Trope-a-Day: Sliding Scale of Robot Intelligence

Sliding Scale of Robot Intelligence: All of it.  Much of the automation, autofac segments, and other component-type robots are bricks.  Utility spiders and other functional motiles are robo-monkeys.  More sophisticated robots, like the coordinating members of a pack of utility spiders, are idiot-savant average joe androids.  Thinkers and digisapiences are Nobel-bots, which puts them on a similar level to people augmented with the usual intelligence-augmentation technology.  And, of course, the Transcend, its archai, and certain other major systems qualify as Dei Ex Machinis.

This is, of course, complicated via networking (all those bricks and robo-monkeys are part of/under the command of more sophisticated systems all the time), the existence of systems which are themselves parts of other systems, and so forth, but is true enough for approximation.

FAQ Followup

And we have a follow-up FAQ. Mark Atwood asks:

Follow up question: how compatible are various worlds and polities nanofacs and slurrys? Polities that are colonies of existing polities will likely use compatible slurries and facs, but independent invention and/or long-enough separation in time will lead incompatible tags, inline data packages, and physical designs of nano-scale cages and gripping points. I can see things getting Interesting on worlds that have to deal and trade across polities with different nano, and interesting issues when trade fleets and military fleets with incompatible nano have to interoperate.

The answer there would be: for the most part, if you think of it as Internet software, you won’t go far wrong.

Most of the Worlds runs nanofacturing protocols that are cross-compatible and function according to the Imperial Nanofacturing Standard V.Whatever, IOSS1 somenumber through IOSS someothernumber, for the same reason as most of the extranet runs over IIP2; namely, it may be an Imperial standard, but at least it’s an open standard, and more to the point, it’s an open standard with plenty of legitimate places to plug in extensions and submit them for inclusion.

Even more to the point than that, it’s one with a lot of weight behind its adoption, because:

First, starcorporation-wise, just as Bright Shadow is pretty clear to its customers that its backbone runs over IIP and if you want interoperability, you can run IIP or built your own network gateway protocol, companies like Llyn Standard Manufacturing and Traders in Ideation make it pretty clear that they publish recipes that conform to the IOSS, and if you want to have your own protocol-format for recipes, then translating their recipes to work with your supply chain isn’t their problem.

And second, there is a huge database of free-to-use recipes out there, and by far the vast majority of them are INS-formatted, for reasons including longevity of publishing, a thriving open-development culture, and patent/copyright law that dumps expired, no-longer-manufactured stuff straight into the public knowledge pool. That that’s out there is a huge incentive for most ‘fac manufacturers to build machines that are compatible with it.

This even encourages worlds that invented the technology independently to work towards compatibility, obviously, something that’s made easier on the ego by the people who come around shortly after First Contact looking to grab any particularly good ideas they had independently to put in the next revision of the standard. 🙂

That being said, this is just like TCP/IP stacks inasmuch as when it comes to the core functionality, everything is swell and interoperable, but life may get interesting when one wanders off into more obscure corners, especially when people have interpreted things creatively or cut a few corners here or there. The further you go from basic mechanosynthetic applications, especially where gray-market, low-end, or from Those Companies, You Know The Ones ‘facs are concerned, the more likely it is that you’re going to end up having to contact your friendly local ‘fac-hacker to patch around whatever it is the manufacturer screwed up. Indeed, if you’re on some dark ‘hab out at the ass-end of the Shadow Systems, you’re probably going to need to get that guy out to make anything compile at all on your home-made sort-of-compliant lash-up system.

This is the level of problem that tends to hit most of those trade fleets, and so forth.

Most of the serious incompatibility issues are entirely deliberate – people who specifically don’t want to have access to those things, for a variety of reasons, be it straightforward economic protectionism (which makes even less sense than usual when you have cornucopias, but no-one said those governments were smart), keeping out evil Impie cultural imperialism as reflected in their Stuff, and/or fighting the War on Hedonic Pharmaceuticals Or Whatever Other Damn Thing It Is This Month by trying to prevent their citizens from printing out designer drugs, mass-driver pistols, or whatever other locally proscribed widgets they can download freely off the extranet.

