• 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

« CRN at Five Years Old | Main | Debating CRN's Scope »

February 05, 2008


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

stephen c

"Cameras and microphones in your glasses or shirt buttons will record every moment, upload it, and let you replay the good bits. . ."

yeah, did y'all see this...

Richard Jones

Jamais, I think your commitment to the idea that "the future remains unwritten" is commendable and absolutely right, but I'm struggling to reconcile this with the fact that you've just issued a set of eight scenarios every single one of which takes as read the idea that molecular manufacturing is technically possible, practically feasible and economically viable. It isn't just that you've assumed that some kind of advanced nanotechnology will come along and you've explored the many different ways it might make an impact - each scenario takes as read a single imagined realisation of advanced nanotechnology, the CRN-model nanofactory.

This might be understandable, though probably not entirely desirable, if there was a widespread consensus that this was the form in which advanced nanotechnology was most likely to manifest itself. As you should know, though, there is no such consensus, so this will strike many readers not just as a series of visions of the future which is excessively foreclosed, but as one that looks much less likely than a whole host of other possible futures.

The last time I commented on the CRN blog was September 27th last year, where I wrote in connection with the way various academic centres were thinking about possible impacts of nanotechnology on society:
"What they do start out with a presumption to reject, of course, is a position of technological determinism. This is both from an appreciation of historical arguments against such a point of view, and from the normative position that they generally start with, that it is desirable for technology to coevolve with society in a way that responds to people's aspirations, expressed through democratic mechanisms of one kind or another. So even if they weren't exposed to the scientific consensus against the likelihood of the MNT project coming to fruition in the way CRN imagines, they would be predisposed to imagine a much broader range of possible futures than simply considering the arrival of MNT as a historical inevitability."

To which Mike replied: "We appreciate (sincerely!) the admonition to consider "a much broader range of possible futures." In the scenario project we're working on, we've tried to imagine and pursue outcomes that vary from what CRN originally posited as the most likely path for development of atomically-precise exponential manufacturing systems."

But now I can look at the scenarios, I can see you didn't take the admonition that seriously at all. While you might consider different ways in which CRN-model nanofactories might emerge, you still can't bring yourself even to consider the possibility that they don't emerge at all and the technology unfolds in some completely different way. I'm reminded of two things Jamais recently approvingly quoted on his own blog:

"The biggest mistake a forecaster can make is to be more certain than the facts suggest."

"The future constantly arrives late and in unexpected ways."

In insisting on a future which, far from being open, is absolutely foreclosed and pre-determined with respect to MNT, you've failed to heed these warnings.


I'd suggest that it's fair for CRN to focus on scenarios in which nanotechnology plays a role, since that is the purpose from which they take their name. My concern is that by focusing too narrowly on the nanofactory model, they are missing out on the opportunity to make a contribution to the discussion about other possible nanotechnologies. Concepts such as smart dust or invisibility shells may have major implications for surveillance and privacy, but they are being developed without waiting for nanofactories to appear on the scene. Control of matter at the molecular scale will be a powerful technology even without embedding Drexler's levers and pulleys. CRN would do well to widen its scope and pay attention to the full range of work being done in this area.


CRN, I'm confused, and I think you are too.

Mike says that "CRN never has suggested or even implied that the timelines we offer were intended as predictions". Well what were they then?

Your timeline page has the title "Estimating a timeline for molecular manufacturing". You used to say of molecular manufacturing "almost certainly will by 2020" and now its "almost certainly will by 2020 to 2025". How is this not an explicit prediction?

Am I the only one baffled by this?!

Can I also point out that if 5 years of progress in the life of CRN has been accompanied by a 5 year delay in your predictions (projections, soothsaying, foretelling) then we don't seem to be making much progress. These arbitrary timetables remind me of the many predictions/projections made by the Jehovah's Witnesses last century, and a very good episode of the Simpsons.

Isn't the truth of the matter simply that you don't have an objective basis for your timetable? Its not a projection of anything. Its just made up.

Jamais Cascio


The scenarios CRN constructed last year were not intended (nor were they depicted) to be the full range of possible futures; they were intended to be a set of different scenarios depicting how molecular manufacturing could emerge (or in the case of scenario #6, "A Goal Postponed," what might specifically delay its emergence). You're right that there were no scenarios looking at futures where molecular manufacturing turned out to be a much harder and more distant possibility than the early 2020s, or that it would take a heretofore unexpected alternative form. That wasn't within the scope of what the scenario project sought to explore: the range of drivers, assuming that molecular manufacturing of the form that CRN focuses upon was technically feasible within the next 15 years, that would shape its development.

