• 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

« C-R-Newsletter | Main | Sounds of Silence »

February 27, 2004

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451db8a69e200d8342563e453ef

Listed below are links to weblogs that reference Early Development May Be Safer:

Comments

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

Brett Bellmore

Now, here's a point where I'm in complete agreement. Another advantage of early development is that it minimizes "design ahead"; The more products which have already been designed at the time replicators become available, the more rapidly disruptive the effect will be. Whereas if we suddenly had a working nanofactory today, there'd hardly be any disruption at all, because what would we make with it? You'd still have to pour a lot of R&D into it before you'd even have products.

Janessa Ravenwood

Unless of course it can deconstruct an object and store a record of it. In which case the answer is: anything we feed into it. It becomes a duplicator. The first things to be duplicated (in perfect detail) will likely be dollar bills...purely electronic economy, anyone?

Brett Bellmore

Highly unlikely, IMO. The first replicator is clearly going to have a limited range of structures it can create, chosen for ease of execution. Nothing floppy, nothing with surfaces that are subject to spontaneous rearangement, lots of constrants. While a disassembler has to deal with natural objects NOT chosen for ease of analysis and replication. A general disassembler is a bigger challenge than an assembler, IMO, and the first assemblers will NOT be capable of generating completely arbitrary structures, even if you had the required data.

And convergent assembly's counterpart, divergent analysis, is much trickier to pull off with objects that don't come already constructed of handy little blocks. Fully analyzing and characterizing an object of human scale will be a slow process for some time to come.

Now, granted, a lot of products made by normal production techniques could have their descriptions generated semi-automatically, but a nanofactory is NOT a cheap way to make castings and other products of bulk manufacture. Even if the nanofactory was free, the power requirements for nano-assembly are liable to be much higher than normal manufacturing, and especially with the early models.

No, the first replicator products will probably be exactly what you'd expect: High performance computer hardware, difficult to synthysize drugs, and other extremely high value to weight products. Macro-scale objects constructed entirely of nanotech parts are liable to be a while in coming, even with resort to "leggo" type designs.

Mr. Farlops

Foresight and the CRN make some good arguments for pushing for early development:

  • Removes the potentially harmful credibility gap (at least in the States.) for MNT.
  • Easier to monitor one program then many parallel programs in many countries.
  • More time to think of control and safety.
  • Shortens the time for benefits to reach the public.
I have a few large reservations though. What if announcement of such a program actually drives the formation of crash programs in other countries? Will this program be open to international cooperation?

And a really big one--what about secrecy? There will be very strong pressures from the uninformed and well-meaning both inside and outside the government to keep certain technical aspects deeply secret. They'll say, "We are protecting ourselves from potential terrorists and dangerous states. We are protecting ourselves from economic disruption. Can we really trust those French?"

I want to hear how this push for early development is going to prevent those potential arms race starters from happening? Specifically:

  • How do we keep the announcement by the government, once it's convinced, from pushing similar programs in other countries?
  • Is there going to be international collaboration on this? I kind of doubt Congress will go for that.
  • What is the role of secrecy in all this?

Brett Bellmore

Clearly, a push for early development won't prevent an arms race from occuring. Then again, OUR refraining from such a push wouldn't, either. It would simply assure that we wouldn't win it.

Unilateral disarmiment doesn't prevent other countries from arming, it just boosts the payoff to them from arming.

Now, at present, and not entirely by accident, the nations which are most technologically advanced also have fairly sane and humane governments. If those nations really push to get nanotech first, it's overwhelmingly likely that they will get it first.

If, on the other hand, those nations do NOT make a big push to get it first, it becomes possible that a technologically second rate nation, with a far nastier form of government, could beat us past the goal post by making an all out effort. And that would be bad, very bad.

Janessa Ravenwood

How do we keep the announcement by the government, once it's convinced, from pushing similar programs in other countries?
----
You can’t. At that announcement, they’ll start working on their own and there’s nothing we can do about that.


Is there going to be international collaboration on this? I kind of doubt Congress will go for that.
----
On a matter of national security? HIGHLY unlikely. Ditto for the other countries.


What is the role of secrecy in all this?
----
All the governments will be trying to keep the details of their technologies secret. Case in point: do they post the blueprints for our newest nuclear submarines on the web for all to see? Same thing here. Again, I post my prediction that they’ll make private nanofactory ownership illegal once they figure out that it can actually be done. Then the nano-underground is born and it’s another War on Drugs, just this time the War on Unregistered Nanofacs. Viva La Resistance...

Janessa Ravenwood

Brett (re: your first post to this thread).

