• Google
    This Blog Web

October 2011

Sun Mon Tue Wed Thu Fri Sat
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31          

RSS Feed

Bookmark and Share

Email Feed

  • Powered by FeedBlitz

« We Love Lovy | Main | nano-techno-logy »

June 22, 2005


Feed You can follow this conversation by subscribing to the comment feed for this post.

Tom Craver

Not casting any disrespect on CRN's excellent and interesting work on nanofactories based on nanoblocks - but the more I read about them, the more obvious the gaps above and below that level become. It would be interesting to see nanoblocks taken to the next level of detail - the internals of a block, and how functional structures can be built from them. The work on nanoblock connectors is a good start. Can that be taken further?

How about designing the internals of a "simple" structural nanoblock design (i.e. one that can only be made to attach or release all sides in contact with another nanoblock - no other functionality)? Assume such a block need only be powered and controlled during assembly, and only by direct external mechanical inputs, and that only the block being inserted will have power and control. Assume some mechanical feedback is required, to insure correct positioning "by feel" prior to (or in the process of) activating connectors. Power and control might come in on one "special" side - which must still be able to have other nanoblocks attach to it.

Then, take it up a level - how could the "micro" power of an axel built of nanoblocks, be transmitted into the nano-mechanical structures of a block? How could signals be distributed mechanically through a structure to a particular addressed block? How do power and signals get transmitted through joints? How can damage in the communication network be routed around, without risk of signals getting lost or echoed through multiple paths? The mechanical approach seems "Rube Goldberg", but the point wouldn't be to make a "final" design, but rather to explore design concepts.

Chris Phoenix, CRN

First, remember that a fairly small 100-nm nanoblock contains one million cubic nanometers (each of which has space for 176 diamond atoms). So there shouldn't be any problem in fitting a lot of functionality, even "smart" functionality, into a single nanoblock. Each face has an area of 10,000 square nanometers, plenty of room for power, signal, and grippers. I'm sure some solutions to the structural-gripping-block problem will be more elegant than others, but I just don't see it as a difficult problem.

Second, remember that a human cell is about 200 times as wide as that nanoblock; 8 million times the volume. So there shouldn't be any problem in building products from lots of relatively simple nanoblocks.

To transfer power across a bearing (from the sleeve to the shaft) I might make part of the sleeve drive the shaft at a different speed, twisting it. To transfer signal, I might line up a bunch of retractable pegs around the sleeve, set them in a pattern, and let a mechanism on the shaft read them as it goes by.

When you can fit a complete microcontroller in a single nanoblock, you can treat signals as pure information--don't specify "mechanical" or anything else--just work on algorithms. To send to an addressed block, just design in a "land line" directly to it--you can put simple information pass-through channels in the "default structure" block and hardcode the switching at build time. For redundancy, the "land line" can include multiple physical channels with voting every few microns for error suppression. To talk to lots of nearby blocks, get a signal to the approximate area with a high-speed 3D network with direct spatial routing and nodes every few microns, then route locally with a simple demux. Again, there will be more elegant algorithms, but with machinery that's SIX ORDERS OF MAGNITUDE more compact (maybe even 12, depending on how you count), you can afford to be crude.


michael vassar

Sorry for re-posting, but I wanted to finish the thought/idea associated with my response to Chris's statement, (shown below) putting all the relevant statements together for ease of input.

Chris Phoenix
" by talking about the positive consequences without any acknowledgement of the source (MM), or any acknowledgement of possible negative consequences of MM (of which there are many), they are short-circuiting much-needed policy discussion."

