Challenges and Pitfalls

These early years of the 21^st century already are a time of rapid advances in science and technology. Every day brings news of startling developments in fields such as genetic engineering, neuroscience, and nanotechnology. So what will the near future actually bring us? Human beings that glow in the dark, like our bioengineered pets? Robot servants? Flying cars? Genuine artificial intelligence? Or something even more exotic?

There is good reason to believe that within the next 10 to 20 years, the most significant changes to society will go far beyond glowing people or flying cars. Many of them may result from the introduction of personal nanofactories, a powerful application of exponential general-purpose molecular manufacturing, made possible by advanced nanotechnology.

In this paper, we will explain exponential general-purpose molecular manufacturing: the basic concepts behind it, and why it will be a technological breakthrough of transformative power. We will show why preparing for it is vitally important—and will be very difficult. Along the way, we will explore how several types of social systems may respond to the changes that molecular manufacturing will bring, including unprecedented material abundance and other opportunities. We will take a brief look at the possible timeline (sooner than many people will expect), explore problems in familiar areas such as military conflict, and touch on new classes of problems that humanity will have to face. By the end, it should be clear that the challenges and opportunities created by molecular manufacturing cannot be addressed by any simple solution.

Above are the opening paragraphs of a new paper, "Challenges and Pitfalls of Exponential Manufacturing," by Mike Treder and Chris Phoenix, that we've just posted on our main website. It's a reprint of the chapter we provided for the recently published anthology, Nanoethics: The Ethical and Social Implications of Nanotechnology, edited by Fritz Allhof, Patrick Lin, James Moor, and John Weckert. We encourage you to get the book, or at the very least, read our contribution.

Tags: nanotechnology nanotech nano science technology ethics weblog blog

Comments

[jim moore]

Good article, you have convinced me that nano-factories should never be widely distributed. They are just too dangerous.

So the question becomes how can you short-circuit the process that leads to widespread nano-factories.
I think the key is to release only the products of nano-factories to the public, never the factories themselves.

Let me describe a system that is similar yet, much less dangerous than the traditional nano-factory.

SPOT fab
A Smart Powered Octet Truss fabrication system.
This system uses three differently shaped reusable micro parts to make a very wide range of technological artifacts. Each one of the parts is made by molecular manufacturing in a highly controlled setting then shipped around the world.

There are Hubs, Struts, and Plates. The Hubs and Struts are formed into a light weight octet truss system that fills most of the space that an artifact takes up. The Plates provide a "skin" for the artifact (and an interface between some large parts).

The Hubs would be spherical in shape ,~25 microns in diameter and typically connect to 12 Struts. Their purpose is control the flow of information and energy through the system.

The Struts are rods 10 microns in diameter and 60-100 microns in length. The struts can expand to 100 microns and contract to 60 microns. During that process the rods either expends or absorbs energy. The purpose of the rods is energy storage and transmission. (they also process and transmit information)

The Plates are hexagonal plates 10 microns thick and ~150 microns wide. The plates provide surface features and and membranes for artifacts. They can also provide and interface between moving parts in an artifact. Plates would also be a good place for sensors, lasers, solar cells, and other miscellaneous functions. The plates connect to either Hubs or Struts.

The octet truss system fills the form and provides the information & power infrastructure for the artifact that you are creating. And because the Struts can change length the truss can also provide movement.

The SPOT fabber uses a Cradle to Cradle design philosophy to dramatically reduce the environmental impact. Any "new" artifact is a rearrangement of reusable micro parts thus saving large amounts of energy and eliminating material waste.

I will leave for another day why the Guardian, Commercial and Information Systems might prefer a SPOT fabber to a nano-factory.

(PS, I know the SPOT fabber is like Utility Fog crossed with a nanofactory)

[John B]

Jim -

Your design's "just" a utility fog, really.* Some of the problems with ufog include:

- the need for constant power to retain shape against load - something that would probably be detrimental in a building, or even a car.

- it'll be porous - won't be able to contain pressure differentials or fluids. This greatly complicates many jobs - like drinking glasses! *grin*

- This is a high-surface-area-to-volume solution. As such, it is likely to be very reactive, probably significantly more reactive than a similar mass or block of solid material. While most mechanochemically interesting materials are pretty chemically stable, not all are.

- There remain significant command bandwidth as well as potentially cooling and power distribution issues to be addressed

* Note - UFog is potentially of enormous use. Don't take this as a slam against ufog capabilities, please! Just note that there's a lot of problems to be worked out with the practical application of nanofactured materials, including UFog.

[jim moore]

John B.
Differences between Utility Fog and SPOT fab:
Utility Fog - one identical part repeated many times.
SPOT fab - 3 differently shaped parts (not all the parts that are the same shape have identical internal structure, for example you can have different types of plates for different functions, examples; solar cells, color change surfaces, rollers embedded on the surface etc.)

