Military Nanotechnology Debate
From Wise Nano
This page chronicles a debate taking place on the CTF mailing list. The debate started in April 2007 and has not yet ended.
Contents |
Michael Anissimov
A book that everyone on this list should read:
http://www.amazon.com/Military-Nanotechnology-Technology-Contemporary-Security/dp/0415371023
Proposals in the book include banning all missiles and mobile platforms under around 0.5m in size, banning all sensors lower than a few cm in size, upholding the ban on space and biological weapons, extending chemical weapons treaties to encompass anything under the size of a cell, etc. Discusses preexisting treaties and verification methods.
The book is short and very informative. Downside = somewhat expensive. Did you know that US investment in military nanotechnology outweighs the investment of all other countries by a factor of 4 to 10?
The book does a comprehensive overview of military research in NT. It addresses NT and MNT technologies and their military applications in some detail. Dr. Altmann is definitely one of those people "with a clue", but in this book he primarily focuses on NT rather than MNT, although the issues blur somewhat.
Michael Anissimov
Brian Wang
abstract on the altmann book http://www.ep3.rub.de/bvp/milnttocab.html#3
http://advancednano.blogspot.com/2006/12/military-nanotechnology-book.html
The bans do not seem realistic. It involves rolling back a lot of stuff that has already been made including commercial and medical use sensors and devices.
Brian Wang
Michael Anissimov
Note: by "a factor of 4 to 10", I meant, the US funding of military nanotech projects is 4 to 10 times greater than that of all other countries combined, depending on what you count as "nanotechnology".
Brian - you are right, some of the bans, such as against small sensors, seem unworkable. But many are. Bans on mobile robotic platforms under ~0.5m for example, seem almost essential. Other bans discussed in the book include treaties against qualitatively new forms of nuclear weapons delivery - i.e., walking robots. I hope to publish a larger review of the book that goes into the bans in more detail.
I in particular would like to see regulations putting a ceiling on available computing power in the MNT age, but doubt that such a ban would be enforced in the absence of some big event (thousands or hundreds dead) as a direct result of computing power availability. This could occur, for instance, if a non-state actor were able to use high-fidelity simulations of nuclear explosions to design an effective nuclear warhead without IRL testing.
If such a ban is not enforced, it eventually become become child's play for anyone to kill anyone using a microbot equipped with a powerful neurotoxin. The ultra-rich may be able to create secure perimeters around their homes, but for everyone else, we'd be at the mercy of one another's whims. It will not be possible to trace assassin bots back to their owners.
Michael Anissimov
Tihamer Toth-Fejel
Michael Anissimov wrote:
>Note: by "a factor of 4 to 10", I meant, the US funding of military
>nanotech projects is 4 to 10 times greater than that of all other
>countries combined, depending on what you count as "nanotechnology".
As far as I can tell, it's mostly near-term stuff.
Brian - you are right, some of the bans, such as against small sensors, seem unworkable. But many are. Bans on mobile robotic platforms under ~0.5m for example, seem almost essential.
But all medical robotic platforms will be under .5m. I'm with Brian on this one; the robots are under development now; by the time the political will (and the mechanism for enforcing the ban) emerges, it will be 20 years too late. In fact, I would say that Brian Wang understates the ineffectiveness of bans.
>Other bans discussed in the book include treaties against qualitatively new forms
>of nuclear weapons delivery - i.e., walking robots.
Why are walking robots quantitatively different than a 4x4 driving over the border?
>I in particular would like to see regulations putting a ceiling on
>available computing power in the MNT age, but doubt that such a ban
>would be enforced in the absence of some big event (thousands or
>hundreds dead) as a direct result of computing power availability.
Even that won't do it. How many people have died in fire since we started using it?
>This could occur, for instance, if a non-state actor were able to use
>high-fidelity simulations of nuclear explosions to design an effective
>nuclear warhead without IRL testing.
