« Holding pattern ... | Main | Happy Birthday »

"Where do you get your ideas?"

One of the perennial questions that I (and every other SF writer) get asked is, "where do you get your ideas?" Harlan Ellison once famously replied, "Poughkeepsie", but the truth is a bit harder to swallow: ideas are all around us, ripe for the plucking, and usually all it takes to come up with something wild and new is a magpie's eye for the bright and shiny, and a willingness to bang concepts together until they stick in a new, unfamiliar, agglomeration.

Unless, of course, you're a visionary like Bruce Sterling.

I've just been reading a slim, non-fiction book by Sterling, titled Shaping Things. This book is not like his other books. For one thing, most of them are novels; for another thing, this one is serious. You can tell it's serious because it was published by the MIT Press, and, for another thing, because it made my headmeat explode. In fact, I am deeply annoyed that it took me so long to get around to reading it (it came out in 2005, around the time I was writing Halting State, a novel that might have been subtly different had I been up to date on my to-read bookcase).

To summarize: designers are the unacknowledged legislators of the human condition insofar as they design the objects that populate our environment, and we are tool-using, object-wielding, primates. The history of human-made objects has evolved through a series of epochs: artifacts, machines, products, gizmos, and spimes. To provide examples: flint hand-axes and carpenter-made rocking chair are artefacts. Rifles and reciprocating engines are machines. Model-T Fords and Coca-Cola bottles are products — of a mechanised culture. The iMac I am typing this on is a gizmo, or perhaps an embryonic spime. Each step represents an increment of complexity, and each category of designed object embodies and encapsulates the functionality of its predecessors while adding more stuff.

Spimes move beyond gizmos like today's smartphones and laptops because they primarily exist as a software representation; when you want one, you request (or buy) a physical instantiation of the design. Spimes rely on improvements in distribution chain management technology (such as computationally active — not passive — RFID chips with GPS positioning and time-binding capability) to monitor their own progress, order supplies and maintenance work, notify their owner when action is required, and, at the end of their life, arrange for their own collection and despatch to a suitable recycling point. Spimes are the logical ouput of a logistics infrastructure based on fabbers (primitive versions of which are just now dropping through the price threshold defined by the first primitive consumer laser printers in the mid-1980s). They can be anything from a passive object like a chair (but a chair that knows who it belongs to and where it is and how to disassemble and recycle its parts) to a jumbo jet, by way of an iphone — the iphone is dangerously close to spime-hood already — but the key insight is that they represent a whole new way of thinking about the not-entirely-post-industrial society we live in.

Around this concept, Sterling has built a gigantic, beautifully articulated vision of the role of design in shaping our environment in the near-term future — one that makes sense. There is clearly something old and dangerously creaky about a supply chain, in the age of expensive energy, which relies on shipping raw materials to factories, processing them centrally, then shipping finished products to consumers — rather than relying on consumers gathering raw material locally and zapping them into the finished product via a fabber.

This book is only about 150 pages long. And I only finished it 48 hours ago. And I was already familiar with many of the ideas in it. But it has already injected the fertilized eggs of at least two kick-ass short story ideas into my neocortex, where they are already incubating, and I can already feel it warping my sense of where my next novel (after the current one I'm writing) is going, because of the sheer visionary drive of Sterling's cognitive engine which is fine-tuned to ingest any and all design-related raw material and build weird new mental structures from it.

<irony>Please don't buy this book, especially if you're an aspiring science fiction author, and especially if you're trying to wrap your head around the near future. Because I'm a selfish old fart who can only watch Sterling's output with mixed awe and envy, and I don't need even more competition.</irony>




And don't subscribe to the RSS feed for his blog. Nothing interesting ever happens there.


And here I was, completely convinced that Shaping Things influenced you in The Halting State.


oooh don't get me started. oh too late? i've just come from a seminar on prehistoric distribution and spheres of exchange... part of a term of 'technology with in society'
Why is a hand axe merely an artefact: an object intentional modifed by a sapient being?

and when do the production of them become a technological process with standardisation?
http://en.wikipedia.org/wiki/Levallois_technique she niggles (not the best explanation sorry)

At any of your levels of 'epoch', I don't think designers or users really understand their things the way they think they do?

The attitude that people have to things in pretty strange and is very old. http://journals.royalsociety.org/content/u42h172302468181/
How differently are people going to cope with the explict knowledge of the inter-relationships between their things? Do I want a disappearing chair if I don't know if I want it to dissapear? what would I really do with a cornicopia machine, how long have people been doing that?

What's quiet fashionable in some parts of archaeology is the idea that artefacts (either iPhones or hand axes) have agency [they tend to ignore the structure bit]. That is they affect their owners [kind a like feline overlords] and influence available choices.
If you treat things as weakly person-like, will you make them more personlike? and how will you react when they do answer back [talking paper clips excepted]?


I saw the two lectures he did last month on Kosmopolis, one about Ballard and one about Hypermedia and Literature, and he was brilliant! I think I'll buy this book (don't worry, I always buy yours too).


Just how complex and large a 'fabber' do you need to take sand as a raw material and create an x billion gate chip to provide your spime with intelligence? Let alone creating your GPS receiver.

If you know where I can get one for £2.50 I'd be interested, otherwise there's a tiny flaw in the idea.