Michael Vassar
Yes they are Chris, but it's possible that the level of nuance required for such discussion simply can't be handled by institutions of public discussion in anything like their current form.
Think of some issue like the war on drugs where policy is pathological, everyone intelligent and informed knows it's pathological, and where perfect policies may be elusive but major improvements take very little effort to conceptualize and explain. Issues where the basic science is not in doubt. Issues with centuries of historical precedent.
Then tell me how effective efforts have been to implement sane policy. Efforts backed by huge political movements. Efforts with some of the world's wealthiest people behind them. Efforts backed by a huge part of popular culture... There may be forums where rational discussion and deliberation is possible. Certainly Enlightenment thinkers like Jefferson expected a lot of rationality. Maybe what they had seen of it's effect justified their faith. Maybe the Netherlands or Denmark or New Zealand is capable of usually adjusting its behavior when said behavior is blatantly stupid, but the US isn't. I think you should seriously consider focusing effort on a populace/administration/bureaucracy with less inertia.
I want to clarify that I am not simply complaining and
saying "the government is irrational", much less "the
people are irrational". Yes people are irrational,
and we have to learn to live with this, but
governments are non-rational. They really do have
emergent properties, and so long as all of the people
within play their parts, corporations fighting for
subsidies and inviting politicians to lavish
fundraisers, local government chasing pork, leftists
marching, Christians voting and donating money,
scientist begging for grant money, etc, it can follow
it's own course. Just the fact that the majority of
the muscles in your body are contracting doesn't mean
that you will contract. Just the fact that the rear
leg is moving north doesn't mean the body is moving
south. Just the fact that the majority of interest
groups want to avoid a war doesn't mean the war will
be avoided.
I'll point out that if it was more "democratic", if
"the will of the people" prevailed, then we'd have
school prayer, creation in schools, no abortions,
socialized healthcare, destructive taxes on the
wealthy and on corporations, very little legal
immigration, regular executions, no foreign aid, etc.
The Christian right would sweep every issue because
they combine money and effort, appearent intensity of
caring. As it is, deep down we all expect them to
loose every issue, yielding gay marriage in the US
before most of Europe and eventually leading to
polygamy and other items that aren't on the table yet.
Why? Because the Evangelical Christians fight the
trends of the day. They try to move in a direction
opposite that of society. And because the rural
people always loose political battles which aren't
resolved simply by bribing them. And possibly because
they and their allies get energy from Christain anger,
so their allies make them loose and hence get angrier
and angrier and more and more powerful (no-one ever
taught them that if something isn't working you should
stop doing it).
At any rate, we should learn to stop acting like the
Evangelicals. We should learn to pay attention and
make sure we see the actual consequences of our
actions rather than just moving in the direction we'd
like society to move in. Some of the behavior of
governments can be predicted by public choice theory.
Other aspects cannot be so predicted, but even
thoughtful informal guesses may be better than blind
conformity to patterns that are known not to work.
Trying to invluence key individuals may be a better
strategy, but even key individuals may lack real
power. The Free State project is an interesting
proposal. It's an effort that hasn't already failed.
Smaller alternative communties might be a better idea.
There's still lots of land in the US. At any rate, I
hope this comment gets some reaction.

Chris Phoenix, CRN

Michael, no problem about reposting--seeing your thoughts together made it clearer what the connections are.

Your point about policy-forming mechanisms being simply incapable of making good policy in some areas is one that I'll be thinking about. Your example of the War on Drugs is weakened slightly because it does benefit some people and interests--it's not a lose-lose but a win-lose policy. Obviously it benefits drug smugglers, but I hope they aren't influential in national policy! But it also benefits people who want an excuse to exercise police powers. And it benefits people who want to convert urban-blighted property into private luxury hotels via eminent domain. (Yes, this is now legal.)

But anyway--I'm not sure whether you're saying that moving in the direction we want to go is actively counterproductive, or just not guaranteed to work. And I'm not sure what your advice is. Forming small sensible communities will not solve most of the issues that MM creates.


Tom Craver

I'm not questioning that powerfully useful nanoblocks will be possible.

I'm pointing out that, by ignoring their actual construction, we're TOO free to assume that nanoblocks with arbitrary useful functions are available, and hence we can design just about anything quite easily with all the major difficulties taken care of by the (unknown) stuff inside the nanoblocks.

But real world nanoblocks will initially be few in design, limited in their set of available capabilities, non-standardized (resulting in incompatibilities), etc. Just setting standards may be a complex task, and likely there'll be competing standards, just as there are competing computer languages and competing communication protocols.

