Again, many thanks for your papers.

It was very interesting to actually see numerical estimates for the time and cost involved in building the fine motion controller (FMC) and potentially a programmable nano-assembler (PNA).

The main place where I'd like to hear more information from both you and Chris is on how well established the experimental work on positional control paths of your paths are. In the case of the wet path, I agree that many enzymes have been shown to catalyze a remarkable variety of reactions, and that the ability to modify those enzymes has been developing nicely.

One place that concerns me, however, is the ability to design enzymes which can connect larger and larger substructures. In the FMC project, the final target structure is, as you write, ~30,000 Daltons, and the final assembly step would be performed by an enzyme with a weight in the 10,000-100,000 Dalton range. Particularly if the enzyme is in the lower end of this range, it can't englobe the reactants like enzymes building smaller molecules can.

This isn't a showstopper, after all, DNA polymerases are proof that enzymes can construct atomically precise structures larger than they are. Nonetheless, I think that the design problem for the enzyme is likely to get harder in this regime. The enzyme has to grip parts of its substrates' surfaces firmly enough to give atomically precise relative positioning, rather than the whole surface. The gripper may have to distinguish between somewhat similar chemical sites on the surfaces of its substrates (particularly on the larger of the two substrates). Landmarks can be designed in, with some (mild) constraints on the FMC design (not nearly as severe as in non-enzymatic self-assembly) – still, ensuring recognizable binding sites is still necessary in this approach.

I think that a good demonstration project, before the FMC project itself, would be to design one enzyme that attaches some type of building block to a well-defined location on a rigid substrate that is larger than the enzyme itself. The substrate doesn't need to actually do anything, but this would be a good demonstration that the tasks involved in designing an assembly stage enzyme don't become unreasonably harder as the fraction of the substrate that the enzyme can bind to drops. In order to avoid a chicken-and-egg problem the substrate could be a large atomically precise naturally occuring structure such as a ribosome, possibly with cross-links added to increase its stiffness.

If I read Chris's paper correctly, it looks like more of a "dry path" approach, using "covalent chemistry done by STM" as one of the key enablers. I was aware of the Fe-CO work a few years ago but would like to hear more about what sort of positionally controlled synthesis is now possible with an STM tip. There has been lots of theoretical work, but there is always a concern that a theoretical model might have missed an important interaction of some kind. What sort of reliability has been shown for adding groups to a substrate? Have multilayer structures been built? Cyclic structures?

At one of the Foresight conferences several years back someone (Smalley???) challenged the STM people to add one ring of carbons to the end of one fullerene nanotube, extending it by one lattice spacing. How near or far does that milestone look today?

Many thanks to both you and Chris for your
respective papers on wet and dry paths to
assemblers. I'm still digesting the papers,
more substantive comments to follow, hopefully.

I think it will be a while before you see that type of letter from someone as informed as they were. We already have moderately concrete answers to the dangers of nanotechnology, for example here.

I also agree about not counting out Russia, so long as oil prices remain moderately high. Taiwan is especially interested in nanotechnology. I would expect Singapore and Malaysia to be hotbeds of activity as well.

An assembler would create far more economic chaos in the developed countries than the less developed countries. People who don't see it coming will be surprised at how quickly the stock of traditional manufacturing companies who missed the boat will drop and how fast such companies will start cutting jobs. Less developed countries will face much less disruption.

But almost EVERYONE misses the boat! An assembler does not a nanotech revolution make. Why? Because as I point out in my Nano@Home proposal we don't have any designs! Nanorobots, for example, are 2-3 orders of magnitude more complex than assemblers. We are going to see this coming far enough in advance to prepare for it. You aren't going to wake up one day, read on the front page of the NY Times — "Working Assembler Created" and then go out the next day and start purchasing all this really cool, cheap stuff to replace the stuff in your house manufactured with outmoded macroscale methods.