Let me explain by analogy here: what you're arguing is akin to confronting a group tasked to look at different scenarios in which a particular business plan unfolds, and complaining that the set of scenarios is bad because it doesn't include one or more futures in which the business plan differs in its fundamentals, or isn't implemented at all.

In both cases (nano and the example above), the intent of the scenarios isn't to answer "is it possible?", but to explore "how might it happen?", with the underlying assumption that yes, it is possible. If you flatly disagree with that underlying assumption, then the resulting scenarios will be at best irrelevant, at worst invalid. Conversely, if you conditionally accept the underlying assumption ("for the sake of argument"), then the scenarios may offer useful insights.

I'm not trying to, nor do I expect to, convince you that nanofactory-style molecular manufacturing is in the offing over the next 15 years. What I'm trying to explain is that, if it is potentially on the horizon -- and CRN and the people it works with believe that this potential is there -- then its impact is likely to be significant, and the ways in which it could manifest are worth exploring. That's what the scenarios set out to do.


Vince, I won't speak for Mike, but I can say that the shift to a fuzzier target date came about in part because of my pushing for it (I'm loathe to assign a specific date for an emerging technology, except in the context of exploratory scenarios), and the five year/five year correspondence was entirely coincidental. Mike & I actually discussed whether there would be an issue with the 5/5, and we determined that, yeah, some people might get stuck on it.

What about this phrasing?

"Based on its research and communications with those working directly with the technology, CRN believes that there is a slim chance that a basic nanofactory could appear as soon as the early-mid 2010s, with gradually increasing likelihood through the remainder of that decade. The probability starts to increase more dramatically by the early 2020s, and we believe that there is every reason to expect to see a fully-functional molecular manufacturing nanofactory by the middle of the 2020s."

That offers the same timeline as the current phrasing, but removes the offending five-year spans.

Brian Wang

There are over 100 million camcorders and videophones now. Each year over 1 billion cellphones are now sold. (Cellphones get replaced on avg every 18 months.) Over 1 billion of the phones are cameraphones. MIT and Texas Instruments have designed chips that are ten times more energy efficient and could run off of ambient energy (thus could be always on.) 5 years away from commercialization for the low power TI chips. A few years back in the lab there has been work that makes digital CMOS cameras 50 times more energy efficient.

Superior Lidar, t-rays and better satellite and other remote sensing too.

Progress with computers and software automatically deciphering what is in the digital images. Quantum computers will also help with pattern recognition and faster image database searches.

Ten to 100 dollars for USB stick gene testing devices (recently made in Canada) not use commercialized but would seem deployment could be rapid. Readout results in minutes.

Printable electronic capabilities increasing rapidly and costs dropping. Will make fabbers/3d printers faster, better and cheaper.

5-6 years away from billions of always on video cameras that are almost always carried by people, plus set up at homes and offices etc.... Already at the 100 million+ level. The 5-6 year vidphones will also have darn good resolution and terabytes of storage (advanced flash or its successors PMC etc...)

I think the even more ubitquitous vidphone, webcams, and remote sensing and software deciphering/pattern recognition should be discussed. The societal impacts will here.

Pre-diamondoid MM tech should be discussed. Almost every big societal impact envisioned for Diamondoid MM world is happening before that gets going. I do think the bootstrapping into diamondoid MM world will happen faster than people think. Advanced DNA nanotech, quantum computers, cheap supercomputers, metamaterials for superlenses to help Freitas and Merkle finish their work.


Jamais, your explanation does go a long way to explaining this significant change in CRN's outlook and I appreciate that. It certainly makes more sense then Mike's talk of non-predictive projections. But the underlying issue still remains. If the original timeline had no objective basis and was wrong, and the new timeline, however you choose to phrase it, was derived in the same way, then why should it be any more accurate? Have you considered this fundamental problem?

The people who are at risk of getting stuck on your projections are those who rely upon them. That does not currently include me.

Brian Wang

There is some instructions and software online for a DIY motion activated digital camera for less than $100 for the pieces

All of the pieces would be fabbable with current or slightly upgraded desktop fabbers (reprap, fab@home etc...) and can be made with rapid manufacturing and 3d printer systems.

Here is how people now can print electronics with inkjet printers today

So less than $100
When it is fabbable and made say usb stick size - $5-10. (2009-2012)
When they are rice grain size (2010-2015) and cost less than $1.

Fabbers and 3D printers expected to improve capabilities and drop in price below $1000 by 2011. Have 100,000+ machines by then. By 2014, there seem likely to be many millions of even more advanced fabbers and 3d printers.