I'm beginning to think the focus on 'mythical' (i.e. as yet unspecified as well as non-existent) nanoblocks may be part of why CRN believes MNT will have such a fast onset. The CRN scenario appears to go something like "MNT breakthrough allows building things with atomic precision; a few improvements later we have the ability to make arbitrary nanoblocks in large quantities; we design the nanoblocks we need and start easily designing things out of nanoblocks.

Missing from that scenario is "We build lots of things directly up from atoms to learn what is possible, useful, powerful, etc. We then struggle to define a broadly useful, compatible library of nanoblocks that saves us time as compared to designing things directly with atoms, without sacrificing much of the performance possible with the latter approach."

Tom Craver

A bit more on nanoblocks:

Some of your own results regarding scaling of nanofactories seem to indicate that simple rigid structure might be built up about as quick or quicker, directly from atoms.

Surfaces built this way would have atomic precision, minimizing friction and vibration, hence simplifying design as compared to designing with nanoblocks.

Strength per mass could be made near ideal for any given situation using parametric space filling algorithms to control atom placement, rather than being limited by the strength of nanoblock connectors.

Complex functionality (motors, signal routing, flexible structures, etc) would be embedded by attaching functional nanoblocks to the rigid structures only as necessary.

Sure, you use nanoblocks and get "good enough" results thanks to SIX ORDERS OF MAGNITUDE - but for what benefit over direct assembly with nanoblock functionalization?

The main potential benefit I can see is that one could implement the scheme I've described for providing a separation between a generally available nanoblock assembly capability, vs a tightly restricted atomic-precision assembly capability.


The issue of responsible MM usage does not necessarily have to be understood by politicians to be applied. It only takes one member of a MM research team to rat out the team's efforts for a project to be halted by convential police or military powers. If the safe application of MM really is dangerous and unlikely to be achieved without following a rigorous set of guidelines as various MM research benchmarks are reached, than all that is required is this danger to be understood by as many potential MM research members as possible. If universities offered brief "hazards of MM" warnings in their chemistry, solid-state physics, microbiology, and molecular CAD courses, it would be tougher for a potential MM-tyrant to assemble a loyal team of hundreds or thousands of personnel. Safe MM parties do not currently have the money or personnel to win a race to MM in the near-term. They must leverage the truth of their message somehow.

michael vassar

cdn: Universities are not GOING to include "Hazards of MM" courses. Researchers are not GOING to worry abou the ethics and applications of their work, they basically never do. We can't even stop biological weapons research and we're not going to be able to.
I assert that the power to successfully leverage the message implies power more than sufficient to win a MM race in the near term. Memetics is HARD. MM is only midling difficult. Trillions of dollars are spent on memetics (all marketing, religion, government, etc), A few million are spent on MM. MM becomes exponentially easier with time due to better tools and computers. Memetics gets harder due to cynicism and competition. Finally, Everyone here is a nerd, and while nerds very occasionally influence politics in the long term (Milton Frideman prehaps? Nietzsche), they never do so either totally or in the short term.
Please don't say that universal understanding of a simple principle among professionals is "all that's required" for something while doctors overproscribe antibiotics and fail to apply Bayes theorem to diagnosis/treatment. Don't tell me Universal understanding is all that's required when many philosophy departments are still full of theists debating questions that are simply empirical from the wrong side. And don't tell me that universal understanding of simple principles is all that's required when almost all early 20th century intellectuals were socialists if not outright Marxists and when most economics discussion as late as the 70s failed to appreciate comparitive advantage, incentives, the role of markets in aggragating information, or the value of experiment!

Chris, I don't think it's counterproductive or productive, simply non-productive, like making a convincing argument in favor of waiting for two banannas instead of eating one today is non-productive when the argument is made to a monkey, or to a person who speaks a different language, or to a person who starts from different assumptions about the role of discussion and reason in life, as almost all people do.

Tom: I think that nanoblocks can be abstractions that the compiler programs work with, they don't need to be actual blocks placed together via convergent assembly.