Well, then I guess from an individual household standpoint, the first replicators won't be of a huge amount of use. However, it also means that we probably WILL have a good window of time to adjust socially to the presence of nanotech as I'd say you need full on-demand disassembly/assembly for things to get REALLY interesting. At the level you're talking about, you might be able to build a few nasty nano-toys, but mass devastation is probably still a bit out of reach at that stage. Which isn't really a BAD thing. I mean, if all you can do is play around with your pre-fab-nano-lego-blocks, I wouldn't start planning for world domination JUST yet.

Chris Phoenix, CRN


Re: limitations of early nanofactories:

My nanofactory paper seems to show that even early diamondoid nanofactories could be useful, not just for computers, but for building structural diamond. The energy costs, using fairly conservative estimates for inefficiencies, came out to $20/kg of product. (Section 8.6) of http://www.jetpress.org/volume13/Nanofactory.htm

Given the very high strength of diamond and the improved material efficiency that's possible with micro-structured trusses, 1 kg of diamond would replace 10 kg or even 100 kg of today's structural materials. And could replace 1000 kg of motors or computers.

So take my car--a ton, more or less. Divide by 10: that's 100 kg, or $2000. But much of the mass is the engine. And a lot more is compressive strength. So maybe you can divide the mass by 100, for a $200 car. Obviously, a 10-kg car would blow away, but you could ballast it with water or sand or pig iron.

Remember, this is with a *very early* nanofactory.

And the same reasoning holds for aircraft, including military aircraft...

Chris

Chris Phoenix, CRN


Re: Whether early development would prevent an arms race...

Right now, it would be hard for any nation other than the U.S. to build a nanofactory. There are several countries that could do it, and if they've already started, they could finish any time. But the U.S. would be likely to finish well ahead of the others if we all started today with well-run programs.

Ten years from now--maybe just five years from now--any reasonably large company will be able to do it in any of a dozen nations.

So if it's best for the U.S. to develop far ahead of the others (which seems likely, but I'm not sure), then we'd better start developing ASAP. If we wait, it becomes far more likely that someone else will be able to start and finish before we can get our feet under us.

Chris

Brett Bellmore

Oh, I agree that you can compensate for the high energy cost per unit mass, by using the very high structural strength and power densities to build products that are very light weight.

With a lot of design work.

I was really pointing out that the high energy cost made the use of a nanofactory as a kind of "matter duplicator" for convential products infeasible. There really aren't any shortcuts; A nanofactory by itself doesn't upset all the applecarts, it's only when combined with the results of a LOT of intellectual effort to design products, that things start changing. And that's what is going to cushion the shock.

Mr. Farlops

So basically the point about there being only one program to monitor is meaningless because, once the government announces the commitment, a lot of other governments will announce similar commitents.

This may be a race between largely friendly powers but it's still a race. There will be multiple programs of research and some things will slip through the cracks as military secrecy is imposed--so much for open development! And the United States, being the largest ecomony on the Earth will probably win that race. Let's hope we are wise with our new power. Let's hope the rest of the world trusts us to behave responsibly.

I guess it's unavoidable but, I'm not too happy about it.

As far as letting the public have general assemblers, sorry, but I only have to think about the level of traffic collisions (These are negligence, not accidents.) in the United States to nix that idea. I guess I just don't trust people.

I can live with n-number of avoidable gun deaths, cigarette deaths and car wrecks a year and so I think people should be allowed to own guns, smoke cigs and drive cars but, I certainly don't advocate letting everyone own a nuclear weapon! The same applies to general assemblers; the magnitude of damage is too great. The number of people allowed to use such technology should be kept small, highly trained and closely watched.

But obviously this situation won't last. Eventually, probably too quickly, the secrets will leak. I can only hope we have blue goo in place and that we have settled permanently in outer space by then to protect civilization from our own foolish and evil whims!

Karl Gallagher

Chris & Brett--

The energy & material costs of a modern car or plane are such a small fraction of the sale price that most likely no one has every calculated it. Certainly less than 0.1%. What you're paying for is people designing the parts, shaping the parts, and moving the parts from point A to point B. A nanofactory will shift a lot of the costs from the 2nd category to the 1st but they won't go away until you're into mass production. That's going to take a couple of generations of the technology.

Brett Bellmore

Actually, I work in the auto industry, designing parts, (Stampings and extrusions, mainly.) and tooling to make them. I assure you that, except for unusually low volume jobs, design costs are a small portion of the piece price; How much do you suppose I get paid, anyway? Not THAT much, unfortunately! :(

For stampings, the chief cost is materials, followed by tooling and machine amortization. For extrusions, about the same, except that tuning in the die can be a major part of the cost.(All the while you're trying to bludgeon it into working, you're puking out expensive materials onto the floor, and tying up machinery.) Energy costs are quite low, except to the extent that they're included in the material costs. (Aluminum is occasionally refered to as "solid electricity", the energy cost to make it is so high!)

