrofluoric acid
  • a box of condoms, lubricated, with spermicide
  • a pearl tie-pin
  • a ruby ring
  • an opal
  • pa’a-shell cufflinks
  • the little thermocouples in the gas stove that detect whether the gas jet is lit, an the one that acts as a temperature sensor in the thermostat of the oven
  • a bottle of Buller’s Calliope 2004 Rutherglen Durif (wine)
  • my medicine — the anti-arrhythmic, the antihypertensive, the glaucoma drops
  • a recent Australian $100 bill.

Also, though I don’t have one, a carbon-fibre golf club. One reinforced with buckminsterfullerene (carbon nanotubes) would be even harder. I don’t have a rifle cartridge with a subcalibre AP bullet either, but there’s another challenge for you. Also, I can’t think of anything in the house that had to be annealed, though there must be something.

None of those things except the iPhone and the golf club is high tech, but thinking about the manufacturing and formation processes involved indicates problems for the approach that may apply even more strongly with high-performance high-tech gadgetry. Many of the examples are kind of frivolous things that you could say people don’t need — but they indicate manufacturing problems for a 3D fabber that could arise in things that people do need.

Edited to add

I thought of a couple of others that indicate specific difficulties:

  • a mousetrap
  • a smoke alarm
1 Like

There’s at least one earlier SF version, in three stories of the Venus Equilateral series, all published in 1945. George O. Smith was a serious engineer, but his stories were fairly light-hearted.

Clarke had read all of those stories as they came out, I believe; he wrote a preface for one of the collected editions acknowledging Smith’s influence.

In my vision of reasonably hard TL11, you can make the tonic water by printing the bottle, its cap, and a block of ice with CO₂ dissolved in it, as three separate processes. But that’s energy-intensive (or perhaps one should say cooling-resource-intensive) and takes time on the machine that could be making vaccines or solid gold humvees or something people actually want to pay for.

(I can’t remember the details, but there’s an explosive you can make by compressing carbon in just the right way. One reason it’s not popular is that any press powerful enough to make it can also make industrial diamonds, so making this stuff has a substantial opportunity cost.)

Similarly, you can print the steel blade, but it’s not precise enough to give you a good temper. It’ll put you in a good starting position but you really need to apply the heating and stresses yourself.

You can make the CFL, but not with the filling gas in place.

And so on.

That kind of exposes another hidden problem of this approach: Your machine may need to be able to solve the famous Problem of Economic Calculation. That is, it has resources it can use to print that bottle of sparkling water, all right. But each of those resources has alternate uses, and there are probably multiple ways it could go about the process of making the sparkling water, too. How does it decide which process to use? How does it enable you to decide how badly you want that bottle of water?

Perhaps all the printers on a planet are communicating with each other constantly, bidding for different raw materials. But then you just have to decide somehow how much your particular machine is allowed to bid for whatever you want.

1 Like

You can make a perfectly good explosive out of carbon black and liquid oxygen.

Cryogenic liquified gasses are another challenge for 3D printing, but I happen not to have any in the house right now.

Similarly, you can print the steel blade, but it’s not precise enough to give you a good temper.

I think you’d get a fragile sintered steel that would need forging as well as tempering, but the point is made either way.

It’ll put you in a good starting position but you really need to apply the heating and stresses yourself.

Right, but supposing that you aren’t making a sabre or a chisel. Suppose you’re making some sort of high-tech gadgets that needs millions of tiny precisely-tempered bits and pieces all through its structure.

You can make the CFL, but not with the filling gas in place.

You can make the case, the electronics (maybe) and the tube. But then you have to install the right mix of phosphors, and the only way the machine will do that is if it has stocks of all the necessary compounds on board, just in case anyone who ever buys one ever wants to make an archaic CRT or CFL. This goes further to the issue raised by the wine and the banknote.

Then you have to evacuate the tube, put in the right trace of the right gas to produce the discharge, and seal it.

It’s not going to be a make-anything machine. It’s going to be a very expensive way on making things that yet “require some assembly.”