As time passes more memory for storage and more energy efficient and longer lasting and higher resolution cameras etc...

so a few final points:
1. sousveillance and powerful fabrication from clearly seeing what is available today and where it is obviously headed
2. Full blown nanofactory level MM will delivering something a lot more powerful. Those who have difficulty accepting that what they ridicule will be delivered by mundane means will not be able to accept the logical conclusions and likely consequences of full blown MM.

Robots in the bloodstream - Israel, S Korea and Japan

In vitro millimeter surgical robots

Swarms of robot weapons: All the mini-UAV work. The previously mentioned cigarette size plasma jet UAV.

So Vince, Richard: Do you not believe in the billion dollar rapid manufacturing and rapid prototyping industry ? Do you not believe in 3D printers ? Do you not believe in inkjet printers ? Do you not believe in printable electronics ? do you not believe in UAVs (tens of thousands in use) ? Do you not believe in the video camera cellphone ?

Do you not believe in DNA nanotechnology ? Do you not believe that the 500,000bp sequence will be placed into an emptied cell ? What do you think will be used with those DNA manufacturing capabilities ? Sure, sure we can position millions of nanoparticles to within a nanometer in a 3d space but something will prevent this from scaling to many trillions of times. They won't be able to figure out how to make the chemical swaps so that the structure will retain its shape after the water is removed (even though the researchers say they will do that). No way does this lead to anything more in terms of molecular manufacturing.

Sure 500,000 custom base pairs, but we could not make a custom ribosome with 2.3 million base pairs that is 460% more (500,000 base pairs I believe because they did it, but 2.3 million? that is crazy talk. Don't believe that projection). The 2000% jump in synthesized length would not get repeated.

Sure they have made two replicatable synthetic bases to go along with four in nature but we will not make more synthetic bases and those bases will not prove useful or lead to more interesting DNA nanotech capabilities. Nature is tops, those synthetic bases are manmade and therefore useless.

Brian Wang

Besides the USB gene testing, Spectrometers on a USB stick seem close as well this will allow for monitoring of chemicals and pathogens and other things in the atmosphere by individuals.

Brian Wang

Jamais Cascio, founder of Open the Future and a director at CRN, says nanofactories will have a huge impact: "If it becomes cheaper and more efficient to have something printed out locally instead of made in China, it will have a big effect on things like trade balances, international labor, and ... our national economy."

Future Fab@home local production is not assured victory over centralized production

Transportation costs to ship 140 million 100 gram UAVs do not put centralized production at much of a disadvantage relative to local fabrication. So do transportation costs and other localized advantages outweigh economies of scale ? Big photocopying jobs still go to Kinkos and bigger printing runs of marketing materials still get outsourced to China. I see the relationship with local printable electronics and fabbers to be the same.

It is not just that you can have bigger machines with higher economies of scale, the economies of scale also have to do with having the customers to keep the machines busy and running all the time. There are economies of constant operation and amortization and economies of specialization. The manufacturing specialist will be better at it because that is what they do all the time.

Tom Craver

Economies of scale - sure, that's been good to us for generations now. But at some point, you just don't care any more how efficiently you're using certain machines.

Otherwise, we wouldn't all have PCs - we'd still be using timeshared central computers. But there the PCs sit, with us using less than 1% of their cycles, even counting the processes that the OS and virus scanners and such run.

Mike Treder, CRN

Vince, I'll say it again: there is a big difference between predictions and projections. CRN has always insisted that because we don't and can't know for certain how soon molecular manufacturing will be developed, it behooves us to prepare for the earliest plausible time when it might happen. It's also essential that serious studies to better understand the technology and its development pathways be funded and undertaken. If it comes later than we think it might, that's fine -- but if it arrives too soon, the world may not be adequately prepared, and the consequences could be severe.


Mike, it feels like we're going round in circles here. On your web site you write that exponential general-purpose molecular manufacturing will almost certainly become a reality by 2025. This is a prediction and I will tell you why.

Here's how the oxford english dictionary defines predict "To state or estimate, esp. on the basis of knowledge or reasoning, that (an action, event, etc.) will happen in the future or will be a consequence of something; to forecast, foretell, prophesy."

Here's the definition from merriam webster "to declare or indicate in advance; especially : foretell on the basis of observation, experience, or scientific reason"

Your statement clearly indicates and estimates that something is going to happen in the future. I don't know if you have a "basis of knowledge or reasoning" or "observation, experience, or scientific reason". You haven't actually provided one. But I would find it funny if you argued otherwise.

Would you have me believe that writing "almost certainly" changes the meaning of the sentence to indicate that, on balance, you don't think molecular manufacturing is going to happen by then? Sure, it stops your statement from being an outright assertion, but your statement is still more than indicative and at least an estimation.