(…which in turn the Agalmic Praxis Foundation, the Free Fabrication Fraternity, et. al., cheerfully subvert by writing recipes to get incompatible ‘facs to print out the needed parts to assemble compatible ‘facs, and so it goes on…)


1. IOSS = Imperial Open Source Standard. Which is exactly what it says on the tin.

2. IIP = Imperial Interweave Protocol. Looks something like IPv6 on steroids, with added relativistics and light-lag extensions, and using 512-bit addressing3 to allow for conveniently addressing individual elements of nanite swarms, etc. (With currently reserved option to extend to 1024-bit addressing just in case future requirements include addressing across multiple universes.)

3. For anyone wondering, this gives you up to 10154 addresses, which may seem excessive in light of there only being maybe 1080 protons in the universe. Apart from letting you feel comfortable using sparse allocation, I suspect the main reason for this is that at some point in IIP development, the engineers said the equivalent of “Look, guys, we have powerful processors these days and the routers can handle it. Let’s make sure we never have to go through another renumbering ever again.”

It’s FAQing Time!

Yes, folks, it’s that time again for the first time when I answer y’all’s background questions!

We have one question this month. James Sterrett asks:

What precursor elements do autofacs require for fabrication?  The same elements in the same proportions as the finished product (plus waste etc), or can they synthesize required elements?

Well, now, that’s an interesting question with quite a complicated answer, inasmuch as autofacs are rather complicated things in themselves..

Let me first suggest that this might be a good time to re-read Things That Make Things, since it covers a lot of the terminology I’m about to be throwing about.

So let’s start at the small end, with one of the most common working parts of an autofac, and which is also the core component of a cornucopia, including the ubiquitous desktop nanoforge, the portable nanolathe, and the specialized fabbers.

These, themselves, can’t synthesize elements, or indeed produce any other part of their feedstock – which is to say, you can’t just throw trash into them and have them rearrange it into what you want (you need specialized disassemblers for that, that are hardened to the job. Throw trash into a cornucopia, you have a good chance of wrecking the delicate internal components). They’re just glorified 3D printers. They’re absolutely dependent on a supply of feedstock, which is called nanoslurry.

(One exception to this is that you can also get what is called a nanobrick, which is basically dehydrated nanoslurry and formed together with a mass of simple assemblers. You use it together with a programming nanolathe for field construction, after mixing it with a suitable solvent, usually water, to form a nanopaste. But that’s not what we’re talking about here.)

Nanoslurry itself is a complex suspension of materials useful in nanoconstruction, designed to make it as easy and efficient as possible for nanofacs to pick out the bits they need. It comes in a variety of different kinds and grades, most of which are intended for one specialized industrial application or another. Standard-grade, which is what is shipped out as a public utility down municipal nanopipe systems, comes in two forms, informally referred to as “gray” and “green”.

The nanopipe you have plugged into your domestic cornucopia, for that matter, is actually a four-pipe system. The first supplies gray nanoslurry – which is water, long-chain alcohols, sulphur and nitrogen compounds, a suspension of iron and copper oxides, heavy metals, silicates, acetats, nanograins of industrial plastics, ceramics, and alloys, and prefabricated molecular components, or to put it another way, everything you might need to perform “common mechanosynthetic applications”. The second supplies green nanoslurry, which is specialized towards organic synthesis applications – what this means, of course, varies from world to world. And the third is the special-order pipe, which gets aliquots of specialized feedstock shot down it upon request, because while you may occasionally need, say, 2.1 g of technetium, it’s something specialized enough that there’s no point in including it in the regular feedstock.

(The fourth is the return pipe, that pumps what’s left after the nanofac has picked out what it needs back to the nanosource for recycling.)

And what the nanofacs need is, well, exactly what elements are in the finished product. (Plus a certain degree of in-process waste that ends up squirted down the outgoing pipe back to the nanosource.)

So, so far, we’ve just pushed the problem back to the nanosource; after all, nanoslurry doesn’t exist in nature, so it has to be manufactured. Which is what nanosources do: out of a variety of sources. Air mining, for worlds with atmospheres that have useful components. The bactries of chemical companies, refining volatile asteroid-liquor into useful chemicals with bacterial aid. Giant metal ingots shipped from smelters, which are reduced to slurry components. Reclaimed and purified chemicals from recycling plants and biocleaning cascades. In short, from the ends of all the conventional supply chains. Larger autofacs, like the Hive, will usually have their own nanosource(s) to produce all the specialized feedstocks that they want, especially since autofacs use a bunch of those raw materials elsewhere in their non-nanotech manufacturing processes.