The more you know and think about materials and manufacturing processes; about brazing, tempering, annealing, work-hardening, chemical surface treatments etc; the more carefully you consider all the different bits of apparatus the Machine will need to make different materials and shape different components; the more you think about the stocks of materials, pigments, dyes, dopes, phosphors, fragrances, flavorants, hardening alloys, strengthening alloys, magnetic materials, coatings etc. that it is going to have on hand; the more you think about how wasteful it would be to have all those stocks and all those apparatuses sitting idle nearly all the time because they are temporarily superfluous to current need, and when you consider the advantages — obvious to anyone who has thought about assembly lines — of having each of those processes operating in its own workspace and either moving the workpiece for successive operations or making components and delivering them for assembly; the clearer it ought to be that a make-anything machine is an entire economy in a box. It won’t be a thing the size of a fridge, it is be a thing the size of an industrial estate. And it would be perfectly capable of making hundreds of thousands of assorted article per day if you didn’t almost all of it capabilities idle at any particular time because you’re using it to make one thing at a time.

Can you, though? I have an impression that dissolved gasses come out of solution as water freezes, ending up as bubbles or migrating like salts to the un-frozen phase of the slush. And that gets to the point of the example: can you 3D-print a compressed gas, or do you have to first make a pressure container and then fill it. Does your nanofac have an array of tanks of compressed this and that, or does it have a built-in gas liquefaction capability?

Solid CO₂ in an ice matrix?

In my mental model you need at least liquid phase, and preferably solid, to make it work. You can make monoatomic gases by, er, opening the tap on the reservoir; polyatomic gases need the first atom to stay in place while you’re putting the other ones down round it, so in effect you’re making them as solids and then letting them boil away.

Ah! You’re thinking of assembling things atom-by-atom, making molecules as you go. I have serious doubts about that on thermochemical grounds and grounds of the stability of intermediates. If I were crafting molecules from supplies of elements I’d use chemical synthesis methods, with enzymes acting as my “nanotech”. In fact, I think that a “tap” and “tweezers” to assemble molecules would have to be catalysts, because you can’t make gadgets that small that aren’t molecules.

But in any case, my list of challenges above was for additive manufacturing imagined as some sort of descendant of 3D printing, working with materials as though they were continuous rather than delving right down to the molecular scale.

In the his essay Aladdin’s Lamp Arthur C. Clarke imagined assembling everything out of hydrogen and oxygen, re-arranging subatomic particles as required. I don’t think that that is possible, either.

CO2 hydrate clathrate is probably what you want, but it’s only stable under pressure. I don’t thinking your first monomolecular layer of the stuff will stay put long enough to let you get the second layer on.

1 Like

I was thinking of atom by atom, or molecule by molecule, too, as that’s what “atomic epitaxy” seemed to suggest. I can see the argument about the tools being catalysts, and in fact I’ve edited articles on catalytic chemistry that talk about building nanoscale structures with catalytic effects.

I am sorry not to have been more clear. What I intended to say in the second paragraph under “The Account of John H. Watson, MD” was that atomic epitaxy is useful, indeed necessary, approach for building/synthesising the picomechanical parts of high tech, it’s vastly too slow and expensive to be economic as a way of making the millimetre-scale parts of a device.

In the next paragraph i meant to go on to dealing with approaches based on 3D printing rather than atomic epitaxy. I thought I was clear and explicit, but since I have failed to communicate both to you and to @RogerBW I evidently wasn’t.¹

¹ When my first academic paper came back from the reviewers, Reviewer 2 had complained that a particular passage was unclear. I protested to my supervisor. And he said something that I have never forgotten: “If anyone says that they find your writing unclear, they are always right.”

1 Like

I can’t speak for Roger, of course, but I don’t think you were unclear. Rather, I was thinking about the specific question of whether you could build a block of ice with embedded carbon dioxide molecules to make it sparkle when it was defrosted, and about what sort of tools you would use to do manipulation on that scale. I had just bracketed out the whole issue of speed. I was more concerned with energetics, and was thinking about the idea that at cryo temperatures perhaps you could plunk down a carbon dioxide and have it stay in place long enough to be caged in water; and also I was thinking about the difference between “machine” and “catalyst” being largely erased at a small enough scale, as indeed you suggested. There is a difference between your reader being confused about what you said, and their having lost track of it while focusing on a different aspect of the question, and I think the latter is more what happened with me.

Meanwhile I see atomic epitaxy as it’s usually presented as being conceptually similar to 3D printing in that you’re shoving atoms or molecules into place and trusting them to stay where they’re put at least in the short term. This to me works something like the various “active” applications of scanning tunnelling microscopy, broadened to cover more materials and work faster.