I saw Bruce do the keynote at a conference somewhere or other (might have been South by Southwest a few years ago) where he talked about his theory of spimes. I came out of it feeling halfway conflicted. Half of me thought: wow, this sound cool, but it's all rather theoretical at the moment. When it happens that'll be great! The other half thought: what? this is daft. And then I felt like someone in 1981 saying "what do you mean everyone will be connected all the time? that's ridiculous!"

The big problem with the spime stuff is that it's not concrete. You can't do anything with it except say: when that happens, it'll be great. It's (and excuse the unintended disparaging tone) science-fiction. Fun to read about, but I can't draw a line between "me right now" and "a world with spimes pervasively in". I suppose this is almost because it's *too* real; it feels like it really could be the case and just isn't yet and there's nothing I can do about it. I can't draw a line between "me right now" and a world with laser pistols or teleportation in it either, but I don't feel the lack there, because they're obviously sci-fi. Spimes are sci-fi too but they don't feel like they should be. Or maybe I just don't want them to be; I want them to be real. Almost close enough to touch, but not quite; not yet.


I must confess I've never had any shortage of ideas. Over 25 years of trying, I've realized that I do not have the discipline to go from idea to finished story, though I did publish one short story and a couple of poems along the way. Ideas to me are easy. Making the characters come to life in detailed scene after scene -- that's hard work. As they say, most of us don't want to write, we want to "have written".


His pal Gibson also does this for me. More in the take-this-social-phenom-and-twist-it-35-degrees style, but similar. Everything sounds so familiar, and yet is so fresh.


You should think about starting a branch of this site that is similar to Michael Swanwick's "Unca Mike's bad advice" .
It couldn't hurt to have an "advice column" designed to cut aspiring science fiction authors off at the knees.


I don't particularly write a lot, but every now and then I'll write some lyrics for my blues band ( http://myspace.com/thebluesberriesat - new demos coming soon! ).

I'm not saying I'm a good writer, but I figured that I knew more or less exactly where my ideas come from, so I hope you don't mind if I share.

My ideas usually come from listening, usually in a bar, until I hear something that makes me say "that sounds like a great blues line" and going away and writing it. The last thing I wrote was "Never Put Water In Another Man's Whiskey", although a favourite title of mine was an earlier non-blues thing I wrote called "Three Quid and Jesus" which I came by in exactly the same way. The conversation went something like this:

"We had the Jehova's Witness round the other day asking if I'd found the lord."
"I always ask if they've looked down the back of the sofa. Isn't that's where you usually find things."
"Yeah, I found three quid and Jesus..."


When asked about creativity, Leonard Cohen said: "If I knew where I got my ideas, I'd go there more often."


Just by way of reassurance, Charlie, I have no plans to compete with you, no matter what groundbreaking works I should chance to read.

Back around 1990 or so, Larry Niven pointed out that most people who say they want to write really mean that they want to have already written. This resonated with me; sure, I enjoy and appreciate a good book, and I wouldn't mind basking in the adoration of the groupies, but no way do I feel like working that hard. So, you can thank Larry Niven for my utter lack of literary ambition, if only because of the way in which he helped me to confront and embrace my inner sloth.

all best as always,

Robert A. Ogden II


Artefacts are spimes except that the information is currently kept in our heads (or on random bits of 'paper').
The desire to extract and realify tacit information is scary. To do it successfully will be really hard and then to have to deal with weaky-person-like artefacts will really change the nature of society. So is it likely?
Its bad enough fending off vendors of free newspapers and broken office chairs without them hanking your twitter profile. Ooops; and how do you train up (aculturate) the stuborn, last generation's implimentation of users. I guess that's why Stuart sees it as SciFi
So your list of epoch, artefact, machine, product, gizmo... has as muuch to do with 'marketing' as technological complexity. This list is not chronological neither is it teological.
I know I am being a spoil sport, but I don't believe it is as simple as all that.
So good luck with the short storys, I look forward to reading them.


(1) Ideas are easy. The difficult bit is _dialogue_.

(2) re Sterling, fair enough, but how do we separate the signal from the noise? About 10% of what Buckminster Fuller, Marshall McLuhan and Jane Jacobs (inter alia) banged on about is coming very true . . . the problem was, at the time, who could tell which 10%? Yr plausible futurologist can hit the nail on the head pretty often: the problem is all the other busted thumbs that we forget a few decades down the line... Domes, yes: Dymaxion, no, ta.


I found "Tomorrow now" much more entertaining for a serious Bruce Sterling book.



Not bought any Sterling for a while - but did re-read a bunch of stuff when I turned my shed into my library. Amazing what gems a re-org throws up.

Bearing in mind that I've not read "Shaping Things" - one thing I will say is that I find his work very interesting but not necessarily convincing. Of course, if you read and enjoy a writer's work of fiction you are to an extent entering their mindspace and you are convinced by the authenticity and - erm - referential integrity of their vision but... He don't fink like me fink. Njoy but not inspire.

I grok that you grok the spime. I don't. I find the world of fings absolutely fascinating. I can waste a lot of time reading about hardware and software on Wikipedia, I enjoy my Screwfix catalogue but fings is fings. The term spime is simply special pleading for a particular group of fings. Their provenance is interesting, but they are still fings.

And there is an inherent danger with this technology. If all you have is a hammer, all problems look nail shaped, however clever and versatile your hammer might be.

And I do agree with the posts mentioning marketing. There are some false distinctions, I feel.