Mike, I know that most of what CRN does is projection and I don't dispute that. You start with an assumption about the future and then explore the implications. Thats fine. What I've argued here, in tedious detail, is that your particular statement about 2025 is a prediction.

Why does it matter?

In the first place, its not a good sign when an executive director expresses the fundamental principles of his organisation in such baffling contradictions. I don't know about you, but alarm bells start to go off in my head when I read anything that sounds like doublespeak.

But more importantly, all your projections are greatly weakened if they are based on a flawed assumption. Does this not bother you?

Richard Jones

Jamais, your explanation is very eloquent and would be convincing if it wasn't so utterly at odds with the tone of CRN's own writings. In the page introducing the Nanotechnology Scenarios, there's no talk of conditional acceptance of a working assumption. There's no talk of "if" - it's all "will":
"Future generations of nanotechnology will use sophisticated nanoscale machinery to construct powerful products with molecular precision."
"Molecular construction will lead to revolutionary capacities, including tabletop fully automated factories capable of constructing duplicate factories in less than a day."

I'm struggling to see how you reconcile statements like these with a view that the future remains unwritten.

Brian, I'm not entirely sure what you are driving at when you ask me whether or not I believe in things like printable electronics or DNA nanotechnology. I think all of these things are fascinating and potentially important technologies. In the cases of plastic electronics and DNA nanotechnology, I've been following them closely for many years (for plastic electronics, ever since the field was invented - I was at the time a junior faculty member in the same research group in Cambridge in which the discovery of polymer LEDs was made in the early 90's), and I wrote about them in my book as exemplars, in different ways, of the soft nanotechnology paradigm.

What I don't believe is that these technologies are in any way destined to lead to molecular manufacturing along the CRN model, for reasons I've written about at length elsewhere. In fact, the way CRN constantly views new technologies solely through the prism of the degree to which they might contribute to the development of their view of molecular manufacturing seems to me to be positively deranging, because it blinds one to their possible true impacts. What plastic electronics might (and I stress might) lead to is true molecular electronics, what DNA nanotechnology might contribute to are the sorts of optical and electronic metamaterials that could make quantum information a reality, or the sorts of active devices that could lead to truly nanoscale medical interventions.

Where I do think I agree with Brian is that many of the societal implications that CRN talks about do not actually depend on their vision of molecular manufacturing at all. Certainly, it's developments in conventional and printable electronics that's going to make universal ambient computing and "sousveillance" happen, without any need to wait for molecular manufacturing at all.

And that points to the central flaw in Mike's argument that it doesn't matter when molecular manufacturing arrives, it's enough to be prepared for it when it finally does. Things don't stand still while you are waiting. Either society will have changed greatly for the worse because we haven't been able to adapt to the challenges we face, in areas like energy and the environment. If this is the direction things are moving in, CRN's insistence that technological salvation through MM is just round the corner will prove to be unhelpful, to say the least. Or society will have been transformed by an array of non-MM technologies, whose implications will have escaped CRN entirely, because they've been so focused on their own imagined version of the future.


Richard, We both agree upon one of the points that I was trying to make that the social impacts predicted decades ago and in the following years about what will happen with MM will happen and is happening now without it.
1. the title of this article sousveillance and fabrication
2. Mass production of cheap UAVs as weapons
3. Surgery in the body with robots
etc... all happening now with more "mundane" tech.

So points that you have made in the past Richard, that there need to be more tame scenarios or scenarios where technology does not advance is wrong. Those would be waste of time scenarios. Along the lines of what if we had no smart phones. So there should be no scenarios around Sousveillance that do not take into account billions of videophones and camcorders.

Quantum information a reality does not depend upon DNA nanotech.

Dwave systems looks like it will produce several thousand qubit adiabatic quantum computer in 2008 Dwave has the money to continue to improve the quality of their qubits.

Trapped ion has some scalable architectures which may compete with AQC

Many possibilities with energy

From mundane
Enhanced oil recovery

Bakken oil field

More regular nuclear build

Flex fuels, biofuels and capacitor/battery hybrids

To high potential high tech:
50% power uprating of existing reactors and newly built reactors of current types Westinghouse working on it. High probability of being commercial 2017.

Fuji Molten salt reactor (low level of funding), French, India and Czechs also working on MSR Could be working 2014.

Bussard IEC fusion could have significant prototype proof this year (2008)

Uranium hydride reactor (funded) could be working 2012. Private money at risk.

General Fusion (funded) approach Could have prototype 2012 and commercial by 2014.

Tom Craver


Are you arguing that there will never be desktop scale devices capable of making copies of themselves as well as making other high capability devices?

