Slashdot Mirror

Please see: tropostats here and here.

It's a problem today, but in 20-50 years, solar-powered nanomachines will suck carbon from the air, creating cheap diamond bricks that could be used for building material.

http://www.imm.org/
http://www.foresight.org/
http://www.khanacademy.org/

For those who actually understand real molecular nanotechnology, aka "Drexlerian" nanotechnology, you may understand that one of the real "breakthroughs" comes when you can computationally simulate the function of a 4 to 8 million atom molecular nanoassembler. Because if you can simulate one and prove that it does not violate any laws of physics then one of the classical oppositions to real molecular nanotechnology falls [1]. The argument transitions entirely from "it can't work" (common among people oriented towards "dissing" nanotech) to "you can't build one" . And as DRM, the iPhone restrictions, etc. have all shown "can't" is very swampy territory to wade into.

Now, I know if I've got 8 million cores, such a simulation is probably feasible (and presumably bandwidth limited by hypertransport data transfer rates) so the question transitions to how many atoms can one core handle and that in turn transitions to how effective the instruction set is at performing the math required for molecular dynamics simulations. So, is SSE5 any better than this or should I be lobbying AMD for SSE6 which is explicitly targeted at molecular dynamics simulations? It is not the market for business computing but it is the market that potentially millions of "nanoengineers" will fall into.

It also goes without saying that the chip manufacturers and ubergamers and SecondLife participants all have a high interest in achieving this because pushing below ~32nm using current technology is going to get very dicey at which point Moore's Law is going to have to shift from bulk atom assembly (current lithography methods) to precision atom assembly (real molecular nanoassembly).

1. There is a third argument against the simulation of a molecular nanoassembler. The argument that an atom specific design for a 4-8 million atom nanoassembler does not currently exist. The best one can point to is a few thousand atom Fine Motion Controller (http://www.imm.org/research/parts/controller/) designed by Drexler and Merkle. However the Nanoengineer software (http://www.nanoengineer-1.com/content/) from Nanorex allows one to design elements of an actual nanoassembler. If even a mere one thousand /. readers were to add 1 atom a day to the design in a distributed open source NanoAtHome.org (http://www.nanoathome.org/) type project -- the design would be complete within 1-2 years (there is a significant amount of redundancy and therefore human intellect amplification in the atom placement in a nanoassembler). You can't simulate it without designing it first -- but if one can design 400 million transistor microprocessors then designing an 8 million atom nanoassembler shouldn't be that difficult.

I recall an argument between Drexler and Smalley in which the "fat fingers" and "sticky fingers" problems mean that molecular assemblers are impossible. My impression is that the Van der Waals force drives the sticky fingers problem in which a nanobot's finger will stick to an atom it is trying to manipulate. That seems to be quite an overreaching statement anyway, but considering that the "fat fingers" problem is on shaky ground according to Drexler, and now we have the possibility of nanoscale repulsion, it seems that the sticky fingers argument is also on shaky ground.

From

http://www.imm.org/PNAS.html:
Proc. Natl. Acad. Sci. USA
Vol. 78, No. 9, pp. 5275-5278, September 1981
Chemistry section
Molecular engineering:
An approach to the development of general capabilities for molecular manipulation

K. Eric Drexler
Space Systems Laboratory, Massachusetts Institute of Technology,
Cambridge, Massachusetts 02139

Communicated by Arthur Kantrowitz, June 4, 1981 .... What can be built with these tools? Gene synthesis (3) and recombinant DNA technology can direct the ribosomal machinery of bacteria to produce novel proteins, which can serve as components of larger molecular structures....

natural scientists seek a more general understanding than design engineers require. Science seeks the ability to predict the conformations of all natural polypeptides. In attempting this, protein chemists can search for a minimum-energy chain conformation (in hope that the protein assumes not a local but a global minimum-energy conformation) (6) or can attempt to follow the chain-folding mechanism to find the final conformation (7). Prediction will be easier if the natural conformation has outstanding stability or if its folding mechanism proceeds in a sequence of strongly preferred steps....

Engineers (in contrast to scientists) need not seek to understand all proteins but only enough to produce useful systems in a reasonable number of attempts.... Through use of strategically placed charged groups, polar groups, disulfide bonds, hydrogen bonds, and hydrophobic groups, the engineer should be able to design proteins that not only fold predictably to a stable structure (sometimes) but that serve a planned function as well.

