A couple of interesting articles in C&E News this week: molecular building blocks forming covalent templated structures.
Buckytubes have been used to guide formation of a new buckyball polymer. Start with C60O (epoxide buckyballs--a buckyball with one oxygen atom stuck to it) dissolved in supercritical CO2. Deposit them in buckytubes. Let the CO2 escape. Heat the tubes. The oxygen, and a neighboring carbon, form bonds between adjacent buckyballs. Mechanically confined chemistry, in the absence of solvents.
Branched polymers allow the creation of three-dimensional structures rather than the long floppy chains formed by linear polymers. Dendrimers are an example, but they're basically spheres like Koosh balls. Now a branched polymer has been discovered that'll self-assemble on micro-scale surfaces to form templated structures. The surfaces are created by ionic surfactants or triblock copolyethers. What this means is that the configuration of the polymers can be controlled by a very weak guide.
All that's needed for molecular manufacturing is one material that can build shapes which can catalyze or guide the programmable building of shapes and machines in that material--and of course, lots of R&D to learn to control it in the lab. The more we learn, the easier it will get. Could this polymer be an adequate building block? We can't tell, because no one (that we know of) has developed the research infrastructure to study the question. It wouldn't be that hard to do. Sooner or later, a good building block will be invented. Whoever is ready will find it surprisingly easy to develop MM. And everyone else will be unprepared for it.
Of course, the first MM may use a relatively weak material, or rely on extremely expensive feedstock, or only work underwater, or have some other limitation. But how long will it take to get from there to diamondoid or graphene MM? Well... again, no one knows.