Or are you saying that CRN's "projection" of that capability for around 2025 (plus or minus maybe 10 years) - is too precise or soon, and therefore amounts to an incorrect prediction, whether intended as such or not?

CRN projects potential technologies enabled by MM in order to illustrate the impact, and yes, some of those impacts may come earlier. But that doesn't mean that there won't be NEW technologies that we can't conceive of until we've got the underlying technology of MM desktop fabbers. CRN obviously can't use those for illustrative purposes - they have to use known potential technologies.

I also don't buy an argument that we won't get MM or that it won't be useful, simply because alternative technologies may produce many or even most of the projected economic/social *product* impacts in advance. If nothing else it'd be extremely useful as a lab tool, for use in a bootstrapping space colony, and for hobbies.

Even if it is slower and less efficient than every single-product manufacturing technology, convenience alone could lead to MM displacing much of the latter.

Richard Jones

Brian, you make my point for me - for a quantum computer, for example, there are maybe half a dozen quite different ways one could imagine implementing a quantum computer, all quite plausible, but none absolutely assured of success, either technically or from the point of view of making economic sense. It would be very unwise to prematurely focus on one particular imagined version of this future technology, but equally one should try and think through the consequences should any of these possible futures come to fruition.

Obviously it makes no sense to think of scenarios which don't take account of technologies that are already here, and I've never argued that. What I did argue was that it made sense to consider scenarios in which, say, Moore's law plateaued out, so the rate of hardware innovation slowed. This doesn't at all mean that there wouldn't be a lot more innovation in general in this kind of world, it's just that it would have a different character to what we're used to. I do insist, though, that it is prudent to consider situations in which we get technological regression due to unfavourable societal changes, perhaps due to energy shortage, war or environmental disaster. I'm not at all a Peak oil-style doomster, but I do think its important to remember that what we think of as progress isn't inevitable at all, and there is a certain fragility to the socio-economic system we've got used to. Frankly, I think not to appreciate this is to live in a fool's paradise.

Tom, there's more than one question in there and they have different answers. Briefly:

Is the prediction of CRN-style nanofactories by 2025 likely to come to pass? In my opinion, almost certainly not. This answer doesn't really depend on an opinion on whether they are technically feasible; it comes from an appreciation of the scale of the tasks proposed and the amount of work that would be required to turn the very sketchy ideas about how it might work into concrete, tested designs. A good calibration of the likely speed of progress comes from realising that Eigler's demonstration of writing IBM in atoms was made in 1990; this is a good date to mark the beginning of the mechanosynthesis project, and we're now pretty much halfway between 1990 and CRN's projected finish date. There has been progress since then but it's been slow and we still don't have an experimentally verified candidate for the basic elementary operation of mechanosynthesis. Now look forward and try to project all the work that would need to be done on those many parts of the CRN vision that remain at the level of rough sketches, at best ... making designs for motors and systems to handle energy inputs, designing machines and mechanisms that account for the different properties and responses of materials at the nanoscale, handling and sorting systems for the inputs, all the systems integration aspects. When I read sweeping assertions that it could all be done in 10 years with a few tens of millions of dollars, my reply is simply, fine, break the project down into tasks and show me a Gannt chart - I think those sort of statements simply betray an ignorance of what's involved in big technical projects of any kind, and of the specific concrete steps that would be involved in this project. And as I stress, this is even assuming that everything works as projected.

Which brings me to the second question, which is whether a CRN style nanofactory is possible at all. No-one knows the answer to this; what we can say is that the often used, and apparently compelling, argument from biology is wrong, because it is now clear that the physical principles used by biological assemblers are quite different from those envisaged in the mechanisms of CRN nanofactories. (Of course, this leaves open the important and interesting possibility that one can make complex functional nanoscale devices using the "soft nanotechnology" principles that biology uses - and this, of course, is what underlies the fast developing field of DNA nanotechnology, among many other possibilities). We also have a number of "known unknowns" that need addressing - I've written about these elsewhere, and I haven't seen any satisfactory answers to these challenges yet. And we can be sure that there'll be plenty of unknown unknowns as well. One important question is the issue of the environment in which these devices will operate - it is possible, for example, that people will make simple Drexlerian-style hard nanotechnology work in ultra-high vacuum conditions at 3 K. This would be a fascinating achievement and worth pursuing, but it would probably be very limited in terms of its economic impact.

Which leaves the final question of, if it's possible, will it be economically viable enough to make people invest the money and effort to make it work? Again, as I've written elsewhere, I'm puzzled at the emphasis CRN always places on making things. It seems to me that new industries in the past have generally arisen from things that have entirely new functionalities, rather than from things that do something that was possible before in a more convenient way.

