It's 2025. I want a bicycle to use to visit my friend Yésica in La Plata, a city near Buenos Aires, where I live. I type "recumbent touring bicycle" into a model search engine and get a page full of pictures of different models. I click on one of them, a model called the Matagallos, and adjust the parameters on the resulting page: my inseam is 95 centimeters, I weigh 100 kilograms, I will be carrying 50kg of luggage with me on the trip, I prefer dull brick red to the default bright blue, and because I'm pretty strong and not in a hurry for this trip, I prefer lower material cost to lower weight. My computer consults databases of parts available nearby to fulfill the model's requirements; only 1.8mm spokes of dubious metallurgical quality are conveniently in stock near where I live in Argentina, so it ups the spoke count on the wheels from 36 to 48 to compensate.
I review the model from different sides on my screen, using a finite-element simulation to simulate how it will travel over bumps and potholes, and find a problem: because this model was originally developed for lighter riders, when the rider-weight parameter is set all the way up to 100 kilograms, it happens to work out with an unreasonably low ground clearance — I'll be bumping my ass on the ground. I call my friend Violeta for help, sending her a link to the model. She shows me how to adjust the formula for one of the strut lengths to add a ground-clearance constraint, and then we test the modified model to ensure that the results are reasonable across the range of parameters.
Having validated her modification, I thank her and write up a couple of sentences explaining the problem and the solution, and publish the modification so that other people can try it out. After making a couple of cosmetic modifications (I want the struts all to look like bones, and I don't want to display the model name, because the police near here are nicknamed "gallos", and I don't want them to feel threatened), I look at a list of prices and delivery times from manufacturing services in my neighborhood. I choose one that's 400 meters away and offers me US$102 with a delivery date of three hours in the future.
I click the "submit" button, confirm the transaction on a handheld smart card, and switch back to chatting with Violeta about the vacation she's planning to take with her husband next month, visiting Arecibo in Puerto Rico.
Three and a half hours later, my doorbell rings, and there's a truck with my bicycle, assembled by robots from whatever materials and parts happened to be on hand, using more or less traditional processes of drawing, forging, cutting, drilling, lashing, gluing, nailing, and welding. By chance, a substantial part of this particular bicycle is made of wood recovered from discarded pallets. I key in the delivery PIN to get the truck to release the bicycle, lift the bicycle off the truck, and press the "accept" button on the truck, so that it can move on to its next stop. Mine are the first human hands that have touched it.
My ride to La Plata is uneventful — the full-body fairing keeps the rain off me, and the speakers have amazing bass — and the bicycle mostly performs well, but I notice a grinding noise coming from the front wheel. As I pedal, I search for "grinding noise matagallos front wheel" in a search engine, and it turns out somebody else had the same problem last year. They figured out that the problem was an over-tightened wheel bearing, and explain how to check, so I stop to check. Turns out my Matagallos got assembled with a cartridge bearing, so it can't be tightened or loosened at all, but it seems to have some kind of sand in it. I do a search for "sand bearing" and the number of the manufacturing service, and it turns out that several other people using them have had the same problem, in products as diverse as drills, fans, and food processors, although I'm the only one to run into it with a bicycle.
Once I arrive in La Plata, I stop by a manufacturing service that I'd contracted with on the way; they have a new front wheel ready and waiting, and in a couple of minutes install it on the bicycle, taking the defective wheel as partial payment of US$5. Since I've reported the bearing grit problem to the service near my house, they accepted the blame and paid the remainder of the cost, US$3. I leave them a good review, explaining the problem and resolution, and tag the other people who had reported grit problems so they'll see it.
At Yésica's house, she and I pick a couple of interesting recipes off a recipe blog, tone down the spiciness for her Argentine palate, and submit it to the robot chef around the corner, paying it US$1 for the service. Half an hour later, a bell rings to notify us that the food is ready; we stroll around the corner to pick it up and bring it back home.
Waking up in Yésica's bed in the morning, I'm pleasantly surprised to find that my (Violeta's) improvement has been accepted into the mainline Matagallos model, and I'm now listed as one of its contributors. The maintainer accepts my argument that the credit belongs to her, not me, and replaces my name with hers. I publish a short experience report, describing the performance of the Matagallos on the pothole-laden road from Buenos Aires to La Plata.
