I'm not impressed. The burden of proving it is science lies with the claimant. Drexler's site appeals to physical scaling laws, which to me makes it pseudoscientific speculation, or perhaps scientific speculation (with a little more convincing). Thorne and Braginsky's massive LIGO paper is an example of one that progresses to scientific status. It follows curves already established experimentally, and so is much more believable.
It's not scientific to say "all flying machines crash"; the scientist must (and is usually more than willing) to provide physical laws and scaling suggestions that back up the assertion. Peter Woit had a good point these last few days--real science is hard work.
Drexler seems to be proposing a microscopic cooperation of molecules which has no precedent. I have no reason to believe his scaling is obeyed into this regime until he can show "curves already established experimentally" that extrapolate to his complex motors. Real successes in nanotech have been incredible: nanotube resonators, SETs, self assembling DNA--very real, and (except for the third) very useful.
There's a standard picture in a bunch of experimental presentations. It's a log plot that has an up-right diagonal line. Towards it, a line burrows to the bottom-right, until it curves upward, "repelled" by the line. This represents a scaling law which is eventually foiled by backaction, the tradeoff response. New techniques approach the line from a different height/width, or pass below the line before being repelled by another limit.
Warning signs here are: PhD 30 years ago. Not associated with a university. Lack of published progress. Any fool can tell you the consequences of such a technology, which is what he wastes most keystrokes on.