Brian Wang

If $343B in the USA is being spent on R&D ($40 billion of which is VC money) then the better R&D alternatives should be advocated.

R&D spending in the world

I look at most of the alternatives and advocate higher impact winners.

Unwise for who to focus on a narrower set of possible futures?

Some groups and companies provide future scenario generation services. Wide a range as possible. There are some who exclusively focus on climate change (Al Gore and his followers) or peak oil or environmental disaster and do quite well financially and with membership.

Moore's law plateau is ignoring work that is already here. Only a narrowly defined set of CMOS lithography timetable delays could be plausible, but in terms of progressing computer power and cheaper prices then any imminent plateau is wrong and ignores the competing solutions that are available.

>technological regression due to unfavourable societal changes

This statement ignores the lives that the economic growth progress machine of societies and nations have repeatedly shown that have been willing to sacrifice for economic growth and tech progress. China lives with 2 million pollution deaths to achieve 10-12% growth. Europe 200,000. The UK 40,000. The USA 60,000. The USA is willing to go to war to secure necessary oil resources.

Tech progress is not inevitable. Africa has made choices that have allowed tech progress to pass them by benefit wise. China made choices that caused 60+ years of delays. But it is also folly to think that the US and Europe and now developing China will not do what it takes to keep tech progress and economic growth rolling. Talk of peak oil ignores several recent big finds of oil and new enhanced oil processes and that places like ANWR will be drilled if things are desperate enough.

Richard, you go broad on possibilities when it suits your argument (many ways to solve quantum computing) and then narrow when it suits your argument. Only diamondoid mechanosynthesis desktop factories.

Could DNA nanotech or other processes produce molecularly precise building blocks for assembly by distributed and highly productive systems ?

Could current 50K-750K rapid manufacturing machines retain most capabilities but be produced at a personal price of $5K or less. $1K or less.

Could there be advances in the cost and ease of creating and working with vacuum and low temp conditions ? Magnetic refridgeration, robotics,

So although particular tech progress is not inevitable, how limited would some of the scenarios with some problem be ?

I agree that it is the new products and capabilities that are possible that matter more than more and cheaper current products. Ralph Merkle has a slide that he shows where what we currently can make is a dot and what would be possible is a larger blob that does fill the sheet which represent the total build space. This slide which he has been showing for over a decade addresses the incorrectness of the statement about emphasis on making things and not on new industries.


A clarification about my point about the lengths that people will go to maintain economic growth and progress. Clearly it is not the first choice to make a few bucks more and sacrifice other people. However, any scenario where people forecast "and then hardship and deaths ensue when Peak oil or Climate change or something else happens" then ignore other scenario options of "and then one tenth of the suffering, cost and death were encountered when growth was maintained but hard choice X, Y and Z were taken to avoid the worst effect of Peak oil, climate change etc..." Or where no death and suffering resulted but politically distasteful and unpopular choices A, B and C were taken but were made possible because of higher costs or where high prices created a market opportunity for other innovations to be introduced.

Tom Craver


Regarding the 2025 date, I think you're neglecting the "take-off effect". Progress on integrated circuits prior to 1958 was zero. It took only 13 years to go from the idea to the 4004 microprocessor, 16 years to the MITS microcomputer. Lack of progress to date means little, given how little lab work has been done.

For "CRN style" nanofactories the rough equivalent breakthrough might be figuring out how to consistently bond atoms to form a designed 3D crystalline structure.

From that point, anticipation of nanofactories would drive progress toward that goal. I'd say about 4 years of prototyping and learning design rules, and a year to design and build the "4004" of nanofactories - an externally driven and controlled unit (no nanomotors, minimal internal control logic) able to build a copy of itself.

The first generation microprocessor wasn't much help for making the second generation, but the first working nanofactory will enable rapid prototyping of its replacement.

Likely we'd jump from the "4004" to the "8086" in 2 years - a nanofactory suitable for use in massively parallel, scaleable arrays. Another 2 years of working out the kinks to accomplish that scale up to desktop size.

By that time, things like nanomotors will have been worked out, and a second generation will likely follow within a year. So about 10 years to get to the desktop nanofactory equivalent of the MITS microcomputer.

Back-timing from 2025, we'd have to get the "bonding breakthrough" by 2015 to be on schedule. But since work is already under way toward that end, I'll optimistically estimate 2010 as the breakthrough date, 2015 as the first nanofactory unit, and 2020 - plus or minus a few years - for the desktop nanofactory.

Sorry, that's as close to a Gannt chart as I can do - but beyond that first breakthrough step, I don't see any obstacles that couldn't be overcome with standard engineering practices and applied research. Perhaps you can name some?

