• Google
    This Blog Web

October 2011

Sun Mon Tue Wed Thu Fri Sat
            1
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

« BBC Interview Tonight | Main | Press Coverage »

June 09, 2004

Comments

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

Richard Jones

Austin's device sorts 800 nm particles with gaps of order 1.6 microns. Should we expect it to be compromised too much by Brownian motion? Let's do the sum. The time taken for a 800 nm particle to diffuse its own diameter is roughly 2 seconds, which is comparable to the residence time of the particle in the device (40 seconds overall). And indeed the paper reports that making the residence time faster on this timescale does indeed make the resolution better, so the sum tells us why the device works, despite Brownian motion, but not by a huge margin.

Now ask how well this device will work if we shrink it to a true nanoscale, say for a 1 nm particle. Let's do the sum: the time taken for a particle to diffuse its own diameter by Brownian motion scales as (diameter)^3. A good measure of the importance of Brownian motion in a device like this is the ratio of the time taken for the particle to diffuse its own diameter divided by the residence time of the particle in the device. For a fixed applied pressure the residence time is actually independent of size for laminar flow. So Brownian motion is 500 million times more important for the nanoscale device than for the micron-scale device.

This isn't a question to be resolved by appeals either to rhetoric or conventional wisdom. You've just got to do the sums. These sums reveal, of course, that Brownian motion gets much more important as your system gets smaller. This is why elegant nanoscale design will exploit Brownian motion rather than trying to design around it.

Mike Deering

I would like to register a complaint about some features of your plan that I find too restrictive. I think you have gone well beyond what is necessary for safety and moved into the realm of politics. By this I mean, making recommendations designed to preserve the status quo in economic, political, or social structures is a political action unrelated to safety.

1. The idea that the nanofactory must check with a central authority before producing any product. I find this restriction unacceptable on the grounds that it makes the individual completely dependent and vulnerable to the central authority. It should be possible to make nanofactories smart enough to be able to approve some classes of products without checking with or reporting to the central authority. The nanofactory should not disable itself in the extended absence of communication with the central authority.

2. The idea that nanoblocks would be produced in central factories and supplied to the population by the central authority or an economic infrastructure, and the nanofactory would merely assemble the pre-made nanoblocks into products. This is objectionable because it makes the user dependent on the central factory which may decide to charge money for or some other unfair method of rationing the supply. This is not a nanofactory, but rather a microfactory.

3. The idea that nanofactories would not be able to utilize available natural raw materials. One of the central features and advantages of nanotechnology is its ability to disassemble any raw materials or recycle any unneeded product. The nanofactory needs to have a section that disassembles, or a separate disassembler needs to be supplied to produce the purified stocks used by the nanofactory.

4. The idea that there can be only one supplier, designer, producer, of nanofactories. Whatever regulatory system is eventually established can as easily be applied to multiple nanofactory suppliers. Monopoly stifles innovation. Remember when only the phone company could produce telephones? Different companies will produce nanofactories with different features, sizes, interfaces, appearances, optimizations, capabilities.

5. And finally an editorial note: To state that some of the restrictions mentioned in your documents may be excessive and counterproductive but not to clearly label which ones they are leaves you open to criticism. To say you are not endorsing everything in your writings is disingenuous unless you plainly separate the parts you do and do not recommend.

Janessa Ravenwood

But Mike, you've said you're all for AI's taking over and keeping us around as pets. Why is it you like that authoritarian scheme and not this one?

Mike Deering

I'm sorry you got that impression. That's not exactly what I meant. I think if we remain un-upgraded humans that A.I.'s will inevitably take over, but that is not my first choice. Like CRN, it is a possibility I mention but don't endorse. My first choice is for us to have the freedom to innovate in the fields of nanotech, A.I. and molecular biotech, so that those of us that want to, can improve our capabilities, intellectual as well as physical, and therefore insure our own self actualization and the continuation of the post-human species. But we can't do that if a central authority controls all technological innovation. CRN's plan to achieve security through placation will not work. I can guarantee that, unless I am in jail, there will be at lease one independent MM development effort.