***

PROTEIN ENGINEERING:
A 1988 view of some 1981 predictions
K. Eric Drexler
Visiting Scholar, Stanford University. Box 60775, Palo Alto, CA 94306

A 1981 paper [1] discussed de novo protein design as part of a long-term strategy for developing complex molecular devices and systems. It presented arguments against the view that the fold-design problem is an extension of the classical (and still unsolved) fold-prediction problem (i.e., predicting folds from sequences without homologous models), a view which has discouraged efforts at design.

Fold prediction is a scientific problem: it must deal with naturally evolved sequences, but natural selection's 'design goals' enforce only the physical reliability of folding -- not its human predictability. This results in folds of only minimal stability. Fold design, in contrast, is an engineering problem. Protein engineers, exploiting their freedom of design, can work with sequences artificially selected for superior predictability and stability of folding. These observations indicated that "the difficulties encountered in predicting the conformations of natural proteins do not seem insurmountable obstacles to protein engineering" [1].

In accord with the implications of this argument, we have seen the successful, de novo design of a globular protein (alpha-4) [2,3] while the classical fold prediction problem remains unsolved [4]. Likewise confirmed has been the suggestion that design can increase protein stability beyond that enforced by natural selection. In recent years, deliberate single-residue modifications have raised protein stabilities through a variety of mechanisms [5,6]. Owing to design choices consistently biased toward stability, the protein alpha-4 has a stability of 22 kcal/mole, substantially greater than the 4-9 kcal/mole of typical natural proteins of similar size [3].

Successful protein engineering marks a milestone in a research agenda leading toward capabilities of broad technological significance [1,7].

Full Disclosure: I'm a Senior Associate with the Institute for Molecular Manufacturing http://imm.org/.

I have to say that this article seriously misses the mark.

Recombinant DNA research self-regulation has been in place for 30 years now, and it has worked very well to prevent "Andromeda Strain" style accidents. The most recent full overhaul was in 1994:

http://www4.od.nih.gov/oba/rac/guidelines/guidelin es.html

There are people who are holding debates about similar regulation for molecular nanotechnology already: The National Nanotechnology Initiative http://www.nano.gov/, The Foresight Institute http://foresight.org/, The International Council on Nanotechnology http://icon.rice.edu/, and many others, including the IMM. The intent of these organizations is to establish guidelines for developement of nanotechnology, and to explore applications.

Here is the first set of guidelines which have been established:

http://imm.org/guidelines/current.html

I fully expect that this will be updated, as the technologies involved become more capable.

A good analysis of the actual societal implications is available from NNI here:

http://www.nano.gov/html/facts/society.html

Don't blow things out of proportion until they are actually implemented; the amount of regulation of any technology has historically always been as much or even much more than was necessary at the time.

-- Terry

Full Disclosure: I'm a Senior Associate with the Institute for Molecular Manufacturing http://imm.org/.

I have to say that this article seriously misses the mark.

Recombinant DNA research self-regulation has been in place for 30 years now, and it has worked very well to prevent "Andromeda Strain" style accidents. The most recent full overhaul was in 1994:

http://www4.od.nih.gov/oba/rac/guidelines/guidelin es.html

There are people who are holding debates about similar regulation for molecular nanotechnology already: The National Nanotechnology Initiative http://www.nano.gov/, The Foresight Institute http://foresight.org/, The International Council on Nanotechnology http://icon.rice.edu/, and many others, including the IMM. The intent of these organizations is to establish guidelines for developement of nanotechnology, and to explore applications.

Here is the first set of guidelines which have been established:

http://imm.org/guidelines/current.html

I fully expect that this will be updated, as the technologies involved become more capable.

A good analysis of the actual societal implications is available from NNI here:

http://www.nano.gov/html/facts/society.html

Don't blow things out of proportion until they are actually implemented; the amount of regulation of any technology has historically always been as much or even much more than was necessary at the time.

-- Terry

It really is amazing to live in a time of such progress and have the means to observe it, and occasionally participate.

The sooner molecular manufacturing is advanced, the sooner global poverty, and most economic inequality can actually be eliminated. You can donate a fish today, or donate towards the tech that can assemble a fish using free solar and recycled molecules in the middle of the desert...

(Anyway, I just had to add you to my friends-list for being in the position you're in without also having the exessively-greedy me-me-me mentality.)