So now we’ve just pushed the question back another level, haven’t we, to “can the people the nanosources use as suppliers synthesize elements?”

To which the answer is, finally: yes, but they usually don’t.

Nucleosynthesis is possible. There’s an entire engineering discipline, alchemics, that specializes in this sort of thing. But it’s neither cheap nor convenient, inasmuch as it still involves banging nucleons together and trying to get the wee buggers to stick, a process that tends to involve particle accelerators and nucleonic furnaces and isotopic separators and mucking about at absurdly high energy densities and low efficiencies. That said, it is now regular non-experimental engineering, and a large enough autofac might well include the equipment.

…but economically, it is almost always cheaper to dig the stuff up and have it shipped to you for nanosource processing than try to manufacture it on site from other elements. Nature’s production process may be slow and uncomfortably explosive for anyone within a couple of hundred light-years, but, damn, does it have economies of scale.

This effect is only amplified, of course, by the fact that alchemics equipment is also what you use to produce gluonic string,  muon metals, and various other kinds of exotic matter that genuinely don’t occur in nature anywhere. Now that’s what you call comparative advantage!

Things That Make Things

Since we’ve just passed the Matter Replicator trope, and since it may be relevant to an upcoming FAQ question, I thought I’d throw out some definitions relating to such things that may make things clear. Well, clearer.

nanofac is the basis of nanofacturing technology. Think of it as essentially a 3D printer which can handle arbitrary molecular components with single-atom resolution. (It doesn’t have to: a lot of the time it can simply place pre-assembled multi-atom components picked out of its feed, but the point is that it can.) While it can use free-floating assembler nanites as part of its operation, the vast majority of the work is done in a supercooled vacuum chamber by an array of distant descendants of the atomic force microscope. The materials supply it needs is fed to it as a suspension of molecular components called nanoslurry available in a variety of forms, supplied as a utility from a central nanosource that makes the stuff from raw materials and recycles the return feed of all the stuff that the nanofacs don’t use.

Most important to note is that a nanofac is not a discrete thing you can buy itself – it’s just the term for the central construction array as a module.

What you can buy, on the other hand, is a cornucopia, which is a general-purpose construction device that comes in sizes ranging from desktop-printer-sized (the ubiquitous nanoforge) to dishwasher/fridge size. These are common household, etc., appliances, packaged as vending machines by companies like Valuematic Vending, and are basically a user interface/power supply/etc. wrapped around an appropriate nanofac. They can make pretty much anything you can describe in a recipe, or conceptual seed, to give it its formal name, although if it’s something too big to fit into its vacuum chamber what you’ll get is a heap of parts over several runs which you have to assemble manually following the v-tags after you get them out. (They may or may not bond permanently once you do this.)

specialized cornucopia, on the other hand, is a fabber. These exist because in nanofacturing, there’s essentially a scale with versatility at one end and efficiency at the other. A cornucopia is a magical device that can make everything, but isn’t the fastest or most efficient way to make anything in particular.

So there are fabbers, which trade off that ability for greater speed and efficiency and customized user-friendliness in doing one particular thing. So while you want a cornucopia available to you, certainly, what you want in your wardrobe is a clothing fabber, in your kitchen is a food fabber, in your sickbay is a pharmafabber, in your wet bar is a cocktail fabber, etc., etc.

And finally, it’s worth noting that assembling things atom by atom, or molecule by molecule, is not actually a terribly efficient way to do things in the first place. It works fine for small objects, sure, where the convenience outweighs the inefficiency, and especially for those made out of lots of tiny components with fine detail to assemble. But large things, especially large things with large areas of relatively homogeneous structure, you really don’t want to make that way.

Which is where autofacs come in. An autofac is a automated assembly system that contains an array of nanofacs for making individual detailed components, but which also contains lathes and drills and presses and kilns and extruders and all manner of other macroscale manufacturing-process equipment, along with plenty of motile robots whose job is to do the assembly of all the different outputs of these processes into the end product. (So they take in nanoslurry for the ‘facs, but also metal ingots, ceramic powder, plastic granules, etc., etc., as their raw materials.)

These vary in size from the relatively modest autofacs you’ll find in most neighborhoods, belonging to companies like Ubiquifac, whose job is to construct large goods people have ordered on-line at a point relatively local to them for immediate delivery, up through larger factories – such as the ones that take nanoslurry and sheet metal in at one end and have finished vehicles drive out the other – all the way to truly giant many-square-miles really-can-build-anything complexes like the Hive.

