The text of the speech will be in italics, my annotations will be in brackets in the color orange.
Let me start with nanotech, because the nice thing about the economics of nanotech is I think I understand them. In fact, I worked them out in 1987, when I gave a talk on nanotech at MIT, shortly before the Foresight Institute was founded, if I worked out the dates right.
I will then go on to AI where I am lot less sure of my views, but where I have some useful things to contribute.
Nanotech from an economist's standpoint has several interesting features which it shares with a number of technologies we are already familiar with. The first one, is that you have a very large fixed cost if you design a nanotech object, if you're ambitious enough about it a nanotech car, where you've got to figure out where every atom goes and there's going to be a certain amount of repetition of the diamond windshield, you don'r really have to figure out atom by atom, you can analyze what a cubic micron of diamond looks like and you can just duplicate it. But, there's still going to be a very large cost to doing good nanotech stuff.
But if you're talking about Drexlerian Nanotech [This nanotechnology is named after Eric Drexler who popularized nanotechnology in the 1970s and 1980s at MIT. He wrote a now famous book titled, Nanosystems: Molecular Machinery Manufacturing and Computation in 1992. He founded the Foresight Institute in 1986. He is no longer connected with the organization. He has been ostracized by many in the scientific community and his "molecular assembler" has been seen as unfeasible.] which is the kind that's fun to talk about, even if it's not being done much at the moment, with general purpose assembler and all that stuff that the people in this room and six other people in this world know about, then in that world, once you gotten the design done the hard work is done. We assume we've got someway of transmitting that design to your assemblers and you can now produce as many copies as you like at essentially the cost of raw materials and dumping a pint of gasoline every once in a while, to be disassembled for energy and a little extra carbon, to build whatever you're building. [Objections to Nanotechnology are not rare. A classic critique was published in IEEE Spectrum in 2008, written by Richard A.L. Jones. He views molecular nanotechnology as "...the reduction of all materials things to the status of software." You may read his article here. An excellent response has been written by our friend Michael Anissimov here.]
Fixed & Variable Costs in Nanotechnology
There are things at present that have just those kinds of economics. It's typically true of the things we think of as intellectual property. So that the cost of writing a book is many hours of labor, but once you've gotten the book written it doesn't cost much to make the second copy. The cost of creating Microsoft Word was presumably some huge number of hours of some programmers labor, but the cost of my making a copy is let's say thirty seconds on my computer. There are other things which have that technology.
So my prediction would be that in a functioning Drexlerian nanotech economy, that would be a typical pattern. You would have firms producing multiple products, multiple firms in the business, there would be more than one company producing nanotech cars, but they would be different nanotech cars. You would end up with an equilibrium, where each firm would specialize in particular characteristics, just as you get with software at present. Quite likely, you would get the pattern that we observe for software, which is that at any given time, in some niches there's a dominant product, when somebody else brings out a product, which is actually superior, the dominant product switches very very fast. So if you'll think about word processors and spreadsheets, there was a time when Visicalc was it. And then, there was a time when everybody knew that Lotus owned the spreadsheet world. So forth and similar things with Word. There's an interesting book by a couple of economists that look into a bunch of this stuff, where the basic view is looking at the software economy and observing that what's going on is not force externalities which is sort of the standard explanation but rather sequential monopoly because if the cost of using one more unit is low and my product really is better than yours, I can replace you very quickly, which is what's happened with software.
That's sort of the first element of this story. The second element carrying on with my analogy to software and similar things, is that it may well turn out that not only is the cost of the firm designing the duplicating technology low, so is the cost for everybody else. And you're then in the area which the firms selling music and videos are, scared of where they are unable to enforce their intellectual property rights. If we have a situation where we could imagine that we worked out general purpose assemblers, but general purposes dissemblers. So, you buy one copy of a nanotech product you set your army of dissemblers to work on it, and you end up with a complete description of the product, which can then be fed to another army of assemblers which can then produce an exact copy. This is the world where a company spends a hundred million dollars designing a really cool nanotech car, and they sell for twenty thousand dollars one copy. After they discover that they have competitors that are selling them for five hundred dollars, having dissembled that once copy. What you're selling really is the instructions for making it and those are very cheaply copied indeed.
Copyright & Nanotech
That raises exactly the problem which some of us refer to as the death of copyright. A problem being faced at the moment in the world of due to the fact that technological shifts have made copyright law much harder to enforce that it was thirty or forty years ago. Some may know the story that a century or so back you could enforce copyright without copyright law. A century or so back the great pirate nation of the world was the United States, which was not a signator to the international copyright treaties. Nonetheless British got sizable revenues from U.S. sales. The reason was that with the technology of handset lead type of printing, there a big first mover advantage. So the author sent the manuscripts to the authorized American publisher, before it was out in England so before it could be pirated. The first point at which the pirate can start is the point at which the authorized publisher is already delivering printed books, he gets all of the early sales which for many books are the a large part of the sale. If a pirate edition appears, the authorized edition brings out what was known as a fighting edition meaning a low cost edition at a marginal cost to make sure that the pirate edition can never recover it's fixed costs. So that was a technology that favored protection of copyright type rights. We've now shifted to a technology which we're all familiar with which is exactly the other way around, in which it is very hard to enforce those rules.