There are other possible paths - e.g. using random assembly methods to create some sort of nanoblocks. Work on DNA appears to be heading in that direction. Probably within five years nanoparticle assembly will be common enough that someone will start stacking them using an AFM and get the idea of making a tiny "robot" to do the same thing.

Three years of prototyping to make a first crude nanofabber unit and attract sigificant investment. Another six to refine the whole concept and get to a crude desktop nanofabber in 2022, plus or minus.

Or that first crude nanofabber might be improved over maybe 5 years, to the point that it enables a bonding breakthrough, followed by another 5 years to get an atom precise nanoblock fabber scaled up to a desktop nanofabber - 2023 plus or minus.

So maybe CRN won't do it, but I will *predict* that by 2025 we'll have desktop nanofactories of one sort or another. Likely much sooner.


Mike and Jamais, I think that the CRN website suffers from having too many unsubstantiated claims and being heavily biased. The timelime prediction, unsubstantiated by any clear reasoning, is an example. Richard's gantt chart suggestion is a good one. Even a simple list of a dozen milestones would be a start.

Here is a link to the 6 challenges for molecular nanotechnology written by Richard over 2 years ago. Does CRN think that any of these have been tackled yet? What milestones have been reached with any of them? In fact in two years have you even bothered to update your list of 30 ESSENTIAL studies and subquestions to reflect them?

I've already pointed out that after 5 years of CRN's existence you've had to move your predictions back by 5 years. This only confirms how very slow progress is.

Another example of an unsubstantiated claim is in your essay from 2 weeks ago "On the point of whether or not molecular manufacturing is feasible, CRN and our allies apparently have won the argument." When did this happen? How can you claim to be responsible if you don't give a fair and balanced view?

Tom Craver

Regarding Richard's 6 challenges:

1-3 and 5 can be mostly answered, from an engineer's perspective, by simply "So make it bigger!"

Drexler's original vision and designs certainly involved the idea of actual molecules shaped roughly like machine components, but if that proves to be unworkable, it's perfectly reasonable to decide to develop a nanoscale technology that backs up perhaps an order of magnitude in scale. It may have lower performance than the hypothetical performance of the smaller scale - but real performance will always beat hypothetical performance.

Or if rotating axels or gears turn out to be a problem, design with sliding parts. Etc. Engineers adapt designs to the possible, and then push at the boundaries to see what else may be possible.

#4 on Richard's list - a motor - certainly has to be addressed...eventually. But a first primitive nanofactory could be completely powered from outside, mechanically distributing power in a manner analogous to the way early factories distributed power through a system of axels and pulleys from a centralized steam engine.

And his #6 amounts to saying "no one has made the breakthrough necessary to kick off "hard" nanotechnology." Yep, quite true, but that's hardly a reason not to try to accomplish that goal!

He also adds a point about possible difficulties of Drexler's concept of bootstrapping "hard" nanotech from "soft" nanotech, but again, the comment amounts to "no one knows how to do this". True enough, but that should hardly be a reason not to research it.


Vince, look through the comments to that 6 challenges article and you can see many of my responses at the time. Here is one of the many responses that I had of why the 6 challenges were mostly wrong headed

Boostrapping from soft/wet to hard. DNA used to place millions of metal nanoparticles. Not to quite the scale and precision that we might like but it is bridge/a path from soft and wet to hard and dry. The other is that make a structure with DNA and then substitute with chemicals and bonds that will retain shape when dry.

DNA pistons can shift whatever they are attached to by a few nanometers.

The more comprehensive list of nanofactory challenges

Specific progress to Diamondoid MM
the Freitas and Merkle work

They have performed computational chemistry for the complete basic toolset that is needed. They are working with the experimentalists to bootstrap the toolset. They have worked out the issue of placing the dimers so that the surface does not reconstruct.

XYZ-Piezo-Scanner for AFM Provides 25 Picometers Resolution The minute P-363 PicoCubeĀ®, together with its ultra-low noise E-536 driver/controller, provide significantly higher resolution and positional stability than previous multi-axis scanning stages.

Continuing advances are being made, but are ignored by those who are only looking for why something won't work. You can also come up with points about why a combustion engine or an airplane should not work, but are only stopped by the fact that they are in everyday use.

In terms of plans for getting there, there is the recently published several hundred page roadmap

Far more than a list of a dozen milestones in terms of the list of work to be done and the published roadmap. also, freitas and merkle have a private roadmap for their own work. Those who fund them get to see it.


Tom, the challenges outlined are not wholly related to the size of the thing built. Building bigger things does not always help and sometimes it makes it worse.