Various militaries don't have it any easier than a consumer products manufacturing company. The first robust nanotechnology designs, e.g. items of a billion atoms or more, are going to require Boeing sized companies to do the designs if they want to be on the early end of the curve. That level of activity isn't going to go unnoticed.

Jeeeesh.

Hardly a month goes by now without some new governmental funding initiative for nanotechnology. So how long will it be before the equivalent of the Einstein-Szilard Letter crosses the desk of some leader and a huge military research effort to develop assemblers is launched?

I would not be so quick to rule out Russia or even some of the smaller or newly industrialized countries like Taiwan, India, Israel, Brazil, South Africa, etc. The effort may cost much less than Mr. Phoenix imagines so some of these countries should not be ruled out.

I would not be so worried if the assembler breakthrough happens first in Japan or in Europe as these states are largely stable, properous and democratic. But China, Russia and the others mentioned worry me. What if political and economic chaos arises in those countries?

Chris provides a fairly realistic summary of the probable timeline in my opinion.

First a couple of points where I disagree. He predicts the development of "nanometer-scale machines, or micrometer-scale complex molecules, within a few decades" based on technologies he lists. While I agree with the nanometer-scale point, as I think supramolecular chemists are within striking distance of molecules like the fine-motion controller, I disagree with the concept of "micrometer-scale size complex molecules". That would be a molecule the size of bacteria. Even the ribosome is only 30 nm in size. A complex micrometer-scale molecule is a nanorobot!

Chris also thinks, "the first built-from-scratch bacterium will happen within four years of the built-from-scratch virus". I'd disagree unless Robiobotics gets funding. I've studied this problem for over 3 years. Its still way too expensive for a small lab to even consider doing this, probably by 1-2 orders of magnitude.

I've done a fairly complex analysis of the "wet path" to assemblers based on engineering increasingly complex enzymes and bacterial genomes for nanopart assembly. See Protein Based Assembly of Nanoscale Parts. While designing a system to build a nanorobot now would cost trillions of dollars, there are ways to leverage other developments, particularly the increases we can expect in computing capacity to get us to the stage where companies would be designing and building nanorobots by 2015-2020. One crucial thing that is needed is for the government to fund nanotechnology centers to educate more people in enzyme design and develop "computer-aided enzyme design" and then "computer-automated enzyme design". This is similar to the path followed by the microelectronics industry over the last few decades as computers now handle all of the low level IC design tasks.

Anyone who doesn't see this should look at Table 16.1 and read section 16.4 in Nanosystems and read the papers Eric wrote in 1994 and 1995 (after Nanosystems was published).

This is one of the best short articles I've read on where we are on the road to assemblers, Chris…great stuff. What everyone here, but not everyone out there, knows, is that Real Soon Now the public is going to be confronted with the realities of proteome monitoring or freakishly strong and light materials or pervasive sensing/surveillance or advanced robotics or the consequences of terabit chips or…something else that feels like science fiction. It may be easier at that time for us to sell an assembler project as the safest choice among many scarier futures. John Boddie and I hosted a get-together for a local congressional candidate the other night, and none of the non-scientists were informed on or interested in future science…but in a more science-fictiony world, I bet we'd have an opportunity to reach them.

You make a solid analogy to secrecy and the Manhattan Project…but a difference here is, almost no one in 1943 expected that a nuclear bomb would be a significant addition to our arsenal. In 2002, nanotechnology news is pervasive, including speculation about assemblers. Any government or non-government attempting an assembler project will try their best to "hide it in plain sight", to pass it off as incremental technology advances, but the odds are small that a large number of people are going to be able to avoid giving any clue concerning such a big, juicy secret.

I think its time to start on the detailed design and simulation of an assembler. Unlike the direct experimental route (which I participated in a bit at zyvex) simulation is relatively cheap, easy, and can be done now instead of waiting for a huge government or commercial program. Feel free to comment on the Sci.nanotech thread or nanocad sourceforge list if you'd like to get involved.

John