--

So here's an idea. Put a captured asteroid into an elliptical orbit. Perigee is at about 200 miles, going about 10 km/sec, apogee is at about 18000 miles going about 1900 km/sec. As the asteroid approaches perigee, it lowers a cable (made of space-elevator rope) into the upper atmosphere. As the cable gets into the atmosphere, the asteroid starts paying it out very fast, so that the end moves slow enough to be grabbed by a high-altitude airplane and attached to a spaceship. Once attached, the asteroid pays out cable slower and slower, accelerating the spaceship to the asteroid's velocity, and very slightly slowing the asteroid in its orbit. Eventually the asteroid starts reeling in the cable faster and faster, accelerating the spaceship further.

The spaceship only needs to be accelerated a little past the asteroid's velocity to reach escape velocity. There are a few possible ways to correct the energy loss of the asteroid's orbit. The simplest is for the airplane to attach a fuel tank to the cable along with the spaceship so that after the spaceship detaches, the asteroid can reel in the fuel and do a burn to pump its orbit back up.

Of course there's a big PR battle to be fought, to make people feel good about a big rock in a relatively low orbit over the earth. But if it worked, it would use a lot less rope than the space elevator, and it would get you into space quicker.

For humans, J. Storrs-Hall (of sci.nanotech fame) proposed a space railway that could be built sooner and more cheaply than a space elevator. It's a linear induction motor laid along a 300km-long track, 100km above the ground, where the atmosphere is thin enough to take a few orbits to decay your orbit. You drive your spaceship up a ramp to one end, and the motor accelerates you along the railway at about 10G for about 90 seconds, putting you in a slightly elliptical orbit with an apogee on the other side of the Earth. When you hit apogee, you do a burn to get into a higher orbit.

Relatively little radiation because you cross the Van Allen belts much faster. You get to LEO without burning any of your own fuel, which is a big energy win. The railway is low enough that orbits still decay slowly, so there's no space junk to worry about at that altitude.

The structure is a collection of A-frames, built like a radio tower. Like the space elevator, only a tiny fraction of the height is subjected to significant weather. The structure is under compression, not tension, which widens the choice of materials. According to Storrs-Hall, existing synthetic diamond would be suitable.

For humans, J. Storrs-Hall (of sci.nanotech fame) proposed a space railway that could be built sooner and more cheaply than a space elevator. It's a linear induction motor laid along a 300km-long track, 100km above the ground, where the atmosphere is thin enough to take a few orbits to decay your orbit. You drive your spaceship up a ramp to one end, and the motor accelerates you along the railway at about 10G for about 90 seconds, putting you in a slightly elliptical orbit with an apogee on the other side of the Earth. When you hit apogee, you do a burn to get into a higher orbit.

Relatively little radiation because you cross the Van Allen belts much faster. You get to LEO without burning any of your own fuel, which is a big energy win. The railway is low enough that orbits still decay slowly, so there's no space junk to worry about at that altitude.

The structure is a collection of A-frames, built like a radio tower. Like the space elevator, only a tiny fraction of the height is subjected to significant weather. The structure is under compression, not tension, which widens the choice of materials. According to Storrs-Hall, existing synthetic diamond would be suitable.

In the long run one would like to be able to get such simulations from the 10,000 atom level up to the billion-to-trillion (or more) atom level so you could simulate significant fractions of the volume of cells. Between now and then molecular biologists, geneticists, bioinformaticians, etc. would be happy if we could just get to the level of accurate folding (Folding@Home is working on this from a distributed standpoint) and eventually to be able to model protein-protein interactions so we can figure out how things like DNA repair -- which involves 130+ proteins cooperating in very complex ways -- operate so we can better understand the causes of cancer and aging.

But AFAIK no brain damage from diamond has been reported

That's because diamonds don't get flushed down the drain, and if they did they would sink to the bottom of the lake and become part of the "muck".

If you Read The Fine Article, that's what the scientists thought would happen to the buckyballs. But in tests they remained suspended in the water and fish and small crustaceans became exposed and subsequently were affected.

There are a couple of other things to remember. Diamond is a crystalline form of carbon, which does make it inert, as other atoms are not attracted to form bonds with it. Buckyball molecules do not have this lattice structure, and are going to be more reactive. Here is a tutorail on the different aspects of carbon chemistry.

There are industrial processes that use diamond (like saws), and the resultant powder can be dangerous. But this is the case for any fine powder that might be inhaled, and the toxicity is going to be dependant upon the powder.