Trope-a-Day: Matter Replicator

Matter Replicator: The cornucopia machine or autofac, which can build matter into pretty much any object that you want and have – or can write – a recipe for.

Sadly, they are required to obey the laws of thermodynamics and conservation of mass-energy.  They also tend to incorporate – especially in larger models designed to build larger objects – arrays of specialized nanofactories and macro-scale tools, and require plenty of energy and specialized appropriate feedstocks for whatever it is you want them to build (so mining, refining, recycling, bactries, and the rest of the industrial supply chain haven’t gone away quite yet).  You can make them increasingly general-purpose in these areas at the cost of greater inefficiency – field autofacs are a lot less elegant and more energy-hungry and expensive to run than standard household models.

Living things generally have to be grown in a medical vat instead; simply because most of them tend to die when only half-printed.  Yes, this is exactly as gross as it sounds.  (Also, some dead organic matter – well, let me put it this way.  While you can print up a steak in an autofac, steak is still made in carniculture vats, because first, self-replicating steak is cheaper, and second, it gets boring eating the exact same steak hundreds or thousands of times.  Similar although aesthetic considerations are why vatwood is generally preferred to directly replicated wood – although vatwood planks are seen as input to larger autofacs.)

Nonetheless, they’re more than good enough for post-material scarcity purposes.

Trope-a-Day: Eternal Engine

Eternal Engine: Quite a few of the Empire’s – and other advanced civilizations – larger factory complexes (mostly in space, to avoid heat-dissipation and environmental problems, but including things like most of ecumenopolitan Qerach) are like this, including city-like size.  The interiors are not particularly friendly to biosapiences, mostly because they’re very, very automated, so they’re not actually designed with the thought that anyone might go in there in person in mind.

Manufacturers

The Llyn Standard Manufacturing autofac, informally known as the Hive, sprawled over a hundred square miles of Seléne’s surface, a vast complex of industrial machinery stacked upon more industrial machinery, gleaming in the crystal vacuum and the harsh light of its floodlamps.

To the north, a ruddy glow mixed with the floods’ blue-white, where a thousand furnaces and smelters turned shipments of raw metal and stone coming in from the asteroids into bar stock and other materials for the inner manufactories, secondary forges pounded, cast, carved, and drew the purified metals into thousands of gross components, and more specialized factories spun stone into specialized clays, ceramics, glasses, and the wafers from which nanocircs were cut.

Off to the east, a tangle of pipes and tanks surrounded the bactries, where volatiles brought downwell from the outer system were fractioned, refined, and fed to reactors containing myriad industrial catalysts, fabzymes and genetically engineered maker cultures to produce a million different chemicals, all the feedstocks necessary for all the industries the complex supported.

In the south, the triple containment buildings of the power plant dominated the skyline, housing three of the system’s largest fusion reactors, gulping deuterium from the buried slush tanks at their feet – mere buffer tanks, kept constantly topped up by a stream of automated tanker-ships coming in from the gas mines of Melíeré; and to the west, the mass-driver launch complex which delivered containers full of any of the autofac’s unthinkable array of finished goods and modular components to any world, hab, or drift in the system rose like a mountain.  The warehouses around their feet were a mere scattering of toy building blocks by comparison.

And within this ring, the heart of the autofac: factory after factory, specialized tooling, nanofac growth chambers, and robotic final-assembly plants, and the thousands of pipelines and conveyors connecting them – a crowded collection of plain geometric cubes, geodesic domes, and polished spheres, in the simple ascetic style favored for those areas not intended to ever be inhabited, or to be more than rarely visited.  Scattered among them, vehicle garages and robot hotels housed and tended to the automation, the driverless trucks and frenetic utility spiders that scurried throughout the complex, carrying its lifeblood and tending to the machinery.

At the center of the great autofac, a single tower rose above all these buildings, its lower floors containing the hosts for the artificial intelligences that ran the complex, and its uppermost level housing the operations supervisor, Lilse Varenna-ith-Varenti, and his dozen department heads – the only sophonts anywhere within the Hive – reclining, eyes closed, in their command chairs.

Bodily functions shifted to autonomic maintenance, minds vastened and placed in synnoetic AI-symbiosis, and senses filled with input streams gathered from sensors, they did not run the complex.

They were the complex.