You can do some amazing things by knocking flakes off of rocks. Check out these examples (scroll down) of modern flint knapping: some of the pieces even have holes in the middle, and look rather mechanistic. Of course, those are the less functional examples of the flint knapper's art; a wide variety of tools can be made if you know just exactly how to hit or press the stone.
Today's nanotechnology and stone-age flint knapping can be compared on several levels. Most trivially, both involve the production of extremely small features. But the correspondance goes pretty far beyond that--and, as you might expect if you've been reading this blog for a while, molecular manufacturing comes out looking quite good in comparison to present-day achievements.
In knapping, the edge is produced indirectly, by hitting the stone in another location so as to remove just the right flake. In today's nanotech, the structures are produced indirectly, by adjusting reaction conditions, ingredients, and sequences.
Flint knapping requires a lot of skill; it's hard to learn to control a finicky process using such indirect methods. Nanotech researchers will surely agree that this applies to them as well.
Knapped stone edges can be considerably sharper than today's manufacturing tolerances. Such direct use of a natural phenomenon produces impressively powerful results, in narrow but useful domains. Many of today's nanotechnologists say this proves the superiority of the natural (especially biomimetic) approaches. However, direct, manipulative, general-purpose access to weaker phenomena may be better than indirect use of powerful phenomena. Metal-working can't make an edge as sharp as knapped obsidian, but it can make shapes that would be ridiculous with knapping techniques. In fact, metal can be bent, cut, welded, stamped, and molded into almost any shape imaginable.
The most important property of metal, in this context, is that metal can be shaped by direct contact with tools. Not only can metal make a broader range of shapes than stone, but this can be accomplished by a small set of metal-working techniques which can be applied by technicians, rather than needing to be invented anew by experts for each new class of shape.
Molecular manufacturing proposes to make molecular shapes by direct application of molecular tools. A few reliable operations, repeated in chosen locations and/or sequences, will be able to make a vast and engineerable array of shapes. Someone who wants to invent, say, a new machine part, will not need to be an expert at chemistry, quantum phenomena, or any other specialized knowledge.
Today, non-experts can design and create shapes in at least two molecular systems: DNA--which can be "stapled" together according to very simple procedures--and Christian Schafmeister's polymers, which take a rigid form without even needing to fold. But building these molecules takes significant time, because they are still built using large tools that take a relatively long time to do each step. Molecule-scale manufacturing tools (Ned Seeman's polymer builder is a step in the right direction) should eventually speed up the process thousands or millions of times per molecule, as well as parallelizing it.
It is not an insult to today's nanotechnologists to compare them to flint knappers. It takes a lot of skill to work with such a difficult process, and the results are quite useful. But working in such a difficult domain might tend to limit one's expectation of what could be achieved with other materials and processes. Let's imagine a prehistoric conversation between an expert flint knapper, and an inquisitive non-expert who's been playing with a lump of meteoric iron:
Iron: Look at this stuff. It's amazing! You can hit it and dent it, and it doesn't shatter!
Flint: Well, then what good is it? You can't make anything out of it.
Iron: No, wait, I bet you can make *anything* out of it if you hit it enough!
Flint: Well, if it doesn't shatter or flake, you can't get sharp edges. Stick with natural phenomena; they work better than that effortful, artificial pounding.
Iron: Not everything needs sharp edges. I bet I could make a waterproof, fireproof basket out of this!
Flint: Yeah, if you hit it a million times. It doesn't scale. It's not as efficient as cleaving flint.
Iron: I just need a bigger hammer.
Flint: Quit bothering me with your science fiction stories. If you try to make dents by direct impact, you can't make any dents smaller than your hammer. I'll call that the "fat hammer" problem. And besides, that stuff is so hard that you'd damage the hammer as much as the workpiece. That's the "fragile hammer" problem. Both of these problems are fundamental, and I say that as an expert in materials. Besides, you haven't showed me a single shape yet!
Iron: Well, I just found this last week, and the chief won't give me a break from hunting to experiment!
Flint: Of course not, I'm making all the arrowheads he needs, and you're going nowhere. Show me an experimental proof--even one arrowhead that's sharper than mine--and then we'll talk. Meanwhile, quit distracting our young flint knappers, asking the chief for special favors, and scaring the kids with tales of super-knives.