I mod therefore I am.


I, too, thought you had been influenced by Sterling's Spime narrative soon after it came out. Now that I know you hadn't (directly), I am just confident that your idea-spinning will improve.

In any case, you shouldn't worry about competition from Sterling; most of his fiction is a little irritating.


Speaking as an aspiring sf author - so many other places to go that are equally as, if not more interesting and inspiring.


I was blown away by "Shaping Things" when it came out and it still shapes my thinking. Sterling's blog every so often has items related to the implementation of those ideas listed as "spime watch".

IMO, while Sterling can be a great inspiration for ideas, his science fiction has gone nowhere after "Holy Fire" and the best, most inventive, work of his that I ever read, "Schismatrix", is over 20 years old (updated as Schismatrix Plus in 1996).

Fortunately Charlie, you seem to write well in both fiction and non-fiction - I particularly liked "Shaping the Future" which I still refer to when the subject of "whatever happened to X?" comes up.


I cam across this book a couple of years ago in my local library of all places. It's pretty amazing stuff, that's for sure. Local hero makes good, eh?


My favourite response to the idiotic question "Where do you get your ideas from?" was in an interview I saw somewhere with (Sci-fi artist) Chris Foss:

"I just take photographs of them as they fly overhead."


I don't want to collect the raw materials to make my own furniture, nor for that furniture to develop any form of intelligence whatsoever.


Apropos @22: rolls eyes.

Martin, Maggie, just go read the book. My attempt at summarizing a 150-page polemic doesn't do it justice.


I've been commenting in various places on my NaNoWriMo efforts.

The basic idea is to write at least 50,000 words in November.

I'm within 4,000 words of the target,

The plot so far has included Nazi sexret agents, fights on the boat train from Harwich, giant airships, parasite fighters, nests of pirates on the chinese coast, and an embarrassing revelation about the activities of the Imperial Japanese Navy

And now the circus is in town. It's going to be one big climactic action sequence, if I can just get the images down on paper. Clowns, knife-throwers, and the flying trapeze.

I'm never going to be able to pull this off.

But the thing about NaNoWriMo is that you have a target, and 50.000 words isn't a huge thing


Charlie: I read Shaping Things last year, and I'm still trying to get my head around the implications and the potential unintended consequences (the basic stuff of sf world building, after all). One thing that's clear is that spimes don't enforce a particular topology for distribution or recycling: it can all be local, partly-centralized, all distributed over various geographic areas, or some mix and match of those. Not only that, the raw materials could be part of a kit of standard materials which is itself a spime, as could the parts of an assembled chair. If your fabber has bins for parts and material, and a robotic arm or other delivery system, all you have to do is plug your kit into the bin system and the fabber will choose the correct materials, email you if something runs low or if the new spime it's building requires something you've never stocked (or maybe order it, if you've given it the authority to do so), and accept recycled material that's easy to use in assembly when other spimes reach end-of-life.

The intelligence doesn't have to be in the physical product, e.g., a chair. All the chair needs is the ability to identify itself uniquely, and to record a time-stamp for certain events (and it needn't have that, but it probably would be more convenient). All the storage and computation can be online, to be accessed by a watchdog process that wakes up whenever the RFID on the chair moves or is deliberately scanned for some reason.

Some parts aren't going to be constructable by personal fabbers for quite awhile; this includes memory and processing chips. That's a reason to keep the computation on the physical product minimal, but it doesn't have to be removed completely; it just requires that the fabber use cheap, standard parts it can program for the application and insert into the chair.

Aside from all the "green" reasons why spimes are a good idea, why would we want them? Having software keep track of the wear life of the mechanical parts in your personal vehicle, and scheduling purchase and replacement of new ones would be handy. Or replacing the perishables in your first aid kit on a regular schedule (I really suck at that, myself). Or upgrading your TV set to the 3D standard when it becomes available.

Essentially spimes are a whole new way of getting, maintaining, using, replacing, and throwing away your possessions; one that allows easy access to responsible use of resources without having to spend all your time at it. Green without the hairshirt, I call it.


Oh, bother. That post started me thinking about security, and I realized that every spime needs a minimum amount of processing to support some basic authorization and access privilege controls. Some spimes will need a lot more: a door lock, for instance. Not much point in a lock if someone can write to its RFID to tell it that it's been recycled and shouldn't lock anymore. And yet another reason why spimes are useful: you can have the security parts replaced when they've been compromised, all without ever reading a crypto journal or a Risk SIG's newsletter, or even knowing about it for longer than it takes to hit a "y" key response to an automated email (which has been authenticated via credentials you accept, of course, otherwise phishing gets a whole new dimension).


I've got to disagree with you on the future of supply chains.

There's always been a tension between centralisation and decentralisation in every production process, from crofting to making x86 processors. It is amazingly cost effective to put four and half thousand shipping containers on one ship and amazingly time effective to be able to get a package across the civilised world in 24 hours. Given those two factors, decentralised production has got to be very, very close in price, time, ease and energy consumption to beat our current plan of just making everything in one big factory in China.

(And yes, the price of transport fuel may peak, but the cost of transport fuel is only a small part of the overall cost of the energy used to make and deliver a product. If anyone wants to claim that it'll become uneconomic to ship something from China to here, then they'll first have to explain how it's still economic to manufacture anything at all.)

