|
Nailing down conclusionsWe are ready to conclude Part One. Instead of telling you what to think about all this, I'm going to indicate some questions that need to be addressed and let you arrive at your own conclusion. This Web site is like a math book, where the reader is supposed to read the book with pencil and paper in hand, and prove some theorems. The methods of argument are as important as the subject matter -- probably more important, in the long run. Time will settle the argument about nanotechnology, just like time settled the argument about the Y2K problem (except it will take longer). Now that January 1, 2000 has come and gone, it is obvious to everyone that TEOTWAWKI didn't happen. The lights stayed on, the banking system didn't collapse, and the world is going about its business just as before. The Y2K survivalists who moved to remote areas of Arkansas or New Mexico, with their generators, guns, and two-year supply of millennium food, look like idiots. Eric Drexler has not made a specific prediction of when the nanotechnological TEOTWAWKI is going to happen. He gives a range of dates. So there won't be a dramatic embarrassment for him like there was for the Y2K survivalists. Nevertheless the day will come -- maybe 20 years from now, maybe sooner -- when it is obvious to everyone, including him, that nanotechnology isn't going to take the form of universal assemblers that make everything without human labor. Meanwhile, the important thing is to understand why the idea of a universal assembler is an illusion. I learned a lot by thinking about the Y2K problem and trying to anticipate what would or would not happen. I learned even more by thinking about Engines of Creation. It's worth taking some time to go over Dr. Drexler's argument line by line and get it in full focus. |
|
Exercise 1. In Chapter One, in the section on Disassemblers (page 19), we find this:
The exercise is to calibrate the idea of a disassembler. Can you think of any system (of any size, localized or scattered) that can take anything apart, while recording what it removes layer by layer? (There is now a hint at the bottom of the page.) Exercise 2. In Chapter Five, in the section called Accelerating the Technology Race (page 80), the following sentences occur:
The exercise is to map this idea onto a different domain, in such a way that the mapping sheds light on the question of whether the idea makes sense. (If you choose software as the new domain, and you say something about programming in assembly language, you have to be careful not to let the pun on "assembly" confuse the issue. However, software is not the only possible choice.) Exercise 3. In Chapter Three, in the section called The Assembler Breakthrough (page 49), the following sentence occurs:
The main clause is almost a truism. If it is assumed that a design must stay within physical possibility, then anything that can be designed can eventually be built. Of course, why not? An "assembler" may be an economy comparable to Japan, and "eventually" may be a long time, and the question of cost isn't considered here, but no one can deny that something that is physically possible could eventually be built. But what about "sidestepping the traditional problems of materials and fabrication" -- what on earth does that mean? The exercise is to explain why an assembler will be able to sidestep the "traditional" problems of materials and fabrication -- or why this is nonsense. Exercise 4. In the same section, two paragraphs later, we find this:
First you need to go back to Chapter One and review the difference between first and second generation nanosystems. The exercise, of course, is to calibrate this idea. What would be involved in designing an assembler? How big a design project would this be? Can you think of any other project, in any domain, that would be comparable to this? Can you give an order of magnitude estimate of how many engineers would be required, and how long it would take? Exercise 5a. The paragraph quoted in Exercise 4 continues:
Is this how engineering projects are done now? Do engineers (or programmers) design entire systems in advance, and find that they work when first built? Obviously not. (Can you spell "beta"?) The question is why. Will it ever be possible to design entire systems in advance, and find that they work when first built? Will it ever be possible to dispense with the experiments with which we ground our language? This is a nontrivial question, but it has a definite answer. 5b. Moving on to the next paragraph,
Well, other questions suggest themselves: how far would it have to go, and in what time frame, in order to result in an explosive breakthrough? Suppose the design-ahead process in nanotechnology proceeds in an incremental fashion, as it normally does in other industries -- i.e. design something and then test it; design something else and then test it; and so forth. Then what? Is someone going to design the entire "truly general fabrication system able to make anything that can be designed," in advance, in such a way that it will work when first built, and in such a way that it can "sidestep the traditional problems of materials and fabrication"? If not... 5c. The next sentence is where it all comes together:
Is this going to happen, or not? It isn't a matter of opinion or speculation. It's a matter of fact and logic. We can draw a conclusion about this that is as certain as whether the sun will rise tomorrow, or whether programs have bugs. The exercise is to state the conclusion and prove it.
|