Janessa Ravenwood

Ah, well that's different. And good to hear! Also, you will NOT be alone. I'm sure that those people oppose a "single central total authority" scheme are (or will be, when this all gets closer to coming about) legion.

jim moore

"Elegant nano-scale design will exploit Brownian " Richard Jones

I guess the real question is how to exploit the Brownian motion.
Drexler's approach to nano-system design is to use very precise and very stiff ( and potentially complex) three D solids that are vibrating in a vacuum. It appears that the main way Drexler uses Brownian motion is as a "lubricant" to keep the parts from locking up.

Now nature works with a very different nano-scale environment. Its not stiff solids vibrating in a vacuum its soft flexible solids jiggling, flipping and flopping in a constrained liquid.
Now Richard, am I correct in thinking that Brownian motion causes the proteins to flip and flop, and that the conformational changes in the protein is how the protein does its "work"?

(by the way thanks for your comments on the "wet" side I always learn something from them.)

jim moore

Mike,
1.) I agree, at most nano-factories should need to use approved designs. You should not need to seek approval for every use of your nano-factory.

2.) I disagree, I think that there are many good reasons to use pre-made nano-blocks.
a) It reduces the time it take to make an object (from an hour to a minute).
b) The nano-factory uses much less energy.
c) By centrally producing the nano-blocks it allows you to keep track of how much nano-stuff has been created.
d.) Makes the sudden generation of large masses of nano weapons much more difficult.
e) Makes run away replicators almost impossible.
f) Allows you to incorporate safety features into the nano-blocks.
g) The dependancy that dislike can also provide the motivation for inter group (international) cooperation.

3.) Two points here, I completely agree that we should have nano-recyclers that break down products of nano-factories into their component block for later reuse.
I don't think anything would stop you from using a product of a nano-factory to process local material. But on earth we probably do not want everyone to have the ability to turn locally available material into very large amounts of nano-stuff.

4.) If all of the different designs of the nano-factories used pre-made nano-blocks I think we be both secure and have a good variety.
I also think you could have a variety of nano-block producers but these organizations would need to be monitored and regulated.

Mike Deering

Jim, there are going to be thousands of different nanoblocks, maybe tens of thousands. Making the nanoblocks in a factory and selling us kits of nanoblocks for specific products completely eliminates all of the advantages of having a nanofactory. Why not just sell us the product? It will be like the MM revolution never happened. We will all still have to go to work every day, if we can find jobs, to make the money to buy the necessities that would be freely available at our fingertips if the technology were within our individual control.

I agree that there are risks and some level of restrictions seems reasonable, but the level of centralized control envisioned in CRN's plan is completely out of proportion to the threat, and way beyond what is required for safety. No, these measures are designed to maintain the economic and political stratification under the guise of safety and security. What this plan does is insure that the elite remain on top and the oppressed remain oppressed.

Karl Gallagher

Y'know, there's two different aspects of the MNT revolution that should be considered seperately. One is the possibility of kitchen-table manufacturing, which is a big revolution. The other is that we're going to have a huge improvement in capability and cost of almost all products. So even without decentralizing production we'd have big changes and improvements in peoples lives as we get much more product for our money. That'll include radical new capabilities such as medical treatments and security devices, which are just as revolutionary if you have to go to the mall for them as if you make them at home. This will change society before the issue of how to regulate home nanofacs ever comes up. Regulating businesses with them will be an issue from the beginning.

Mike Deering

Karl, I agree completely.   I'm sure, in the capitalistic free enterprise environment we currently occupy, the very first implementations of MM will not be home nanofactories.   It will be industrial nanofactories producing vastly improved products.

Firstly, hugely powerful yet cheap computers.   Next, consumer electronics: cell phones, TV's, portable music and video players.   Last will be large products: household appliances, furniture, cars, houses, and buildings.   And somewhere in the middle of all this the development of increasingly intelligent humanoid robots.