For example, yes, we've all got laser printers at home. But big printers still make up 90% of the market because the cost per page of offset is still far less than the cost per page of laser.

Add on to that, for some goods, the thermodynamics of production favour economies of scale. If I just say steel, aluminium, and cement, then I've probably covered 50% by value of our manufacturing, and 90% by mass and environmental impact. For each of those, bigger is better, to the point where NZ's own steel mill, making 650,000 tonnes per year, is positively boutique. Yes, someone will be selling some funky machine to make one-off gubbins in your desktop from steel powder, but the steel still needs to be made in the first place.

I don't see decentralised, desktop manufacturing taking over the world. Yes, it'll be huge. Yes, it'll be cool. Yes, I want one myself (selective DPSS laser sintering, please). But I doubt it'll make up more than niche products and I think that the overall environmental impact of manufacturing isn't going to be substantially changed by how we make these niche goods.


Here's a link to Sterling's 2004 SIGGRAPH speech in which he talks about SPIMES:



Having been born in Poughkeepsie (NY, USA) and done most of my growing up in that area, I'm fairly sure Mr. Ellison was, um, exaggerating (or talking about a Poughkeepsie I've never heard of). I mean some areas of the mid-Hudson valley can be quite beautiful in certain seasons, but as far as Poughkeepsie itself providing ideas or inspiration for science fiction stories? Nah, sorry, that's way to unlikely for me to be able to suspend my disbelief.


Bleyddyn: I have this word for you to look up: "sarcasm".

(Not to be confused with the portmanteau neologism, "sarchasm". As in, "oh look, he seems to have fallen down the sarchasm.")


#11: Oh, that Cohen line is great.

It's been printed out and stuck outside my cubicle.

According to a Cory post on Boing Boing, Sterling announced the end of the Viridian movement today.


> The history of human-made objects has evolved through a series of epochs: artifacts, machines, products, gizmos, and spimes.

So what would be the next item in this progression?

Maybe it's entities that can design objects, including improved versions of themselves. Which might lead to a version of the Singularity which does not require true AI?


@22 - It's not about what you want, it's about what the furniture wants. ;-)

Incidentally, another good link to throw in here is Jamais Cascio writing on Malware for Materials.


Bruce @25. Once you put the storage on-line somewhere, you have the situation where Megacorp know exactly what underwear you are wearing today, who else's underwear it's been in close proximity with recently, which ointment tubes you've been...

I'm endlessly fascinated by the way we get all excited about some losses of privacy (the destination of our IP packets, say) and yet seem far more relaxed about far greater ones.

I'm pretty sure I'd rather the Government could read all my email than that the 17 year old across the road who works for Tesburys knows which particular combination of objects were in close proximity in my kitchen last night.


...but as far as Poughkeepsie itself providing ideas or inspiration for science fiction stories? Nah, sorry, that's way to unlikely for me to be able to suspend my disbelief.

Surely that's because Harlan Ellison has exported all the ideas to California, leaving behind... um... er... the prison? (my only knowledge of Poughkeepsie being from The French Connection).


Charlie I really thought you would already have read Shaping Things but anyway. Two brothers from New York are beating a murder rap because their transit card knew where and when it was used and they also used a automated payday machine which photographed one of them, story from the New York Times. They are going to beat an eyewitness identification. Now I just need a cardboard cutout and a magnetic card read/write machine and I will be able to establish an alibi that really rocks. Lookout snitches!


Although I like Mr Sterling's work I'm afraid he's far too enthusiastic for my tastes.

When you actually have to deal with RFIDs you discover that they're not nearly as good as people make them out to be - it's all down to receiving signals and that dratted inverse square law.

Suffice it to say that I don't think we'll be getting every little thing RFID tagged _and tracked_ anytime soon.

p.s. @37 Please post a link so we can decide the issue for ourselves, your tone is decidly biased sounding


Nick @ 35: I think judicious use of encrypted storage and careful (open source, reviewed) application design would keep most information private. It's never been possible to remain private from someone who really wants to know about you specifically and is willing to spend the time, energy, and money necessary to dig things up. But it is possible to keep your privacy against the average hacker, and against the average marketing organization. As long as the government doesn't have all the keys to the Internet's back door (and at the moment, they have some of them) it's possible to protect against them in the sense that you can make it very expensive to dig up and make readable the information you most want to protect.

Think about wireless networks. The only reason wardrivers have so little trouble breaking into them and reading the information being transmitted is because most people haven't a clue about network security, or can't be bothered to implement it. My network is closed, and has access limited to only the physical machines I actually use. Yes, it's possible for that to be cracked and/or spoofed, but very few people have the capability.

Most of the government's, and some of the corporations', methods of information-gathering are passive and indirect; if they want to know what's inside your house they don't have to break in because they have IR cameras and teraherz radar. Spimes needn't give them any more information than that.


Never mind, I found a link:


Read it and make your own mind up


Robin @ 38: Have you been following technical developments in the RFID field? I had to do some research recently in support of a manufacturing/resupply system, and was rather surprised at the amount of work that's been done on them that hasn't had much public exposure. There have been advances in readers as well as in the tags themselves. The tags have gotten smaller, but not lost much range due to using higher frequency RF, and the readers have gotten very good at quickly reading large numbers of tags in a small volume. I expect a lot more work to be done on them in the next few years; the stated objectives of the manufacturing and sales industries have not yet been met, but they probably will be.