For challenge 1, surface reconstruction and intermediate structure stability, scale is irrelevant. The atoms in the diamond surface cannot get bigger. This challenge is not one with a single simple answer. Every atomically precise structure designed will face this ongoing challenge.

Here's an analogy that might help you understand the challenge. In the game of Pick-up sticks or mikado you let the sticks fall into a random heap. Think of this structure as being the arrangement of atoms you have designed using nanoengineer. Can you build an exact copy? If you have very steady hands and can take each stick off in turn then it should be possible to do the sequence in reverse and build it back up. But often the sticks lean on each other in a complex way. When you move one, all the others move too. So you might think you need to hold the others in place first. The problem is that according to the rules of the game you can only use one helper stick, or one hand, to move the others. This might also be the case in mechanosynthesis if the tooltip is large and fat. Sometimes in pick-up-sticks the arrangement of the sticks is so sensitive to movement that its not even practical to hold them in place if you cheat and use two hands and a vice. Obviously in mechanosynthesis you would want to avoid having to make components where the atoms lean on each other in a complex way like this. But chemistry tells us that atoms like to behave like this in molecules and surfaces more often than not.

If you want to explore the scale of the problem further than lets think about the mechanosynthesis simulation work that Freitas has done. These structures are analogous to simple rows of sticks laid side by side. Successive layers run perpendicular. This structure is a good start, but its edge sticks tend to move around easily and roll away. You can make the edge sticks stay in place by making them lean on each other in a complex way, but this is where we hit the problems about not having enough hands. It might turn out that in reality all the good atomic structures, the ones which stay together, have atoms pressing on each other in a complex way making them very difficult to build. This means that the components which we need to build a nanofactory (tooltips, cogs, motors) might be much more difficult to build than the simple arrangement Freitas has worked on.

The late Richard Smalley wrote about this sort of problem and introduced the terminology "fat fingers" and "sticky fingers". Any analogy is limited and should be treated with caution, but if you think of how difficult it might be to do pick-up-sticks in reverse you get an idea of the problem.

For challenge 2, making things bigger will make the problem worse. The deflection of the end of a structural member in a mechanism increases with its length. A good analogy is comparing the ease of walking on two legs compared with walking on stilts. Alternatively think about playing pick-up-sticks with a very long helper stick. It makes it more difficult.

For challenge 3, it is not possible to say how much larger you would have to make something to overcome friction problems until you can calculate what the friction is. Scientific knowledge in this area is still not developed enough for anyone to make these calculations. This is an ongoing challenge.

For challenge 5, making things bigger won't help. This challenge is about controlling how things enter and leave the eutactic environment. You want the gaps to be as small as possible to make the filter pump be more efficient. This challenge is also ongoing.



Why would we want to create arbitrary difficult tasks for ourselves ? If we are constructing a log cabin, do we first dump all the logs into messy piles ? We try to simplify and organize at every step.

There are plenty of ways for semiconductor chips to get messed up but the hundreds of process steps are organized to avoid the problems.

That is why Freitas is working on simple arrangements. By transition from simple arrangement to another simple arrangement then the problem can be solved.

Every problem does not have to be solved if we can engineer around it. If we are making a highway or railway line and we are unable to solve the problem of going across the grand canyon or tunneling through a mountain range then we find paths around the obstacles. The end result is a completed highway or railway between the points that need to be connected.

The Wright brothers did not solve or understand all the aspects of airflow and fluid dynamics but they knew enough and were able to make the planes work.

For eutatic environments, we may not have to make it perfect, make it as good as you need to and improve later if you have to.

Make thing and money with what can be achieved so that you can get the revenue to support making even more money from improved systems.

On Smalley, he had his own agenda which getting his own visions and work funded.

Smalley was pushing (and his colleagues continue to push for a nanotube energy grid) He first proposed it in 1995. He and Rice have received millions in funding to realize this vision. Where is it ? It has been 13 years. Has any nanotube project been put forward by a utility ? this vision would take trillions of dollars to realize and billions in research. So why is it more worthy or credible ? Why are the problems and challenges that are preventing the nanotube grid from happening easier to beat than what must be overcome for nanofactories or diamondoid molecular manufacturing ?

Back in 1996-7, Smalley was talking about being able to grow carbon nanotubes that long enough to need to be placed onto spindles (pretty much unlimited length.) He talked about being able to do it in a year to two. Did not happen. Even though he and that industry had a lot more money and people working on it.

So it is not always clear what will turn out to be the harder of two unsolved problems or whether what appears to be a problem even needs to be solved for progress to be made.

For engineering, there are the choices of find ways to fix it or workaround and minimize the negative impact.

The comments to this entry are closed.