All: MM is clearly the pressing issue of the day. We are right to focus on it, and to ignore most other issues when they don't bear on it, but when trying to get something done, I sincerely ask that people investigate why easier things, things pursued by many good and intelligent people in the past, have failed to get done. The winners from the drug war, other than criminals, are absolutely small-scale compared to the managers and the like who "win" by maintaining the status quo. What should be done with MM is a difficult enough question that sincere and open debate over 25 years, while clarifying some issues, has basically failed to answer most others. Real political processes can't POSSIBLY, in the time remaining, do a better job than 25 years of debate among those who understood the issues. It's us or no-one guys.


Hey I'm not a nerd. I can drink all my buddies under the table, and I'll take on a bar full of bouncers to prove otherwise. Why wouldn't universities include such material in their courses? If I've got nothing better to do in 2006, I'll chat it up with staff and students at the U of Alberta and the U of Waterloo, the two Canadian universities most likely to offer MM courses in the years ahead. I think towards the end of a MM research programme, esp. at the product prototyping stage, thousands of personnel will likely be required. It is hard to recruit this many people for a responsible effort without the resources of a Nanhattan, or very wealthy person. I agree on the imperative need to win the MM race, but without money/tools/manpower, the most effective actions a bunch of visionary but poor people can take now, might be to lay the groundwork for encouraging defectors in future unknowingly suicidal MM research programmes.

Chris Phoenix, CRN


Nanofactory planar assembly scaling: It only works as long as the deposition machinery scales with the stuff being deposited. That's probably not true down to atomic level. So there's reason to want physical nanoblocks; in fact Drexler has encouraged me to think about a bit of convergent fabrication to build the blocks; I agree it's probably more efficient especially if you can mill-build some of the parts, but I don't think it's necessary.

Design of nanoblocks: I am indeed assuming that it'll be possible to design a set of useful recombinable nanoblocks fairly quickly. Quickly after what? Well, after (if not before) we have the ability to make a nanofactory and enjoy exponential manufacturing. See below.

Standardization of nanoblocks: If we judge by the electronics industry, where it takes years to agree on an extension to the DVD format, then you're right, it'll take a long time before we can design products. But that's for an industry. Think back to computers in the 70's, where every computer company designed their own bus. It would be fairly easy to route 100 uW of power and 10 MHz of bandwidth through every face of every block in a standard bus.

Flexibility of nanoblocks: Each nanoblock, regardless of purpose, can have a few hundred or even a few thousand gates of digital logic embedded in it, so it won't be hard to program its functionality. 100 uW is enough to power 10^14 gate ops per second (at 100 pS switching speed). Running the gates slower saves significant power, maybe even proportional to speed. At some point you run into bulk cooling constraints--you don't want to run thousands of gates at GHz speeds in each nanoblock in a volume. But you don't need to.

Your rendition of our claims: "MNT breakthrough allows building things with atomic precision; a few improvements later we have the ability to make arbitrary nanoblocks in large quantities; we design the nanoblocks we need and start easily designing things out of nanoblocks."

The only place I'd quibble is the "few improvements" between atomic precision and arbitrary nanoblocks. It will certainly take a lot of work to make a nanoscale molecular manufacturing system that can build its mass in a day. And that's what you need for useful exponential assembly. But once you have that, you have as much mass as you want to build, and you have a fair amount of design experience already.

So I don't think The Step will happen right after we get atomic precision. I think it will happen right after we get exponential manufacturing.


michael vassar

Chris, a minor disagreement, as far as I can tell, a nanoscale manufacturing system that can double its mass in a week or two is good enough for a pretty massive revolution if used intelligently, though one of the first things to do with it is design better assemblers. Price of feedstock is probably as important as doubling time. I'd rather use crude oil feedstock, or even 99.9% ethanol, and double in two weeks, than double in two hours and use an exotic feedstock such as the feedstocks often needed for wet nano.
As for nanoblocks, I see 3 stages to design.
1) Broadly determine criteria and make crude draft design
2) Simulate many variations and evolve improvements using computation cheap algorhythms.
3) evolve the blocks by building trillions of variations and testing them for conformity to constraints.
If 3 fails, go back to 1.

jim moore

Here is something real interesting:

"The findings of microbiologist Derek R. Lovley's research team are published in the June 23rd issue of Nature. Researchers found that the conductive structures, known as “microbial nanowires,” are produced by a novel microorganism known as Geobacter . The nanowires are incredibly fine, only 3-5 nanometers in width , but quite durable and more than a thousand times long as they are wide.