But generally, these are "microparticles", not "nanoparticles", which may react differently in a biological system. Being a magnitude smaller, they will by their nature tend to stay afloat longer. Rather than "clump together" and sink like other particles would.

Here is a study about diamond's biocompatibility.

Their conclusion - "Thus it appears that diamond is extremely -- indeed outstandingly -- biocompatible with living cells."

Foresight Institute

Institute for Molecular Manufacturing

Singularity Institute for AI

--

Do you have any links to any NanoTech think tanks or anything which talks about your claims further? I couldn't find any critisms on any nanotech websites

The primary MNT websites (primary because the people who really invented nanotech work there) are those of the Foresight Institute and the Institute for Molecular Manufacturing. Their people are the editors of the Slashdot-like MNT discussion site Nanodot. Here are a couple of news articles they've posted there about the NNI funding issue:

TNT Weekly: deletion of MNT study from nano bill is "a farce"

Nanobusiness Alliance spokesman attacks MNT

And that foundation will also carryover into a Free Hardware world once "molecular manufacturing" tech allows nerds (and regular folks) to manipulate atoms like bits. The big difference in that world is that there'll be far fewer greedy people complaining about not being able to "put food on their solid-redwood table" once the food and the table it's sitting on can be "copied" almost as easily and cheaply as software.

Want a GNU/Burger? (then use some solar energy to power your "GNU/anything-box" to rearrange the infinitely-recyclable molecules sitting in your trash.)

--

I think we'll cultivate a lot more good will if we send "oreos" to Mars. Especially double-stuff.

A space elevator wouldn't hurt either.

I used to think the space elevator was a silly idea, or at least a not-any-time-soon idea, and that we should tinker with tethers or J. Storrs-Hall's space dock idea in the nearer term. But I started reading the info (most of the technical issues are treated in essays in the "Downloads" section of the website) and it's remarkable how thoroughly this guy has considered every possible aspect or risk of the system.

We don't need a lot of new science and technology to do this. We need better nanotube manufacturing and we need an automated system that can shoot down flying garbage in LEO. Aside from that, it's just money and politics. It would mean the end of shuttle disasters, and every space activity would be 10x or 100x cheaper than it is today.

Interestingly, the space elevator guy has recently taken a position at this think tank where he says he'll have the resources to really put up a space elevator. I hope things work out for him. If he succeeds, we'll all be better off.

Sorry for the failure to appropriately curb my enthusiasm. I'll try to do better next time.

Tethers ( 1, 2, 3 ) attached to counterweights can be used to transfer spacecraft from one orbit to another. The first tether has an orbit that skims the atmosphere, where a craft catches and connects to the end of the tether. The craft is lifted into low earth orbit and subsequent tethers help it to reach escape velocity. Using the tethers takes energy out of the orbits of the counterweights, some of which can be put back by using the tethers for descent as well as launch.

J. Storrs-Hall (once moderator of sci.nanotech) envisioned a space dock, a linear motor suspended 100 km above the ground that accelerates spacecraft to an elliptical orbit. He computes an amortized cost of reaching low earth orbit of 42 cents per kilogram. From the elliptical orbit, it's a relatively small safe step to escape velocity.

A space elevator ( 1, 2 ) is an excellent long-term solution. A cable is hung from a weight in geosynchronous orbit, reaching down to the Earth's surface. The elevator climbs the cable, carrying a craft. When it reaches GEO, the craft detaches and spends only a little fuel getting to escape velocity.

Tethers and the space elevator require novel materials for strong cables, probably using carbon nanotubes. The frame to hold up the space dock is in compression, and something we could build with little or no advance in material science. Any of these alternatives would be vastly cheaper and vastly safer than putting human lives on the noses of fuel tanks subjected to unreasonable speeds and stresses.

Video phones exist and are actually reasonably common, especially in business circumstances. The reason most home users don't see them though is that for the most part the quality is sorely lacking, and they're often way too expensive...

a quick search on google netted me this:
a home videophone...
another home video phone...
and, for what appears to be the prevailing standard: h.232

molecular manufacturing is a bit of a different story, but:
a group devoted to molecular manufacturing
some interesting stuff on it
and, last but not least:
IBM does some cool stuff sometimes

hope this helps dispel your mistrust of tech previews (Although i'll admit that at least a grain or two of salt is warranted in many occasions)

There are some variations on the idea though,like this one, that are close to being possible with today's technology, and can even be provisionally costed. Basically the idea is to construct an elevated runway about 100km up, and use mass drivers to hurl stuff into orbit. At that altitude the saving from air resistance is huge and mass drivers become very efficient

At this stage, NASA speanding serious time thinking about space elevators is probably no more useful than daydreaming. Thinking about this kind of thing is probably more productiove though, becuase something might come of it in the medium term, and its almost as efficient as an evelator anyway - with the decided advantage of not being able to collapse and strangle the planet.