Add on to that, for some goods, the thermodynamics of production favour economies of scale.

Yes, but that economy needs to cover costs and make a profit. If I make myself a plastic hair comb using a 3D printer, my cost is just the cost of feedstock. (Assuming the machine has already paid for itself, etc.)


Sebastien @ 42: Distributed manufacturing will be most beneficial to the production of low-volume goods like fancy combs, yes. However, it's not going to be much help for making the plastic feedstock itself. That's going to be much cheaper to make in one big factory than in many small factories.

Similarly, for steel, distributed manfacturing has been tried before, the backyard steel furnaces of the Great Leap Forward. That was a collosal failure for several reasons, but a major one is that small hot things have higher surface area-to-volume ratios than big hot things, so they lose more heat and need more fuel, per kilo of output. That's why both polar bears and steel furnaces work best if they are big.

Thermodynamics trumps politics (and wishful thinking) every time.


Sebastien @ 42: Answering your point on the economics, your cost is the feedstock and machine. A big factories cost is the feedstock, machine and transport.

They buy feedstock by the tonne from industrial suppliers. You're probably buying by the kilo from retailers. That's a win for the big factory.

They amortise the cost of their machine by running it night and day, producing squillions of units. You amortise your machine over the units that you produce for yourself and friends, when you're not busy doing other things. That's a win for the big factory.

Yes, they have to pay transport. Transport costs have plummetted over the last thousand years and show no signs of stopping. What's percentage cost of shipping an iPhone, in bulk, across the world, compared with it's retail cost? Approximately zero.

Okay, oil crisis => oil prices treble OMG!!! we're all doomed!!! Well, approximately zero times three is still not very much at all, so we'll still be making (most) stuff in China.

So for these economic reasons, I don't see distributed manufacturing taking over the world. It'll do niches quite nicely, and there's lots of niches, but overall? No big deal.


charlie @37, you sound like you think they're guilty. Does this mean you believe an eyewitness (studies show witnesses tend to be exceedingly inaccurate) rather than the electronic track?


Jez @44: Word. And if push comes to shove, the reason sail-powered ships went out of fashion wasn't that they were inefficient, but that they couldn't use either of the newly-opened Suez or Panama canals.

Apropos of maggie's comments, I'm really interested in the overlap between supposed epochs of "things". Is my kayak still an artefact, even though it's made in a factory and gets carried around on top of a car? If I made a kayak out of bits of driftwood, would _that_ still be an artefact? The categories seem to reflect the economics of objects rather than their capabilities, so the spime is actually an odd fit there. Yes, yes, will read book...


Leading on from 42 and 44 etc:

The other thing people forget about is time. My time actually has a value (to me at least).

So to use Sebastian's comb example; even if the feedstock were to cost fractions of a penny, chances are that it would be cheaper for me to buy a mass-produced cheapo comb than bugger about for even 5-10 minutes setting up my 3D printer.

Even on the UK minimum wage a 50p comb (and that's an expensive one) is equivalent to less than 6 minutes work!

Time is the big killer with these things and it is the big payoff. Nothing is "free" that you have to spend time setting up, researching or otherwise buggering about with.

For this reason things like Reprap ( http://www.reprap.org/ ) are, for now and the forseeable future, the preserve of hobbyists who ENJOY messing about with arcane scripting languages!

Manufactured objects are so cheap these days that thier material costs are trivial in comparison to the time saving element.

The killer designs of the future will be those that save people time - not material costs.

Actually this is already the case: when was the last time you or your partner made a main meal totally from scratch without packet sauces or some other pre-prepared ingredient?


Brude@41: we were tracking people wearing RFID tags with 2 year old tech (as of now) as they went in and out of 'choke points' (sorry, have to be a bit vague for IP reasons) and we got a lot of missed reads.

We had to completely replace aerials for v2 to get the acquisition rate up to about 90% - and thats over a 2-3 _meter_ range.

When I read Mr Sterlings 'Brasyl' with it's city full if RFID tags being tracked by kilometers up observation platforms I thought he was being rather optimistic...

I think RFID would be great in a manufacturing environment - fixed ranges, no squishy radio absorbing/reflective bodies in the way, you can blast the aerials up to full power without worry abot health and safety, etc, etc.

For use in meatspace they're going to have to get a lot better.



As in the wonderful essay by Ursula le Guin: "From Elfland to Poughkeepsie" published in her collection of essays: "The language of the Night"

Which DOES deal with .. Where do you get your ideas from?


Robin @49: you're confusing Ian MacDonald with Bruce Sterling. (Try not to do that, okay?)

Yes, there's a tension between centralized mass production (with economies of scale) and distributed one-off production (with bespoke customization). We're going to need the former for a very long time, if only to provide refined input materials for the latter. But I suspect most people, given a choice between an off-the-peg pair of generic jeans from China and a pair manufactured by fabber to their personal measurements, will take the latter (once the price premium drops below their threshold for discretionary spending).

Think back to Henry Ford's dictum that you could buy a Model T in "any colour you want -- as long as it's black" and then consider the customization options on just about any modern car. Once the basic need is satisfied, people want customization.


On another note, here's Bruce Sterling on Steampunk:

Steampunk's key lessons are not about the past. They are about the instability and obsolescence of our own times. A host of objects and services that we see each day all around us are not sustainable. They will surely vanish, just as "Gone With the Wind" like Scarlett O'Hara's evil slave-based economy. Once they're gone, they'll seem every bit as weird and archaic as top hats, crinolines, magic lanterns, clockwork automatons, absinthe, walking-sticks and paper-scrolled player pianos.

