Saturday, July 23, 2005

Self Replicators

John von Neumann first formalized the idea of a self-replicating machine in the 1940’s. The idea is still radical because the very definition of life includes the ability to reproduce. People fear the idea of a “sorcerer’s apprentice” scenario of a self-replicating machine running madly out of control. With nanotechnology or molecular manufacturing, this open-loop condition leads to ‘gray goo’ oozing over the landscape and absorbing everything in sight.

A “clanking replicator,” which uses a cooperating team of conventional manufacturing machines is a more realistic image. The factory would be able to mine and refine the raw materials then form, shape and assemble the parts. Researchers at the University of Bath are investigating ways to make a universal replicator that will make itself from supplied materials. Project RepRap (for REPlicating RAPid-prototyper) uses rapid prototype and small machine technology to make a copy of itself and other things. One of the pioneers of the project, Adrian Bowyer, believes that an open-source manufacturing machine could change the world. The universal constructor, he says, would have three world-changing properties.
First, the self-replicating machines would become exponentially more available and therefore exponentially less expensive. The ultimate cost of the machines would only be the cost of the raw materials and a small amount of final assembly labor. I think he over-estimates the contribution to cost of the manufacturing. In today’s modern mass-production, the processing of the item is a very small percentage of the cost. Most of the cost of an item, after the material cost, is in the production overhead cost. Overhead includes the cost to design and organize the production process. Bowyer’s proposal has an open-source design so that cost is less. But even with RepRap machines, as the volume of machines making new machines grew, the organizational costs would probably blossom to the same as any factory.
The second unique characteristic of the open-source self-replicating machines would be their continual incremental improvement – their evolution. The wide distribution of the machines into different hands would lead people to tweak their own machine to improve it. The improvements would then transfer around the community and incorporate into new machines. This is probably a fair hypothesis; it is also true of most manufactured products today. It would only be truly effective evolution if the new machines replaced the old ones.
The third factor that Bowyer predicts is that manufacturing will become so cheap that it will completely change the nature of the economy. Since anyone can make anything they want then industrial manufacturing will be completely obsolete. If someone needs a new item they can just download the design from the internet and make it themselves. In the limit, I agree that this is a possibility. However, there are many more barriers to the end of factories then the ability to make one’s own items. Just as the Xerox machine did not cause the end of the publisher, the self-replicator will not cause the end of the mass manufacturer. Some things will always be cheaper to make in mass. People will try to improve the design of objects to be able to sell theirs at a premium price. People will very quickly learn how to keep their algorithm or unique process details proprietary so that they can make a profit. So while the new method of manufacturing will change many things, I do not believe that it will change peoples’ motivations for wanting to own special things.
Artifacturing falls somewhere between nano-molecular manufacturing and factory-sized clanking replicators. The goal of artifacturing is not for the machines to take over the world – economically or otherwise – but to allow people to expand their idea of what is possible and to create artifacts of the fruits of their limitless digital pallets and imagination.


Post a Comment

<< Home