The advances in A.I. made possible by increasingly powerful and cheap computers combined with MM will rapidly transform the economy to a crisis point where companies can produce wonderful products at almost no cost and with no human physical or intellectual labor for a population that has no money because all the jobs have disappeared.   To think that our current economic model, of producers and consumers exchanging labor for products by way of currency, will continue indefinitely, is very short sighted.

We don't presently pay for air because there is no scarcity.   Nanotechnology makes it possible to remove the scarcity from other necessities and even luxuries if we can avoid imposing an artificial scarcity by holding onto outmoded ideas of economy and wealth.   We need to design social systems to allow individuals independence from economic issues.   One way to do this is through the distribution of the means of production, nanofactories.   Another way, less efficient, is through centralized production and free distribution of product.   The idea of the home nanofactory is a step in the direction of decentralized production.   Marshall Brain in his Manna story describes a community based on centralized production and free distribution.

Janessa Ravenwood

The "centralized production and free distribution" model sounds distressingly like Communism, which I think history has shown us at this point to be a VERY bad idea.

Like the DE-centralized one, though.

Karl Gallagher

Mike, I'm not getting paid for handing someone an airplane, or even a part. I get paid for writing up a description of how part of it should work. "Knowledge work", or producing IP, or services, whatever you want to call it, is a steadily increasing part of the economy. Producing actual physical objects is already shrinking down toward the small fraction of the workforce that works on farms, which used to be nearly everyone. Now we don't have a scarcity of food and people get paid for preparing or serving it in attractive ways. Hardware manufacturing is going down that same curve.

Unless we're reduced to pets we're not going to escape economics any more than we will gravity.

Brett Bellmore

Agreed. I get paid for designing things, too, and that's going to remain the case even if they're manufactured by nanofactories. AI might change that, eventually, but will probably come AFTER nanotech is available to provide the hardware platforms, and that still leaves a number of other sources of scarcity, such as rare elements, and favorable locations.

Janessa Ravenwood

But even so, if the only work available is in:

A) Nano-product design
B) Mining rare elements
C) Real estate

then this will leave HUGE sectors of the populace unemployed. I'm out of a job as soon as the AI's get here so I already know I'm on the chopping block.

Brett Bellmore

Frankly, I'm hoping "AI" ends up meaning "Amplified Intelligenc" rather than "[i]Artifical[/i] Intelligence. I've made the transition from drafting boards to engineering workstations, I'd love a chance to try a direct neural interface. And it would certainly be the best way to keep humans on the top of the food chain.

Janessa Ravenwood

And if your average office manager were offered the following choice:

A) An AI you can speak English (or whatever) to and it does what you want and who you don't have to pay a salary to and never talks back.
B) An upgraded human programmer who does require a salary and has opinions of his/her own.

which do you think would be the choice in 9 out 10 cases?

Brett Bellmore

That rather depends on just how clueless the AI is about human aspirations and motivations, doesn't it? Might end up with idiot-savant AIs that need handlers who are technically inclined enough to understand in general terms what the AI is doing, and direct it into a humanly useful direction.

But, really, what I'm advocating is that we would actually be better off as a species, and a culture, if we directed our energies into self improvement rather that designing our own replacements. The future belongs to something superhuman, but that something COULD be us, if we play our cards right.

Janessa Ravenwood

I'm not arguing that your approach is wrong, just that the business world does not operate on altruistic principles. If they can replace someone with an AI that will do their job at no salary, they will do so in a heartbeat.

Brett Bellmore

Depends on matters like capital cost and productivity. I've designed a fair amount of automation equipment, and it's frequently worth paying wages instead of buying a machine.

The real question boils down to just how good the AI is, and how expensive. If you posit a superhuman AI with full social capabilities, aquaintance with human wants and needs, and perfectly benificient in it's motivations, my job is toast. So's the manager's.