Utility foglets can detach move to a new position and reattach themselves. This gives Utility Fog the ability from morph one object into another. For a SPOT fab the Struts, Hubs and Plates attached during fabrication and don't become detached until disassembly.

Now onto your specific "problems"
1- Need for power to maintain shape against a load. Not true for a SPOT object if you just lock the struts into position during fabrication.

2- Porous- One of the big reasons for having Plates is that they can form a continuous "skin" around an object or a part in an object.

3- High surface to volume ratio- this is true up to a point. The smooth sides of the parts will reduce reactivity. An outer surface layer of silicon carbide or aluminum oxide on the parts would also reduce the reactivity. (The parts MAY be large enough that there is not an explosion hazard from finely divided carbon, typically you need submicron carbon particles to be an explosion hazard)

4- Limits on communication and power distribution through the system. This is a feature not a bug. It limits the destructiveness of potentially dangerous artifacts.

5- I have also heard that programming for Utility Fog is very difficult, SPOT fabbed artifacts will be less difficult because they can't morph into other things.

[Hal]

Going back to the original article, a common problem in discussing the impact of nanotech is failing to distinguish between early and mature versions of the technology. Contrary to the statements in the article, early nanofactories are going to produce waste products. And the cost of energy is glossed over; it is likely that early nanofactories will require huge amounts of energy and will not be able to effectively recycle or reuse it. Energy costs may in fact dominate product costs, so there is no guarantee that even if the feedstock is inexpensive, products will be.

You can talk about improved prototyping reducing design cycles, but as someone who works in software (with a near instantaneous design to test time), I can tell you that producing large but reliable systems is still slow, complex and error prone, and many projects are ultimately abandoned as failures.

I would not plan on quick progress to some mythical perfect-efficiency manufacturing system with no atom misused, every aJ of energy precisely channeled, the whole thing working in a near mystical harmony with nature. Real nanofactories are going to experience real compromises in their designs and their imperfections will undoubtedly play a major part in determining their ultimate place in the world.

[John B]

Jim -

Good differentiation.

Some gotchas, tho' - sounds like you've got alot more than 3 parts - each plate type is going to be a different design, and you've got at least 2 generic types of struts (static & dynamic) with hints of quite a few others - conductive (electrical, heat, data, and combinations), etc.

Additionally, you seem to think that hexagonal 'plates' are going to be designable so as to be pressure-bearing in a static configuration. While this might be do-able in certain configurations, others are going to be rather more difficult. Example - a knife blade edge, or other 'sharp' corner. (Rounded edges I suspect will be easier to make more resistant to pressures)

Question - You state in your initial post that struts are flexible in length. This implies to me that you'll have objects that flex as one possible subset of the technology. If so, how do you maintain good seals around the edges of plates? Or are your resultant designs exclusively static - no flex in materials? (This could go either way, IMO, depending on your intentions/design choices)

Hal -

Good point, and one I was trying to make myself over at the http://crnano.typepad.com/crnblog/2007/08/nanotech-revolu.html post.

I would also suggest, given the purities that'll be required for initial nanofactory (IMO), I rather doubt initial feedstocks are going to be cheap. If self-cleaning sorting wheel designs get worked out in time, this purity requirement might be removable, but I don't think this is going to be an easy fix, and probably therefore going to be a while before that particular efficiency drain gets plugged.

Energy, as you state, is going to be another massive cost factor. Then again, that's also the case with most manufacture today. The difference is that the energy expenditure nowadays is in the factories & the delivery mechanisms, rather than at the endpoints - meaning that another infrastructure upgrade will be needed, or various parts of the old infrastructure will continue to be needed. (IE - either make it at home and require industrial power, or make it at a factory and ship it to stores/homes, or perhaps find some other solution...)

And may I applaud your last paragraph? That should be bluntly put at any nano-future discussion where the focus isn't explictly stated as being beyond the development of nanofactories...

-John B

[Tom Craver]

Hal - while there's truth in what you say about software design being hard, you also need to realize that software complexity is much higher than most hardware complexity. Microprocessor computation cores are about the closest hardware things to software in complexity.

Try this thought experiment - visualize a modest sized software program as a factory. For loops are workstations that take in pallets loaded with parts. They feed those parts into welders (add), cutters (subtract) and dozens of other basic machine tools. That'll give you an appreciation for how relatively complex software already is.

[Josh]

WRT porosity, given a fine enough grid, either UFog or SPOT can be effectively waterproof at ordinary pressures (e.g. the bottom of a water glass) given a sufficiently hydrophobic surface, without the need for plates. (Think of a woven fabric tent treated with Scotchgard.) If the stuff has the ability to change the hydrophilic/phobic nature of its surface under program control, it has the ability to manuver films and volumes of water within its volume with remarkable flexibility. A stack or set of concentric shells of alternating water and air films could in theory resist significant pressure, if anchored in place by appropriate surface characteristics.