“Such long, thin conductive structures are unprecedented in biology,” said Lovley. “This completely changes our concept of how microorganisms can handle electrons, and it also seems likely that microbial nanowires could be useful materials for the development of extremely small electronic devices.”

I think that this discovery adds a lot of capability to the "soft wet" approach to nano-technology. The most likely pathway to productive nano-systems starts out wet.

Christine Peterson

CRN writes:
"A story yesterday at News.com tells us that Foresight divides the evolution of nanotech into four categories:

'[P]assive nanomaterials, such as stronger plastics; active nanomaterials, such as chemical sensors; nano devices, such as transistors; and nano systems, or complete semiconductors.'

It is hard to tell if this evolution includes integrated nanosystems in a high-performance structure like a nanofactory."

I agree that it is hard to tell from the News.com story, but surely folks at CRN know that "complete semiconductors" have little if anything to do with Foresight's concept of a nanosystem. So quoting this added noise, not information.


jim moore

There is something I have wanted to ask about the changes at the Foresight Institute, Why isn't programable manufacturing systems that operate from the nano-scale up a major (publicly stated) focus of the new Foresight?
The six areas Foresight has identified: energy, water, health, communication/computers, space, and agriculture are areas that can benefit from near and medium term advances in nano-tech. I think that these areas are terrific places to focus on, but isn't a nano-scale programable manufacturing system the critical goal needed to maximize the long term benefit of nano-technology?

Mike Treder, CRN

Hi Christine,

You wrote: "surely folks at CRN know that 'complete semiconductors' have little if anything to do with Foresight's concept of a nanosystem."

We thought we did know that. For many years, Foresight's mission statement said: "Our primary focus is on molecular nanotechnology: the coming ability to build materials and products with atomic precision. The development of this technology has broad implications for the future of our civilization."

But the new version of Foresight, as presented on your home page and in the recent press release, makes little (if any) mention of MNT, molecular manufacturing, or programmable productive nanomachine systems.

Can you clarify if the Foresight roadmap initiative is intended to lay out a detailed technical understanding of how and when molecular manufacturing could be developed? If that is the case, then CRN is firmly behind the effort.

On the other hand, if Foresight's new vision is focused only on nanoscale technologies and their beneficial applications, we also support that. It's a worthy role to take.

But if the mandate of the roadmap project falls short of mapping all the steps that will lead to general-purpose molecular manufacturing, then it does not fulfill CRN's greater hopes. It woud be encouraging if you can tell us that our doubts are misplaced, because that work is essential and urgent.


Chris Phoenix, CRN

Michael, it takes about 50 doublings to go from a 100-nm cube to a 1-cm cube. If you can double in a week, then initial scaleup takes a year. You'd do better to spend ten months working on a system that can double in a day, then 50 days scaling up. So it's not that a week is useless if you can't do better, but I expect that any week-doubling system would be rapidly obsoleted by a day-doubling system.


Chris Phoenix, CRN

Jim, the fact that conductive nanowires can be built and used by bacteria is pretty cool. But so far, it only seems to be known in one type of bacteria. It'll take some work to learn how it works, how to tweak it, how to splice it into other bacteria... and even then, it's not immediately obvious what else it can be used for.

How long do you think it would take for us to design, say, a Geobacter derivative that contained an electronic circuit and connections to implement a high-speed cellular automaton? If we could do that quickly, that could lead to a powerful approach to computing. But we have no clue, AFAIK, how to make bacteria grow digital logic molecules. We now have the wires, but we still don't have the transistors.


jim moore

if you go to
You will see that the Geobactor is being used for Bio-remediation and microbe powered fuel cells.

Chris Phoenix, CRN

When I said "not immediately obvious what else it can be used for" I was thinking in terms of enabling technologies for MM. There's no doubt Geobacter is a very cool and useful organism.


The comments to this entry are closed.