On the other hand, we could get AI, and it turns out to be as buggy as any other massive bit of software. Try that on for size: IQ of 250, crashes twice an hour, and nuttier than a fruitcake. THAT I could easily see happening... Then you'd still have engineers, but they'd spend their time coaxing the AIs into doing something useful, instead of coming up with new origami patterns.

Mike Deering

Another problem with centralized control of powerful technologies, such as nanotechnology, molecular biotechnology, and artificial intelligence is its complete incompatibility with humanity's expansion into space. Actually there are two models of our future expansion into space, one I call the "good model" and the other known as the "bad model."

Good model - many independent ventures controlled, designed, managed, and funded by those going. There would be as much variety in the destinations, methods, goals, and structures of these expeditions as there are in the human imagination.

Bad model - government controlled, standardized around a few mission specific patterns. These would be administered by civil servants with close cooperation of the military and under the direction of a government bureaucracy on Earth. NASA goes to the stars; everyone else stays home.

Which model do you think is most compatible with CRN's plan for centralized control of advanced technologies? Which model would you like to take part in?

Chris Phoenix, CRN


To Richard, re Brownian motion and rhetoric:

One of the inventors was quoted as saying that he would not have expected the device to work. Whether this reflects an inadequacy in his intuition, or the discovery of a new physical principle, I don't know. The point is the same either way: people who assume that engineering problems are actually law-of-nature limits are likely to be wrong.

Many engineering problems have been presented as reasons why MNT can't work. Usually, the claim is so simplistic that it's insulting to MNT researchers. As Drexler remarked to me recently: "Gee, Mr. Ratner, I didn't know that robots required power! I'll go tell Marvin right away!" Referring, of course, to his PhD thesis advisor--Marvin Minsky.

Good engineering very often ignores some phenomena while exploiting others. A kitchen-sink approach doesn't make for "elegant" engineering.

Come to think of it, the whole debate about using or resisting Brownian motion is nonsense! Any protein resists Brownian motion constantly, in order to stay folded. If a protein machine "uses" Brownian motion at all, it's for a small fraction of the degrees of freedom of the system. Meanwhile, the rest of the DOF are resisting it. Does it matter, then, whether an MNT machine uses Brownian motion in two DOF, or one, or zero? No, it doesn't. If it's useful, use it. If not, find a way to ignore it. *That* is elegent engineering.

You can claim that biology is inherently better than engineering. But I'll just ask why you didn't ride a horse to work.

Chris

Chris Phoenix, CRN

About the political thread: In future, rather than start an off-topic thread, please email me or Mike Treder and we'll post a new blog entry to kick off a relevant discussion.

In fact, I think I'll do that this time. Please move this thread over to the new post I've just created:
http://crnano.typepad.com/crnblog/2004/06/is_crn_playing_.html

There's a good technical discussion shaping up here, and I don't want it to get diluted.

Chris

Chris Phoenix, CRN


OK, two technical discussions, but I think this one will be quick. Mike wrote: "Making the nanoblocks in a factory and selling us kits of nanoblocks for specific products completely eliminates all of the advantages of having a nanofactory."

The power density of nanomachinery indicates that only a very small percentage of nanoblocks can be active. So you supply a mix of 95% structure, 1% computers, 1% actuators, 2% transmission/network, 1% other. You design a few "virtual materials" that can be built by rearranging these blocks into different patterns. Then you can build any product you want with the virtual materials.

Imagine: you drop a cube into the nanofactory, and it is pulled apart into sub-cubes all the way to the production module. Then the sub-cubes are shuffled for a few milliseconds. Then they're put back together just as though the factory had just made them. In five seconds, out pops a finished product. It's extremely elegant; because it allows stockpiling of finished nanoblocks, it's worth developing whether or not the factory can make the nanoblocks.