Of course, I suppose other costs accumulate by the time the parts we make are assembled into finished cars.

My take on this is that nano-assembly will reduce material costs to a trivial part of the expense, except in special cases, (Sure, building your diamondiod out of pure carbon 12 makes it better performing, but you'll pay through the nose!) On the other hand, energy costs will increase drastically, except perhaps in the case of single purpose factories which don't have to trade poor efficiency for generality. And design costs will increase, too.

See, there's a lot of people out there who can design products using conventional manufacturing techniques, but for quite some time after we have working nanotech, I expect nanotech designers to be in very short supply, and corrispondingly well paid.

michael vassar

I think that a major point of convergent nanoassembly is that design becomes rather easy compared to current design. Also, many products are already designed. You don't need to disassemble them to transfer these designs to a nanofactory. Mostly you just need to know what the mechanical and electrical properties of each piece or part of a piece are and you can replicate them by CAD. Obviously this gives worse products than redesign would, but it saves a lot of work. Also, a small set of pre-designed products with MNT performance can revolutionize the world economy. None of the following seem very complex to design: motors, mechanical power storage, solar power generation, active nanotube protective clothing, sub-dermal heaters (for warmth and to burn calories), nanoelectronics, nano-optics, phased-array optics, general-purpose space-filling and force-exerting nanorobots, space elevators, water-filters, recycling ultracentrifuges and filters, air filters, LED paint, high temperature incinerators, high pressure low friction and high heat conductivity diamondoid for chemical engineering and other applications, lighter materials to replace parts of aircraft, improved biotech instrumentation, sharper smaller mechanically controlled surgical tools, cool chips, high-friction gecko biomemetic surfaces, low-friction easily cleaned surfaces, small sensors.

Chris Phoenix, CRN


On ease of design, I'm inclined to agree with Michael and disagree with Brett. Sure, it'll be a new kind of design. But software engineers are good at that. As long as the design can be at a functional level rather than a physical level, it shouldn't be very hard to learn the new domain. And most designs can be functional, because there'll usually be room for a huge amount of overdesign.

Will nanofactories have huge energy costs? Not really. According to Alcoa, aluminum requires about 12 kWh/kg with modern efficient processes. My primitive nanofactory design takes less than 250 kWh/kg. But the product should be more than 20 times as strong as aluminum.

Chris

Karl Gallagher

Designing one part is easy. Designing 10,000 to work together is very hard (this is my day job). I suspect molecular manufacturing design problems will be dominated by the emergent effects of putting many identical active units together. Pretty soon you'll get "flocking" behavior and have to introduce a lot of complexities to get the system doing what you want it to do rather than going off on its own.

Chris Phoenix, CRN

Karl, what kind of part do you design? I bet it's integrated circuits. How much easier would your job be--how much simpler would the anti-flocking complexities be--if you were allowed to build devices with the same function, but one-tenth the performance? If you could start up your 10,000 devices in ten gangs of 1,000, for example?

I have no doubt that the ultimate level of performance will be beset by all kinds of nasty analog interactions. But the power density of Drexler's diamondoid devices is ridiculous: 10^17 W/m^3 for the electrostatic motors, for example, and 10^11 W/m^3 for rod logic. For most products, cooling will require that we use a tiny fraction of that. So we can spend a lot of space on buffers to simplify interactions between parts.

MNT will be more than an incremental improvement. Even if we throw away several orders of magnitude of miniaturization, it'll still be vastly more compact than today's technologies. So we won't have to cram things close together--at least in early products--but early products will still be revolutionary. Most of what CRN has written is based on early and primitive designs.

Chris

Karl Gallagher

Chris, I'm a systems engineer, currently working on fighter planes (JSF). Before that I did launch vehicles and weather satellites. So complex systems with thousands of parts are my specialty.

I don't worry much about being able to fit together a grid of identical components. But if you want this device to have a useful impact on the human scale you have to assemble the pieces into something big enough for us to touch. At which point you have a size ratio about the same as the Empire State Building to the screw holding a switchplate. The MM designer has to accomodate all the macro-scale interfaces of the final device, none of which will match the crystal perfection of the nanodevice array. It's that transition that causes the problems. So you can build a lovely nanoengine but that only gets you 10% of the way to a device that will actually be useful to people.

Software engineering is a good analogy to this. A simple number-crunching algorithm is easy to implement. Embedding it within a good user interface and making sure it can run on a changing platform is the hard part. Whole libraries can be filled with descriptions of the unexpected complexities that bite people trying to do the latter (I have a shelf of them myself). I expect MM designers to worry more about the analogue supporting different printer interfaces and network protocols than they will about how to optimize their mechanical data storage.

The comments to this entry are closed.