But now it's time for me to get to work. A few blocks away, a couple I don't know are seeking a mediator for an unspecified marital conflict related to their autistic child, and as I happened to be in town and with good reviews for similar mediations from some of my friends, they requested that I meet with them. I consult their reputation — they seem to be trustworthy — so I accept their proposal of a US$800 fee for an hour and bike on over to their house.
There are some things that robots still can't do, you see, and those things have gotten more expensive.
Before I return home by bus, I sell my new bike as scrap to a local manufacturing service for US$97.
(The above scenario is based on some things Richard Stallman wrote, but it isn't an attempt to represent his point of view. Violeta is a real person, but Yésica is not.)
XXX why did you sell the bicycle?
In the last couple of decades, we've seen a really remarkable shift in production from the market economy into ad-hoc volunteer networks engaging in something Yochai Benkler calls "commons-based peer production": people contributing resources on a voluntary basis to a common product that everyone can use without restriction. Examples of such products are Wikipedia; Stack Overflow's collection of technical questions and answers; the worldwide repository of data available through BitTorrent; the worldwide tourist lodging system CouchSurfing; the FreeBSD system that is the basis of Apple's MacOS and iOS operating systems; the WebKit browser engine that drives Chrome, Safari, the Android browser, and the iOS browser (known as Mobile Safari); and arguably the collection of videos on YouTube, the social graph and other information on Facebook, the corpus of photos on Flickr, and the contents of other such "user-generated content" sites.
I say "arguably" because these proprietary sites aren't governed by the same liberal commons-protecting licenses that enable the communities I listed earlier, so I'm not sure if they fit Benkler's definition, but much of the same dynamics seem to apply. Whether you upload zero, one, or a thousand photos to Flickr, you can still look at the commons of all the photos on the site, produced by your peers. (I'm embarrassed to admit I haven't managed to make it through Benkler's book yet, so I'm not totally sure where he draws the line around "commons-based peer production".)
This is a remarkable phenomenon. One after another, we see market-economic projects collapsing when they have to compete with commons-based peer production, much as the Soviet Union eventually collapsed economically when its system of production could not compete with the capitalist bloc.
Apple had 20 years of experience as one of the most innovative and often financially successful producers of computer software, employing many of the best programmers in the world, when it decided to cancel its in-house next-generation operating system project and switch to NeXTStep, much of which is free software. Apple's own internal software development project simply could not compete with the commons-based peer production of FreeBSD — although so far its user interface seems to be holding its own against Ubuntu.
Wikipedia, similarly, completely crushed Encyclopedia Britannica, Microsoft Encarta, and the other lesser-known commercial encyclopedias.
Never before has capitalism found itself face-to-face with a system it can't compete with economically. The only feasible response, taken by Wikia, YouTube, Facebook, and so on, seems to be to try to build a "commons" that in reality is privately owned by you, so everybody else does the work of constructing the economically valuable product, while you take home the profits. People seem to be starting to realize that this is not such a hot idea, but it's going to be a while before we can fix the problem. But that's not what I want to talk about now.
You can also argue that this trend is unjust, because it takes economic rewards away from creative people or from the majority of the population, which is also a really interesting discussion to have, but that's also not the discussion I want to have in this post. Bookmark the idea for later, because I want to talk about it in a later post.
It's reasonable to ask how far this trend of replacing market economic production with commons-based peer production can go, assuming it's not already played out. Wikipedia, for example, has replaced the encyclopedia market, but there still seem to be lots of other books out there. A lot of the fiction books can maybe be replaced by fanfic sites; Wikipedia and the WikiBooks project of the Wikimedia foundation are working to replace other nonfiction books, as is Stack Exchange (the family of sites including Stack Overflow); BitTorrent and YouTube are replacing TV to some extent; Android is largely replacing Microsoft Windows; but what about the implications for things that aren't pure information? Physical things like recumbent bicycles, chairs, highways, swimming pools, clothing, food, housing, clean water, headphones, and antibiotics?