We are a technological society. When we trifle, in our sly, Gothic, grave-robbing fashion, with archaic and eclipsed technologies, we are secretly preparing ourselves for the death of our own tech. Steampunk is popular now because people are unconsciously realizing that the way that we live has already died.


Marilee @45 I have no idea if they are guilty or not, I'm guessing they aren't smart enough to game the system so they probably are innocent. My point is that every advance in tech is an opportunity for everyone who is intelligent to subvert the system. The two edged sword cut both ways.


Oops, slap on the wrist for me there.

Apologies to both Mr MacDonald and Mr Sterling


Charlie @51: True, but why does the customisation have to be done in a decentralised way? I can order a car in a whole bunch of colours, but it still gets painted in the same factory as all the others.

More intelligent production processes mean more flexibility for both centralised and decentralised production. I don't think that it helps one more than the other.


Jez: your choice of car is governed by the exigencies of engineering cars, which are horrendously complex and intricate machines that require exacting QA standards and which are developed centrally because there are a limited number of corporations worldwide which are able to Do That Stuff. (Just like your choice of long-haul widebody airliner is limited to Boeing and Airbus.)

For less complex, or small, items, or items which require extensive customization or tailoring, it's desirable to produce them close to the point of consumption because they're being made for an individual customer. Last year, when I needed 24 new bookcases, I could have gone to IKEA ... but instead I went to a local furniture designer and cabinet maker, who tendered for the job and then produced something a lot closer to my needs. IKEA is cheap because of the big-factories mass-production thing they've got going; but if you want something specialized (with self-levelling feet because the room they're going to stand in has a floor that's bowed) they're not going to help you.

Mass production is the enemy of customization, by and large. (And if you don't believe me, try getting your new car with a colour of paint the manufacturer doesn't see any demand for because their consumer research folks say it's unpopular).


Charlie: Cannondale, who make very nice aluminium bikes, set up a factory that could mass-produce one-offs at the same rate, and for a similar cost, to uncustomised bikes. Laser cutters, all fancy stuff. We can mass-produce one-offs already.

And that brings us on to another question - how to do QA on distributed production? Spare parts for cars is the obvious example, vast range of bits, people in a hurry to get them, ideal for distributed production. But if I can't get a solid guarantee that it's not going to snap and kill me, then distributed production is limited to non-critical knick-knacks.

In comparison, there's silly people saying things like:
"It has been called the invention that will bring down global capitalism, start a second industrial revolution and save the environment", from http://www.guardian.co.uk/science/2006/nov/25/frontpagenews.christmas2006

All I'm saying is that the impact will be more limited, and more delayed, than some expect.

(People who assume that, because you've got a digital copy and a digital printer then your final article is somehow just work, are people who've never made anything more complicated than a spice rack. In my experience, having been one, real engineers spend the majority of their time working out why the thing that works perfectly in simulation doesn't work in real life.)


Ellison gets his ideas from Schenectady, not Poughkeepsie.


Admittedly, I have not read the book, and might be going into really different territory here, but...

1) We already have spime-like characteristics with consumer capital equipment, most notably houses and cars, but certain labor-saving devices as well. RFID and the whole wireless ethoes is only a symptom of other changes towards natively intelligent materials, and the growing logistical ability to support customization throughout our lifespace.

2) I'm not sure how Sterling handles how Power and Society handles choice. Design is most often used to shape individual's awareness of what is possible--You have many choices, but all of them only support a certain ecological economics. However, it's hard to recognize these things because it's pretty easy to design in a way that encourages people to ignore opportunity costs, and support it by advertising and regulations which uses Society to lean against outliers. Obvious examples are housing/zoning policies and cars/transportation policies.

3) From the second idea, I think that it's easy to show that spimes would put a great deal of stress on the average person, and most people would not use it, and leave it to defaults. For too many things, you would have to make active decisions that affect how you set your routine. A chair is not going to get RFID chips in them, and you wouldn't really use them for doors. What I think will become popular will be *point*-spimes. Spimes that are for the "out of the ordinary" situations. Say you're really busy and all of a sudden have to make a change in your schedule--then you "touch" a spime-point and let it cascade as you do stuff (which in the process, inform the intelligent equipment what to do)and stuff that needs to get done quickly, even if sloppily, gets done. Sort of like a computer programmer's toolkit, only with a select group of networked equipment to do a metafunction quickly.


A short but entertaining piece from Bruce, in Rudy Rucker's SF web-magazine, Flurb:



50, Posted by: Charlie Stross

"Yes, there's a tension between centralized mass production (with economies of scale) and distributed one-off production (with bespoke customization). We're going to need the former for a very long time, if only to provide refined input materials for the latter. But I suspect most people, given a choice between an off-the-peg pair of generic jeans from China and a pair manufactured by fabber to their personal measurements, will take the latter (once the price premium drops below their threshold for discretionary spending)."

I expect LL Bean and Land's End (in the US) to offer this service long before home-fabber machines become available at all. And by the time that home-fabbers are affordable, it'll be a quick stop in a clothing store to have them custom-made by the store's fabber, at a very reasonable price.