(Since I heard about this from a NASA researcher, maybe Im being a little harsh to accuse them of daydreaming)

I've recently finished a detailed analysis of what is required to achieve the full vision of molecular nanotechnology via the wet (biotechnology enabled) path (in contrast to the dry path being pursued by Zyvex). It will require significant improvements in both computer capacity and tools for the computer-assisted, and eventually automated, design of enzymes. Currently our abilities to design enzymes is limited, but we can expect these capabilities to increase significantly within the current decade. Within the period from 2010-2020, the costs for the design of assembly lines for nanoscale parts should fall low enough that the design and assembly of nanorobots should become feasible. So Drexler's estimates may yet prove to be right on the money.

There are two problems the chemists have. First, they haven't read the material. Second, Drexler is proposing to precisely assemble millions to billions of atoms and the chemists think that is infeasible. That is why programmers can accept nanotechnology to a greater degree than chemists -- manipulating a million or a billion "bits" is something they regularly have to deal with. For chemists the idea is nightmare.

I'd urge readers to educate themselves with regard to the material before they comment on it. If we have to spend all of our time attempting to erase malformed memes we will never get a chance to work on developing new ones.

Current efforts in nanotechnology are not directed towards making "analogs" of things "found in nature with elements not commonly used in nature". Current efforts are directed towards using the laws of physics and chemistry to explore regions of the phase space for atomic structures that are by and large unexplored by nature. Zyvex wants to assemble diamondoid materials -- that isn't a different element, its a different way of putting carbon atoms together than that commonly used by nature. Nature primarily assembles polymers (DNA, RNA & Proteins) which involve creating 2 covalent bonds -- molecular nanotechnology seeks to create more rigid, stronger materials by controlling the creation of 3-4 covalent bonds.

The development of nanotechnology is likely because of the "existance proofs" provided by nature. The development of "picotechnology" is highly speculative, as documented by Hans Moravec in Harvard Doesn't Publish Science Fiction.

It would be nice if people really knew something about a topic before they commented on it.

But what about when machines have been given control over their own replication.

See here, here and here.

I would like to think that humans could at least limit the amount of technology and capability that goes into its creations. But the fact remains, if we give the machines the ability to procreate and self-replicate, we are also giving them the ability of evolution, which once given will conceivably allow for the time when humans will be rendered obsolete.

BTW, I believe that human/machine/computer integration is inescapable due to what might be just another stage in our evolution.

xen

A few years back, John Storrs-Hall (for many years the moderator of sci.nanotech) was talking about an interesting idea that, like the space elevator, is not very far beyond existing material science. It is also probably more economical. The gist is an airport runway, 300 km long and at an altitude of 100 km, with a built-in linear motor that can accelerate a spacecraft. Over 80 seconds at 10 G, the craft accelerates to 8 km/sec, necessary to maintain a circular orbit. Humans (at least young healthy ones) can survive this acceleration. Current approaches to space launch cost around $10,000 per kilogram. The space dock could allow launches for 91 cents per kilogram, dropping to 42 cents per kilogram as the construction was amortized over the first few decades of use.

The space ramp is about 100 km high and 300 km long, with a linear induction motor running its length. You take an elevator up to one end, hop on the induction motor, and get accelerated at 10 G for about 80 seconds. This puts you in low-earth orbit, at an amortized cost of 42 cents per kg. Prior to amortization, you'd be paying about a buck a kilogram. The cost for space shuttle launches is about $10K per kilogram.

If your question was genuine and you seriously think that we are anywhere near to reaching practical limits on physical merging of appliances then you badly need to spend a few months reading something about nanotechnology, its near-term impact on molecular manufacturing, some wonderfully readable and seminal insight on where it might lead, and if you want more depth, a key text book in this area.

We are on rung 1 of a ladder that extends into infinity. The idea that somebody on a nerd forum could ask a question even suggesting that today's primitive toys are anywhere within a million light years of effective limits in any respect whatsoever is mind-boggling.

