The question is: Given that the animation doesn't show power or control, but only motion, how can a nanoscale production line be powered, and how can its activity be controlled?
A user known only as "anonimouse," who says he's experienced in nanoscale work, asks how we can know what we're building, and whether it is being built correctly. I would have continued the discussion over there, but three days after posting my last comment, it is still "awaiting moderation." So I'll continue it here.
In addition to his more general questions, anonimouse proposes the problem of "eutactic sensory blight." Water, he says, is a useful medium for transmitting information about a workpiece, and a eutactic system can't sense enough about its activites to control them successfully.
Anonimouse says, "Not only can I not see what I’m doing at a 1-10 nm scale, I can’t even see what I’ve done. If I look at it with an electron microscope, I change it. Sure, I can look at a few conveniently-presented outer surfaces with an AFM or STM, but I can’t see below the surface, and I can’t see features that are hidden by other features." I'm not yet clear on whether s/he is worried about development of new techniques, or operating those techniques in a nanofactory, or both. S/he talks about the lack of an "evolutionary feedback loop" but also talks about the need to deal with production lines breaking.
To develop new techniques, we have some very powerful techniques that are not available to evolution. One is computer simulation. Yes, it's imperfect, but it can still be quite helpful, showing us what's probably happening even in the interior of a molecular construct. Another powerful technique is field ion microscopy, which can pull apart a sample atom by atom and report the 3D positions and isotopes of the atoms.
Once a technique is developed and incorporated in an assembly line, there are lots of sensing opportunities. Keep in mind that fabrication will be additive, so every atom (or at least, every small molecular fragment) will have been newly deposited on an accessible surface at some point. Sensing shouldn't have to be too detailed: Did the deposition work, or not? Is the bump where it should be, or somewhere else?
If anonimouse will visit us over here, we will be glad to continue the discussion. We are talking from rather different mindsets, so we may miscommunicate and miss each other's points at first. But this kind of discussion is very valuable to the molecular manufacturing community, as we work from high-level theory (scaling laws, existence proofs, lower bound calculations) to system architecture.