Chris

Richard Jones

I'll start by taking your cheap rhetorical point seriously! I drove to work today rather than riding a horse because cars are more convenient and cost-effective - given the environment that's available to use them in. The last point is important; cars work for me because someone has gone to the trouble of building a well-graded, paved road between my house and the university. Between the two, there is a big and steep hill, and if you walk along the crest of it you can often see deep channels running straight up to the top of the hill from the valley. These are the erosion runnels from old packhorse trails. Between the 15th century, when lead production rocketed, and the late 18th century, when the modern road network was built, huge packhorse trains were used to carry the lead out over the hills to where they could be loaded onto river barges for export across Europe. The point of this long story is that the appropriate technology to use depends on the environment you are operating in, and if a fleet of trucks had miraculously landed in a 16th century lead mine they'd still have used the horses for transport, because horses are a much better transport technology for rough, roadless, hilly country than any wheeled vehicle.

So what's the appropriate environment that would allow MNT to exploit its advantages? I think MNT might work in ultra-high vacuum and very low temperatures - good for space applications, possible but expensive and difficult for centralised manufacturing applications, but very awkward indeed for crucial applications like medicine, where it's going to be very difficult to avoid water and 300K's worth of thermal energy. The appropriate technology for use in this environment is one that is well adapted for its special features - ubiquitous Brownian motion, high dissipation and strong surface forces.

Life's protein-based technology suits this environment well - strong surface forces together with Brownian motion is what you need to make self-assembly (protein folding, for example) a viable construction method. And it's the combination of Brownian motion and conformational flexibility which underlies the operating principles of molecular motors (I know that this is not at all an obvious point but the fact that simulations of biological motors have to use Langevin dynamics to capture their operations makes it clear). None of this says that protein-based nanotechnology (or, for that matter, nucleic acid based nanotechnology, which uses the same principles) is the only possible nanotechnology that does exploit these special features of the warm wet nanoworld. I'm quite sure that with sufficient ingenuity a wholly synthetic nanotechnology using these principles could be designed. But this work has not yet been done.

I don't normally feel the need to defend other people's pronouncements about nanotechnology, but I think your use of Austin's quotation missed out some vital context. Firstly, Austin's sorting device was not a mechanical device at all, it used the fluid mechanics of laminar flow. And as my back of the envelope calculation indicated, it wasn't at all obvious that its operation would not be compromised by Brownian motion, as it's rather a marginal case. In saying that he hadn't expected it to work, he can't be faulted on his intuition, and (for that matter) there's no indication at all that he was talking about nanomechanical devices in general. I interpreted his statement as one specificially addressed to nanofluidic sorting devices, not least because one of his earlier achievements was a MEMS based device that used the principle of a Brownian ratchet to sort DNA.

I don't take any responsibility for other people's simplistic or insulting arguments against MNT! I think my own arguments against MNT, whether they turn out to be right or wrong, actually reflect some quite subtle physics that is not well-appreciated either within or outside the mainstream scientific community. But you'll soon have the chance to judge for yourself when my book comes out. If I can brazenly highjack your space for an advert, it's called "Soft Machines: Nanotechnology and Life", published by Oxford University Press August 2004 in the UK, October 2004 in the USA. I confidently expect that you will pick over it in immense detail and find all its flaws (and that's how scientific debate should be).

jim moore

Richard,
I looking forward to reading your book.

Let me see if I understand you. You think Drexlerian MNT can be made to work but it has two major problems. 1.) It needs a good vacuum to work. 2.) Needs a low temperatures to work.

1.) Yes I believe that a vacuum is needed for MNT to work. But all the designs that I have seen use some kind of casing to keep the inside "filled" with either a vacuum or a low pressure noble gas and the diamondiod machinery. Now are you arguing that you can't construct effective walls? or By constructing effective walls you have severely limited your ability to interact with the environment?

2.) Low temp, Do you think you need very low temperatures for diamondiod nano-machines to do any work? or do you mean that low temperatures will be needed for mechanical positional synthesis?

High dissipation by this do you mean dissipation in the sense of processes in which energy is used or lost without accomplishing useful work?

The comments to this entry are closed.