Barry: there's currently a race going on between high street retailers and internet-based box shifters (with big-box stores like Walmart or Costco sitting in the background). I'm sure LL Bean would like to still be around by the time bespoke made-to-measure garments become a high street option -- but it's not clear what the future of retail is going to be; right now, bricks-n-mortar shops seem to be hurting badly, and with an oncoming recession the picture doesn't look likely to get any better for them in the near term.


having still not read the book yet
Chris L @46
Yes your kayak is an artefact: an object intentional modifed by a sapient being. Its lots of other things as well. 'Artefact' has connotations of ancient stuff, but it is still fings. Archaeologist (anthropologist for the US)generally have an arbetrary cut off point of 300yrs young for piles of fings. This is mainly 'cos after that their sheds fill up so fast with stuff. But if your interested in near history then artefacts can be as little as 90yrs old (WWI anyone?).
So its not the age or complication of the fing, its how you concieve it that makes Charlie's epochs different from each other. You have to invent the clocking-on-machine and the concept that the efficent use of fuels, materials and [a source of excess] labour is a _good_ thing, before you can get a centralized machine economy. Then you can write manafestos about the process. With out that change of attitude we'd still be getting the blacksmith to hand whittle ever wood screw, wouldn't we. Any one know when the British army stopped hand whittle the Royal Enfield?
So back to the kayak - thinking about how and why it was made, who uses it and why, how they curate it, what all that means to the user's society, how users and kayaks interact etc - would turn it in to an artefact in an ethnographic sense. Thinking about it in otherways turns it in to a member of the meta-classes of objects in Charlie's list. And you thought it was just a kayak.
And if it was a spime, think of all the things it could do for you: remind you that you haven't been for a paddle lately, that the weather was perfect for paddling, or that the weather is even better in that place your been trawling on the internet and this is the shopping list of things you'll need if you go...

I wasn't meaning to be rude or hurtful, just a little afraid - people's relations to fings fasinates me.

ps. Re; supply chain, real or data
What do you do about theives opening up 2 manhole covers and walking off with 500m plus of telephone cable over night? Doesn't show much enerprise, one more manhole, one more cut would have doubled their haul. Maybe their arms got tired.

I understand that I don't know much about modern manufacturing, the one prcess I'm familar with ground to a halt in the 19th c, as you know bob. So the idea of having a little aspergers person inside all my things is wonderful (in the old sense).


Hey, Charlie, glad to have found you. Sold the Hungarian translation rights yet?

Concerning Bleddyn and sarcasm: I have a notion Bleddyn may be actually quite good at it. Reread the post, perhaps?

Concerning next step in chain: Vernor Vinge's autons, from the time wars? Call them... autons, perhaps?

Concerning Bruce Sterling: methought mediocre, really, and yes, far to enthusiastic. Islands in the Net, which is just ancient, being the book of his I enjoyed the most. Bit like his pal, Gibson. And not in earnest: a kudos-seeking man with less of a sense of humour than I really like in one. Sounds though (correct me if I'm wrong) as if you were unaware of Neil Stephenson? Snow Crash and sequel The Diamond Age are about as good about this sort of thing as I have read. And Vinge, in a somewhat bizarre, maths professory old-school way, too. An in earnest, and with senses of humour, the both of them.

Fabbing: friend of mine's been playing around with 3D printers: tiresome, expensive, but cool. check out http://spidron.hu, also on the spacecollective, http://spacecollective.org/edanet ... forgive his accent, he's deep, really, honest.

I need to get off this train. Can't type on my bicycle, yet...

Cheers, later


Maggie: not rude at all. LOL at kayak as spime... George Dyson (son of Freeman) once suggested kayaks that know the way home.

I was actually thinking of the comparison between an object (e.g. kayak) in the cultural context of e.g. Inuit, versus my cultural context, even though the object itself is functionally the same.


Chris L @ 63: . . . kayaks that know the way home.

One of the land-based functional equivalents is known as a "horse and buggy." Reasonably well-trained equines can, and easily do, find their own way home (through reasonably familiar territory) even when an H. sapiens model on-board guidance unit is asleep / inebriated / otherwise incapacitated.

Their ability to consistently monitor their own progress, order supplies and maintenance work, notify their owner when action is required, and, at the end of their life, arrange for their own collection and despatch to a suitable recycling point is still a little patchy, but largely in place already. Hmmmmm . . . in this context, I wonder what the comparative timelines might be on development of purely electronic AI to this level of competence, vs. electronic augmentation of, and interface with, existing mammalian cerebra?

Paging Dr. Linebarger and Dr. Brin . . .


The minimal requirements for a spime probably apply to an object that is of fixed shape and zero on-board intelligence -- something that in an earlier context would be a hand-made artefact. The step-stool sitting next to me, for example.

As a spime, all a step-stool needs is some sort of super-RFID chip that contains:

* Production chain inventory and despatch information
* Details of manufacturer's warranty (if any)
* Identity of legal owner (presumably digitally signed by their, and the retailer's, public keys to cryptographically certify change of ownership)
* A user manual/product usage documentation/health and safety warnings
* List of material constituents for recycling purposes
* Shipping instructions for end-of-life reprocessing, pre-paid at time of purchase, to be activated when the certified owner signs off on them and shoves the step stool out the door for the trash wagon to collect.