J. Storrs-Hall, until recently the moderator of the sci.nanotech newsgroup, wrote an interesting proposal for what he calls a space dock. It's a platform 300 km long, at a height of 100 km above sea level, where air drag is much smaller. Your spaceship would ride an elevator to get up to the platform, and once there, a linear motor would accelerate it (at 10 G's for 80 seconds, survivable for humans) into circular orbit (8 km/sec), from which it's relatively easy to hit escape velocity.

This does not require nanotechnology. It would be possible (albeit initially expensive) to do it with existing materials and techniques. Once the construction is amortized, the total energy cost of putting a kilogram in orbit (elevator plus linear motor) is 43 cents. With hourly launches, it would be possible to amortize the cost of construction by charging about a dollar per kilogram.

In estimating cost of construction, JoSH writes: The wildcard is the cost of the diamond (and the ability to fabricate it into structural beams). Diamond is a bit expensive today! If an Apollo style (and -cost) project could do for diamond what the original one did for electronics, we could build the tower in the next decade or so, and with regard to near-term feasibility he writes: Even commercially available polycrystalline synthetic diamond with advertised strengths of 5 GPa would work.

The problems mentioned by Bill Joy in his interview point out how poorly informed he is. Anyone who has been in the computer industry as long as he has, should know enough to "read the manual(s)" before offering uninformed opinions. The problems regarding nanotechnology run amok have been discussed for many years in the sci.nanotech newsgroups as well as at conferences for the Foresight Institute's Senior Associates. The basic solutions involve making "safe" (e.g. reviewed, open source) designs available while at the same time developing defenses against nanotech run amok. The Extropy Institute's Mailing List Archives, for example, contains recent discussions about encouraging the availability of "almost anything" manufacturing boxes (similar to Star Trek "replicators"), while discouraging the availability of "everything" boxes.

Diamondoid or saphire based molecularly assembled nanobots used in medical applications will greatly exceed the capabilities in of "biobots" built on existing genetic machines (DNA, enzymes, bacteria, cells, etc.) because they are stronger, can pack the "code" more densely, and can have more complex programs than the rather "ad hoc" designs that nature has provided us with. Most of the first volume of Nanomedicine is devoted to determining exactly what the physical limits will be on power, communication, mobility, etc. Most of the applications will be discussed in Volumes II and III.

Joy may be right that the technology poses a threat to the "human species", but that begs the question of "Why would you want to run on obsolete hardware?". Anyone who understands even a little astronomy knows that galactic hazards doom biological human forms to death at some point. Only those humans who choose to upload have any hope of living the trillion or so years that seems quite feasible. So while the hopes for biochemical humans are rather dismal even with Nanomedicine, the long term prospects for humanity, based on what nanotechnology allows are quite good indeed.

As far as nanotechnology background material goes, the best (nontechnical) source is Engines of Creation. Other references can be found in Eric Drexler's CV.

Start with the Foresight Institute .
Make sure you read the Nanotech Study Guide .
Then go to the Institute for Molecular Manufacturing .
Also look at Zyvex , a company founded to develop molecular nanotechnology.
For fun, read Neal Stephenson's DIAMOND AGE and Michael Flynn's NANOTECH CHRONICLES .

Good luck.

how can you make a robot smaller than the smallest possible computer core?

Maybe you can find something better than circuits etched on silicon surfaces. Tom Knight at MIT is looking at how to get bacteria to perform useful computations, using genetic engineering methods that have become well understood. You can mail-order custom DNA sequences, graft them into cells, and get the ribosomes to synthesize the proteins you want, if you're smart enough to design proteins. Eric Drexler, generally recognized as the guy who formulated the concept of nanotech, wrote one of his early papers on the possibility of engineering proteins as a step to a more complete form of nanotechnology.

The nanotechnology literature (see Engines of Creation and Unbounding the Future) talks about placing atoms at specific locations as you build up a molecular machine incrementally, in a process called mechanosynthesis. If this works (and I'm not aware of any technically sound arguments that it wouldn't), it might become possible to build almost any object whose existence doesn't violate the laws of physics. At least, it would become possible to build a lot of different things we can't build today.

how the hell do you tell them what to do? ...

how big are these things going to have to start out, since the first generation must contain all sets of code for all generations of nanomachines?

There is a lot of excellent information about nanotechnology at Ralph Merkle's site at Xerox PARC.