(IBM xenon atoms on nickel, pre-2000)

But certainly you don’t use that for everything. Even for the basic “bits of plastic” market, a modern 3D printer costs more per part than injection moulding, and takes vastly longer; the main benefit is that you don’t have to have a run of at least a few thousand to pay for the tooling. so if you only want one or two and nobody else will want any it’s cheaper overall. And if (like me) you want to have a large catalogue of parts to sell to people, you can store your entire inventory in software and run it off as needed, rather than renting a warehouse to store unsold parts and indeed moulds. I see absolutely no reason not to assume that the same principles will continue to apply with new manufacturing technologies.

1 Like

Thanks for posting that picture, Roger: I remember when it was first published, and the conversations my friends and I had about it in the Mathematics Departmental Centre at ANU. I was thinking of searching out for my next essay on technology in Flat Black — “Nanotech is Biotech”. That exercise is exactly what I think of when I use the term “atomic epitaxy”.

I recognise the conceptual possibility of using that approach to reify a Clarkean “Organizer”. But I have profound doubts that it could ever be practically possible for all materials, nor within a long coo-ee of being economic, at least for anything big enough to see. You might use something like this to build the conjugated crystal-diode laser elements in the optical phased array emitter of a OPA laser pistol, but you’d never bother to use it to make the frame to micro precision or the stippling on the grip. When you don’t care that atoms are placed with an accuracy of a few tenth of a nanometre there have to be vastly, vastly, vastly cheaper ways of putting a few times ten to the twenty-fourth of them into place what might as well be a cheap injection-moulded plastic piece.

I doubt that you will be able to assemble any molecule you might need by putting atoms in place and trusting them to stay where they’re put. First, you won’t be able to get them there because of thermochemical issues. Second, they won’t stay put because the half-built things you have before you’ve finished a molecule won’t be stable. It’ll disintegrate or react with the wrong thing or cross-link where you don’t want it to. To assemble a molecule you’ll have to design a synthesis pathway for it, build special jigs for to assemble the reactants, then transport the precursor molecules to a special jig for final assemble, align them correctly, and (in some cases) either supply energy for the final assembly from ambient light or fuel, or carry off energy and reactive byproducts so that they don’t wreck what you’ve just made. In short, you will have to build molecules by designing custom anabolic pathways and building the conceptual descendants of enzymes, not the conceptual descendants of brick-laying robots. Think of tRNA programs for artificial ribsomes, powered with light or ATP. Think of growth media. Molecular nanotech is biotechnology.

Then as you say this is going to be ludicrously expensive and slow in proportion to the 1⅔ power of the versatility and the cube of the size of the product. You wouldn’t use atomic epitaxy to produce any of the things on my list of household items. And that’s when the challenges become relevant that they present to 3D inkjets that print thermoplastic resin and sintered metal.

There is nanotech assembling molecules in Flat Black, vast arrays of it. But it’s not sitting an a make-anything box in the workshop of a village craftsman. It’s in huge self-assembling solar-powered nanotech chemical factories on low-dev worlds. In other words, it is in the leaves of genetically-engineered perennial crop plats, in vast plantations.

1 Like

I am trying to get my friend Ralph Sutherland to join us here, but unfortunately he thinks he is busy with other things that matter more. Ralph’s first degree is in chemistry: when we met in the Maths Departmental Centre at ANU neither of us was a maths student. I understand (though this account might be distorted) that the Department of Chemistry, the School of Mathematics, and the Faculty of Computing all offered Ralph a university medal for his honours thesis.

A friend of Ralph’s who is an artist asked him for help in designing a three-dimension Sekrit Art Project that had a particular set of symmetries that Ralph calculated for her, then printed templates for the parts. But the parts couldn’t be assembled without a jig, and the necessary jig was just as complicated as the Sekrit Art Project. So Ralph designed the jig in twelve identical parts using 3D CAD software. Each part was too large for a hobbyist 3D printer, so he took the file to a place that does commercial 3D printing in plastic and asked them to make twelve for him. They quoted $1,200. “Nonsense!” said Ralph. “I could build my own 3D printer for that.” So he went home and designed and built his own 3D printer.