This stuff is passive until actively queried. And it's basically equivalent to the paper and packaging that came with the step stool when I unboxed it from IKEA. The sole functional difference is that if it's in an embedded, remotely-queryable chip it's a lot harder to lose than a swatch of paperwork. And from a tech viewpoint, the only things stopping us implementing this level of spime-hood right now are (a) the cost of the chip, which is probably still hovering somewhere over the £1 mark (compared to the fractional-pennyhood of a bar code) and (b) the necessary upgrades to the supply chain, the banking system (those digital signatures) and the recycling chain (to make use of the disassembly/recycling instructions).

What's the point?

The point is that the thing can be tracked from instantiation to recycling, can't be separated from its documentation, and can't be trivially stolen or falsified.

You probably don't need to inflict spime-hood on a step-stool made from fifteen pieces of wood and eight steel nut-and-bolt fasteners, but it's a useful example. Ramp up the complexity a bit and contemplate a future bicycle. It's got tyres, made from synthetic materials such as kevlar or synthetic rubber. It has components (the shock absorbers and gear train) that need regular maintenance and contain sub-assemblies. It has a frame made of titanium or high-grade steel. Recycling it doesn't happen all at once (the tyres are changed regularly; other bits may be replaced or upgraded gradually over its life) and requires a number of different pathways (tyres and a titanium frame go in radically different directions when they're no longer needed). It's useful to keep track of this stuff -- to embody the user manual, and maintenance logbook, and change notes in the item itself.

And this is long before we consider something as complex as a mobile phone, or a combined heat and power generator system for a house, or a car, or a fork lift truck.


Charlie @ 66: Thanks for laying that example out so clearly; I haven't had the patience to sit down and do that the last few days, and it's a central part of the usefulness of spimes. Here's another example that shows the usefulness to designers and manufacturers (and thus not dependent on distributed manufacturing) that also addresses some of the security issues: tracking the reliability of manufactured parts. Suppose each of the major parts of a bike has a spime chip, and that, while there may be an overall chip for the bike as a complete assembly (in fact, there should be) that contains the IDs of all the spimes it contains, none of the sub-spimes has a back-pointer to the assembly. Then, should the deraileur fail, for instance, it can be sent to the address of a failure analysis lab stored in its chip, or simply recycled, and its lifetime reported to the manufacturer and designer. And the owner of the bike, or even the identity of the bike does not have to be a part of either of those transactions; the shipment or recycling could be prepaid, as you mentioned.

So problems with a design or with purchased components can be tracked closely and analyzed, with permanent records that don't impinge on customers' privacy available for purposes of regulation and accountability.


Distributed manufacturing has already taken over the globe. It is called the biosphere....

If a machine can build and maintain itself, then the only cost is feedstock. Whether spimes can hit the target is another matter.

I see distributed manufacturing as one of the ways that we might cope with AI driving down the wages of everyone on the planet. Make everything you need from your bit of land.


Schenectady, where GE had bright ideas, not Poughkeepsie, where IBM still has them.


Will: your concept has an embedded assumption which is not true in all cases: that everyone has a bit of land with which to make stuff. I, for example, don't: I live in a top-floor apartment in a building. I own the freehold on the apartment, as do the owners of the other five dwellings on the common stairwell, and we effectively co-own the land we live on. But my ability to grow stuff is non-existent -- and this is the majority dwelling pattern for people in the city I live in.

It's a nice idea, but all you're doing is pushing back to a subsistence-agriculture model (albeit with a much higher standard of living for the subsistence farmers).


Hmm, true. I think it is my extreme self-sufficiency side showing.

More realistically you extend the notion of "you" out a bit, to a country or neighborhood. Whatever the smallest unit is that can sustain the technology and maintain the lifestyle that the people there want.

Rather than being in hock to the region that can produce all the consumer goods at the cheapest rate because all the factories and supply chains and/or entire business is automated, each region makes use of the natural resource of sunlight, air, rock and refuse to build what they want.

Not happening tomorrow. But I can't see any other nice way out of the automation of jobs track we seem to be on.


Spimes? And I'd just got my head around the Noosphere.

If everyone who commented on this trend sat together in a room, I'd simply go David Selig on your minds, and have a clearer picture. Alas :D I'm feeling so dispossessed.

Yes, I never left, yay. Hugo's for all.


I hope you don't treat this as spam, but I wrote a little chuck of short fiction about fabbers tonight over on my blog, and then happened to see them mentioned here.

So in case anyone wants to read it, click here: http://www.sollanych.com/blog/2008/11/the-machine-movement/

I can't wait until we can get these things ourselves. Only $15,000 for Z-Corp's stuff now, and I'm sure it'll be down to 5,000 with five years...


Mike, there's a number of different groups making their own 3D printer. fab@home, reprap (a bit of self-advertising there), candyfab, and so on. The build cost is certainly less than $5k.


So no one commented on the fact that I can't tell a motor bike and a rife apart @62 [Lee Enfield -rife, Royal Enfield -bike (though it does have a picture of crossed rifes on it)] and was probably hand whittled too.
Thanks for the example Charlie @66. Speaking for myself I wasn't sure how much info was capable of being stored locally on the chip and how much was allocated 'somewhere-else'. So as long as the handreader is charged and your in control the user comands the information. But if you 'automate'/network the reading of spimes so the information is actually useful, then there is all sorts of hacks one can throw together to make over friendly kayaks or whatever. And that was the conclusion I jumped straight to.