Well, those physical things are all produced by the following unreasonably general process:
In the case of a bicycle, for example, the raw materials include wheels, a frame, a seat, and a chain; the energy moves the parts into place relative to each other and tightens the nuts needed to hold them together; the existing durable physical things probably include some wrenches and a chain tool; the design consists mostly of the mechanic's choice of which of these raw materials to use; and feedback is used, perhaps, to discover that the chain chosen was too short and needs to be lengthened by adding some links, or to replace the desired double-walled aluminum front wheel with the single-walled steel front wheel that happens to be in stock.
But each of these "raw" materials is itself a physical object, and is produced in the same way. The frame is made from raw materials of metal tubing, consumable welding electrodes, paint, headset bearings, brakes, the pedal assembly (I forget what this is called in English; in Spanish it's the "caja pedalera"); the energy needed to combine them includes the energy to cut the tubing, melt the welding rods, aerosolize the paint, and tighten the pedal assembly and headset bearings; the existing durable physical things include the paint sprayer, the arc welding buzzbox, and some more wrenches; the design describes, among other things, exactly how long each piece of tubing should be, which alloy it should consist of, the shape of the cuts to make, and how much metal to deposit in welding; and the adaptation in response to feedback may involve, for example, re-welding spots on the weld that melted through the first time.
If you carry out the recursion all the way, you find that the raw materials for a bicycle consist of most of the 89 or so natural elements: the tires are made of carbon, nitrogen, oxygen, hydrogen, and sulfur, for example, while a steel tube may be made of iron, carbon, chromium, molybdenum, nickel, and silicon, and an acrylic fairing is just carbon, hydrogen, and oxygen. The existing durable physical things necessary to process these raw elements into a bicycle currently include blast furnaces, zone melting facilities, hydrocarbon cracking towers, other chemical process plants (including their catalysts), welders, cutting tools, and so on, each of which uses energy in its own way, although largely to heat things up, cool them down, or move them around. (We can imagine machinery that used a narrower range of processes and of materials, but making an acceptable bicycle entirely out of them seems like it will be a big challenge.) The design covers everything from the crystal structure of the wheel hub to the overall bicycle weight.
A sandwich is made of the same raw materials, although in somewhat different proportions, but much of the design is provided by genetic material, and much of the energy is provided by the sun through photosynthesis. Plant growth is really interesting because, although it's a fully automatic process, the final design of the plant can be extremely varied as a result of adaptation to its environment ("feedback" or "improvisation" in the framework above), both in its physical form and in its chemical content. The same variety of beans can taste quite different depending on how much you water them and what kind of soil they grow in.
An interesting feature of looking at things this way is that almost none of these raw materials are typically depleted. Helium can be depleted by escaping the atmosphere, and a few of the natural elements (like radium and radon) decay into other natural elements, but they're being constantly produced by thorium and uranium, which are decaying but not at an appreciable rate. And since bicycles and sandwiches are made of the same elements, you can recycle bicycles into sandwiches, or sandwiches into bicycles.
This framework even applies to, for example, manufacturing a copy of Firefox. The raw materials are a small area of ferromagnetic oxide on a spinning disk platter; a small amount of energy is used to magnetize them in a particular sequence, under the control of the disk head and, eventually, a whole computer; the design is the bits of Firefox that have been downloaded over the web and reside in memory, awaiting being written to the disk; and there's a certain amount of feedback involved in choosing which disk sectors to write. This may sound absurd, but the process is essentially the same as printing a book on your laser printer, except that the raw materials are a little harder to recycle.
Traditionally, in a manufactured product, the raw materials were drawn from a relatively small set of natural materials and refined alloys, perhaps a few hundred for most products; the energy was provided by human muscles; the existing durable physical things were general-purpose tools; the design was produced, or perhaps remembered, by an master craftsman; and feedback was also exercised by the craftsman.
In this framework, then, the First Industrial Revolution was, more or less, simply a result of replacing the source of energy with a more inexpensive source of energy: coal producing steam. The Second Industrial Revolution, centered on mass production, was more or less a matter of reducing the need for feedback to a minimum, largely provided by the introduction of many specialized existing durable physical things ("tools") but also by improved measurements that reduced the variation between objects, and a much wider variety of raw materials. The Japanese dominance of manufacturing was largely because they reincorporated ubiquitous feedback, but at a different scale — the workers worked to use feedback to optimize the process, not to fix an individual produced artifact.