So that means that we may end up with nanotech facing that same problem. And then the question for nanotech, which is the same question for authors, musicians today, is how you get paid for bearing the cost of producing that cost of the first copy, given that you can't prevent people from copying it.
The first solution is that you prevent people from copying it. You have copyright law. It might work for nanotech, because after all, a car is harder to hide than a program on my hard drive is. So one could imagine a future world in which if a copy of my car is found I can sue you and you have to pay damages. If you're talking about physical objects as well as structures of information, makes that more plausible than it is for straight IP infringement.
But it's not clear that it's workable, so let's assume that it isn't. There are a number of interesting solutions. Most of which parallel all solutions in the real world, which address the problem. One of the ones I like to imagine, is that I pay some substantial amount of money to download the instructions for building my nanotech car. Having downloaded the instructions, I dig a pit in my backyard, dump in a bunch of assemblers, whatever elements don't happen to be available in the soil, soil doesn't have aluminum, so that takes care of some of the construction, I've got a bunch of garbage dumped in to get me some Carbon (this is the new version of fertilizer) and I go to bed. In the morning, there is my lovely shinning new car sitting there with it's diamond windshield. I get in the car, but they haven't handed me a key, but I am the key. When I grab the steering wheel, the steering wheel checks my fingerprints, takes a couple of dead skin cells, dead skin cells contain my DNA and the engine starts. So in addition to having the car, I don't even have to disassemble it because I have the instructions so I can build my friend a car too and it will sit in his driveway, but it won't be good for very much, because when he tries to drive it, it will not turn on (we're not considering cloning as a relevant technology). So that's one solution. That's a real solution in the software world as you know, sometimes implemented by a dongle, which is a physical object you plug into the computer, and the seller gave you the dongle, which is physical hardware that the software checks to see if it is there.
In any case, that's one solution. Another possible solution is that this time I download the same stuff, build the car; it comes with an ordinary key. The engine turns on and quietly purs as a melodious voice tells me about some wonderful houses for sale along the coast that I might be interested in, the massage services that are available in my town at a very reasonable price and a variety of other things. I look for the switch to turn off the advertising and funny thing, there isn't a switch. I look to see if there is something I can rip out, but, of course, the hardware that is doing that advertising is distributed throughout the structure of the car at a nanotech level and there's nothing I can do about it. I think about it for a awhile and I say hah! that's interesting, how did web pages pay for themselves? How does Eudora in the non-paid version pay for itself? I realize again, that the solution of "we will give this away" is advertising.
It is perfectly clear to any economist that there cannot be a private radio broadcast industry, because a radio broadcast is a pure public good. You can't control who gets it and you can't make them pay for it. Yet oddly enough, there is an industry because the solution I've just described was thought to me by some unknown genius eighty years ago or so, which is that you produce one public good with a positive cost of production and positive benefit to the consumer in a radio program along with one with a negative cost of production and a negative value to the consumer called advertisement. You tie them tightly together and give away the bundle. So that's a second solution for making nanotech economically workable.
Then there are some other solutions. There's one very old solution, which we've sort of lost track of. If you open an 18th century book on almost anything, you are likely to find in the first couple of pages, a glowing description of the most wonderful person you have never heard of at all, who is generous, wise, learned and has been feeding the author for the last three years. This is the patronage system, in which the author gives credit to the person who is basically paying his costs in exchange for what he's been paid for. There's a famous work of renaissance Italian literature, one of the most very famous, the Orlando Furioso, by Ariosto, which is a tongue in cheek renaissance fantasy fiction. In it, there is a chunk I think maybe in the first page or so, praising the virtues of this most wonderful lady by the name of Lucrezia Borgia, one of who's descendants was Ariosto's patron. In fact, I think he may someone who is seeing into the future in the Orlando and sees the great descendants that Lucrezia Borgis is going to have. Anyway, it's a very old approach and presumably the modern approach would be that in some version or other, the nanotech object comes with packaging that says, the cost of designing this red blood cell was borne by Apple Computer in order to show that they love humanity and do not anyone to ever die of a heart attack again. I assume you have the wonderful model for red blood cells where if you have a heart attack you call up your doctor to make an appointment and arrange to come in a matter of hours and you've got enough air in your bloodstream to keep you until then. So you could imagine patronage is another solution.
Final solution that we observe in software is something like open source where you get people who want to use the product who have the time and energy, in a rich society to do things they are not getting directly paid for but are getting certain sorts of reputational credit for producing it.
So those are all possibilities. And all of those lead to what I've called a software economy, with very "hard" software. It is an economy where the production of physical objects, is done in way we have traditionally done in the production of software.