[jim moore]

John B.
I had come up with 3 different ways to approach the problem of water tight, air tight zones for an artifact that gets fabricated.

Josh, beat me to the hydrophilic / hydrophobic surfaces approach. (very clever to use water films to create air tight zones, I had not taken the idea that far).

The second approach is use plates and keep those areas you need air tight / water tight static. Just allow movement in rest of the artifact.

The third approach would be to make expandable plates, which would expand and contract with the rest of the structure.

Just for design simplicity I would like to keep the struts and hubs as uniform as possible and put most of the design diversity into the plates. (So the struts should be able to expand, contract and lock in place)I think that the struts and hubs would make up ~90% of the mass of anything you fabricate so keeping them as uniform as possible would simplify the fabrication process.

[John B]

JoSH -

Given a tight enough spacing, I can see how a hydrophobic material could do this. I'd be a little concerned about the water barriers being used to contain pressure - sounds like an invitation to cascading failure off the top of my head.

Question - How would you propose changing the hydrophobicity/hydrophilicity in a nanoscale material?

Jim Moore - If you move the 'guts' of an object, either you don't disturb the surface(s) (in which case, why're you moving things around?), or you'll need to smoothly adapt the surface. This latter, with plates, is going to be ... painfully detail-ridden. IMO, it would require a very advanced iteration of such a capability to move plates such as you propose smoothly enough not to have gaps big enough to slip through.

Additionally, if you move things around, the power & control concerns mentioned above apply - and obviously you can't lock struts if you're moving them.

-John

[jim moore]

John,
Maybe it will help to think of a specific artifact rather than something in the abstract.
Say you want to SPOT out a propeller driven plane.
We want to have an airtight cabin. But, we can't use Josh's water film trick because the plane would dry out as we fly.
(which is a shame because should be easy to change the nature of the surface of a SPOT part, all over the hydrogen terminated diamond surface you have diamondiod pillars (10-20? nm long) that can be push out from the surface or pulled back below the surface. The sides of the pillars are covered with OH groups or other hydrophilic group. Pillars out and water loves it and wants to stay put, Pillars down water doesn't like it nearly as much and is much more willing to move.)

OK, back to the plane, the cabin is a simple static structure, ( needs some fancy electronics but the structure is no problem right?) For the propeller and motor we have two ridged objects moving relative to each other. Most of the volume is made of hubs and struts the surface is covered in the plates. Some of the plates are simple and just block the air but the plates at the interface between the propeller and the shaft are different. The plates on the inside of the propeller make a smooth track, the plates on the outside of the shaft have powered rollers.
For the wings we want to be able to change the wingspan of the plane. The wing keeps the same profile but can change its wingspan from 6 to 10 meters, You do that by simply expanding the octet truss along one axis and by having plates on the surface that can expand by the same amount and in the same direction.
And finally for the energy source we stretch out the diamond springs in a whole bunch (several kilos) of struts and place them between plates to form a spring powered battery.

[John B]

First off, let me commment on one line of yours. You state, "We want to have an airtight cabin. But, we can't use Josh's water film trick because the plane would dry out as we fly."

If the surface is as truely pressure-resistant to hydrogen as you stated earlier, why would water - many times larger - be able to escape from its layer(s)?

Next, let's take this a step down in complexity. Let's just look at the wing of the plane, with flaps available by 'morphing' the material in one preplanned manner - stretching the trailing edge back and down.

A further assumption that seems reasonable to me - we want to use the volume of the wing to store material - gas for the engine perhaps, as is a common practice in aircraft nowadays. This means we need to maintain a pressure barrier, else we'll be trailing a plume of jet fuel as we fly. This is bad for a number of obvious reasons.

So. We have a surface of hexagonal plates. We need to be able to keep them sealed tightly enough to resist a pressure differential between the fuel and the surrounding air (there *is* a low grade vaccuum generated by airfoil wings after all, plus flying height is typically enough to have reduced atmospheric pressure as well). AND we need to move them around to handle the shape change.

As such, how do we move the nodes and rods under the hexagonal plates to maintain a pressure seal between the plates, and still alter the shape of the object? This is a non-trivial amount of complexity, IMO.

Can you whip up an example architecture showing the kinds of design you're expecting to work in such a situation? *curious look*

-John B

[jim moore]

John,
I guess that I was not clear enough, perhaps I should have wrote "We can not use water films alone to make the cabin airtight because the water would evaporate away as we fly, we will have to join together plates to maintain cabin pressure."

It might also be helpful to make a physical model of a section of an Octet truss. Then think about how it can change shape as some the struts change length.

Now if your point is to say "you can't easily make arbitrary morphological changes and maintain substantial pressure differentials within a SPOT object." I agree completely. But I would add, that you can make some limited kinds of morphological changes and maintain a pressure difference.