So of these five factors of production, all five were originally very expensive; making #2 cheaper created the British Empire; shifting some cost from #5 to #3 created the Second Industrial Revolution and the American Century; and making #5 cheaper instead of eliminating it created the richest country in the world, at least if you measure by life expectancy.
So, within this framework, here's what I predict will happen to the cost of manufacturing in the next few years:
Raw materials, in the sense of elements, are becoming progressively more available, because people keep mining them and not putting the scarce ones back in the ground. Even when they do put scarce ones back in the ground — for example, the indium in indium tin oxide coatings on discarded LCDs, or the cadmium in nickel-cadmium batteries — they mostly put them into specific small locations, where they should be recoverable later.
Energy is already really cheap, as it has been for a century, but it's going to get a lot cheaper soon, because as of 2013, photovoltaic power plants have gotten cheaper than fossil-fuel plants in sunny parts of the world. Without subsidies. And a substantial part of their remaining cost comes from the energy needed to make them. Almost none of it comes from the cost of the elemental raw materials, primarily silicon and aluminum.
Existing durable physical things have their cost determined by the cost of manufacturing them, so they fall out of the analysis.
Designing things is still really hard, but designs are just informational goods, and it turns out that commons-based peer production completely kicks the ass of the market at that. Sometimes.
Eliminating feedback was the core of the Second Industrial Revolution because feedback required human attention. But computers are feedback machines, and now they're really cheap! Industrial process control was a major use of minicomputers starting in the 1970s, and computer-based automation (computer-based in order to handle feedback) has been the key to the dramatic increase in manufacturing productivity per manufacturing worker over the last 30 years. We can predict that this trend will continue, and as a result many of our Fordist instincts about how to optimize productivity will turn out to be wrong. (I say this despite the fact that Toffler-style "mass customization" has been a highly-hyped business failure for nearly 30 years now.)
So it seems likely that the main cost of manufacturing in the near future will be raw materials and energy, with some sort of multiplier for the durable tooling needed; but only insofar as we can sustain commons-based peer production for manufacturable designs.
That is, within a few years, by far the most expensive part of manufacturing a bicycle, or nearly anything else we manufacture, will be designing it. And that design effort will probably be mostly done through commons-based peer production, not through a market, because market economies are generally unable to reach a level of productivity sufficient to compete with commons-based peer production in areas where we have figured out how to do commons-based peer production.
The distributional equality issues I mentioned in passing previously are a big problem. They could keep us locked in an outdated 20th-century industrial production model.
There's the usual problem of diffusion of complex innovations: manufacturing services like the ones I envision above are only useful if there's a sufficient volume of automatically-manufacturable parametric models, people who know how to tweak them for their needs, and software for creating, modifying, and evaluating them; and each of these other three things depends in the same way on the other three. So diffusion could take some time.
Environmental collapse, or other forms of societal collapse, could cut off long-distance communication, which would effectively decimate the people you could collaborate on models with. It could also cut off the long-distance trade needed for the mass-produced "processed" materials that can't yet be produced locally, such as integrated circuits.
Robotic automated custom manufacturing could remain dramatically more expensive than mass production. We see this, for example, in integrated circuit manufacturing: IC customization is certainly possible, using lasers, electron-beam etching, or "programming" (i.e. blowing fuses), and in fact I think most ICs undergo at least one of these processes, but the vast majority of the IC manufacturing process is done with X-ray lithography, the most mass of all mass-production processes, which commonly produces trillions of identical parts (such as Flash memory cells) in a single operation. And semiconductor manufacturing is one of the most highly automated of all current industrial processes.
Purely competitive concerns could result in people sabotaging each other's productive capacities. Israel might have a strong motive to destroy any general-purpose manufacturing service in Palestianian- controlled parts of the West Bank, for example, while the Palestinians might have a strong motive to do the same for general-purpose manufacturing services in Israeli settlements in the West Bank. Taken to the extreme, we could see assassinations of researchers, like those carried out by the Mossad against Iranian scientists, those carried out by the Unabomber, and the famous assassination of Gerald Bull.