Tuesday, March 31, 2009
A Junkyard Universal Deconstructor
Ever read Accelerando? It's one of those science fiction novels that'll annoy the hell out of you while you're reading it with half-thought-out ideas, unfulfilled technological developments, and some superficially clever intellectual candy that's really little more than terms from different disciplines being jumbled together. But for all that, I've been thinking about the damn thing for about 3 years now. I guess that means it's good, and I recommend it.
One thing that can't be said about the book is that it's not ambitious, and ambitious is always good. I should also add that the author, Charles Stross, was kind enough to respond to an email I wrote him on a completely different topic.
A theme of the book is that dumb, cognitively inert matter has been getting less valuable as human economics accelerates, and that as the curve gets steeper, increasingly the main value-driver of any object or being is how smart it is, i.e. operations-per-second per gram. Presumably this is why labor has taken on increasing value-adding power over time, although it would also predict commodities getting cheaper. Yes, Julian Simon won this bet once (but not the second time), but it's hard to argue that in 2009 there's a clear trend for people to value gold less than before. Dumb matter doesn't get much dumber than gold.
In any event, our increasing ability to create cognitively active agents gave us power to reshape matter. And here is one of my (and many readers') complaints about the book: as we approach the Singularity, there are little self-reproducing nanotech agents crawling over everything, doing their thing, assembling or breaking down matter as need be. At the same time, we are told that de-manufacturing becomes an important industry, because the matter locked up in the old technology is much less useful in a Chrysler than it is in a self-assembling computer. See the problem? During the course of the book the nanotech bugs dissassemble the Solar System. A Chrysler shouldn't be harder to take apart than Mercury; especially not so hard as to need its own industry.
Not long after I read the book, I moved. When I moved I got rid of my crappy old TV. The day I took it down to the dump in my city of maybe 150,000, I dropped my TV onto a pile of at least ten others. And what I noticed is that no swarm of nanotech bugs appeared to chew it down into cybercellulose. And the guys working there weren't really de-manufacturers; nothing would be done with the materials from these televisions besides pile it somewhere else.
Now, before you email me with all the specific recycling paths of a TV's components (which, if they exist, I'm glad they do), my point is a more general one. Every day machines break and are discarded, yet in most broken machines, typically only a few components have failed. Imagine that there were a way to systematically mine the working components out of broken machines.
You'd have to find an easy way to catalog the full set of components in each failed machine (worrying later about exactly which ones are shot), and then run through all the possible combinations of components from the inventory of junked machines and whether any of those combinations could be easily extracted and built into a new machine. I'm not designing anything new, just searching through a list of known designs for what components are required. A computer would be best for this, and of course it would need not only a huge list of machine blueprints, but the knowledge of how to construct and deconstruct them. Of course, there are already refurbishing businesses that pull out hard drives or copper tubes to put in computers or refrigerators that still work, and you don't need a computer for that. But the limitation there is human processing power - no one wants to sit and constantly worry whether they finally have that capacitor they need to make a microwave. To make this most cost effective, the ultimate goal would be attaching the system to a generalized assembly line with multiple manufacturing tools.
Benefits: parts of wasted machines that would be put in landfills are put back into the economy. Proprietor makes money. Drawbacks: Skynet. Major challenge: solid-state electronics are very device-specific, and de-manufacturing is apparently not yet as economical as outright recycling of raw materials.
Anyway, it's worth a thought, since there are real generalized constructors being built.
Comment here if you like but I'm also going to put this on halfbakery.com for grins.