Nuclear warheads are so 20th Century. If I was an evil genius with enough computing power, I would take 1% out of everyone's bank account and put it in mine. Or just wipe my target's hard drives. Or their minds, so I have an army of well-dressed zombies at my beck and call. Computing power will not gather the necessary uranium or plutonium isotopes - though nanotech will.
Caveat: some people like destruction for destruction's sake, or because it's the most honorable way to get to those 42 virgins/day (though there may be ways to get to that level of fun in this life, just by turning on their pleasure centers).
>If such a ban is not enforced, it eventually become become child's
>play for anyone to kill anyone using a microbot equipped with a
>powerful neurotoxin. The ultra-rich may be able to create secure
>perimeters around their homes, but for everyone else, we'd be at the
>mercy of one another's whims. It will not be possible to trace
>assassin bots back to their owners.
No, No, Maybe.
1. The microbot has to physically get to the intended victim. That means that it needs to deal with Reynolds number first, and then the Utility Fog defense that provides a perimeter and volumetric protection against anything larger than 100nm. Plus our personal immune systems will be enhanced to combat attacks like that (mine will be :-)
2. Ultra rich? A few months after the first nanofactory can print utility fog (and copies of itself), the price of Utility Fog will be the same as dirt and sunshine. A lot can happen inside that few months, though (or before).
3. Tracing assassin bots will be as easy as tracing computer viruses; not easy, but might be possible in a transparent, accountable society.
What you *do* need to worry about is cyberwar attacks that turn your own Utility Fog defenses against you.
Tihamer T. Toth-Fejel
Michael Anissimov
Tihamer,
There will be a significant period of time between the debut of the first nanoproducts, made based on variants of the machinery within nanofactories themselves, and the release of much more advanced products such as utility fog, personal shields, corpuscle-unfreezing nanobots, etc. Even if the gap of time is as little as 3-5 years, it is crucial that a regulatory framework be in place for that period of time so that the international societal and political infrastructure is not seriously destabilized.
Medical nanobots are small, yes, but not entirely autonomous. The suggested ban was on autonomous robots. The range of the robotics platform is a central issue here.
Keep in mind that the CRN approach, from what I can tell anyway, is one based on government regulation, preventive arms control, and building nanofactories in a certain way that they are restricted from building certain dangerous products. The plan is not to assume that nanofactories will be entirely unrestricted and that comprehensive nanotech defenses will have to developed to deal with every conceivable abuse of the technology. Is your approach more similar to the former or the latter?<
Nanotech ain't what it used to be - we all started off reading Eric's exciting stories and the ensuing rush of speculation and imagination that followed from them throughout the late 80s and early 90s. The underlying assumption in a lot of this was that if some application is possible, it is nigh certain to be developed and deployed. This is good for creating scenarios to illustrate the full potential of the technology, but I don't think it's the way things will happen in the real world. The first commercial nanofactories will be carefully restricted, the question is, "in what ways?", "who will be making those decisions", and "can we influence them to make the right choices?"
Michael Anissimov
Nato Welch
On Thursday 12 April 2007, Michael Anissimov wrote:
>
> Keep in mind that the CRN approach, from what I can tell anyway, is
> one based on government regulation, preventive arms control, and
> building nanofactories in a certain way that they are restricted from
> building certain dangerous products. The plan is not to assume that
Again, it would probably be a lot safer to restrict the hands that dangerous capabilities can be put in than to restrict the actual product applications.
Far be it from me to advocate not empowering the lowest common denominator, but it's becoming apparent that, once the first actual molecular manufacturing capabilities are available, there is an enormous amount we can do to design nanofactory-//like// fabrication systems that can provide the majority of safe, high-demand applications along with user-customizable design capabilities without really needing to stick atomically precise fabs in every home.
This is what seems to be unfolding out of the "Staggered Nanofactories" thread recently.
> nanofactories will be entirely unrestricted and that comprehensive
> nanotech defenses will have to developed to deal with every
> conceivable abuse of the technology. Is your approach more similar
> to the former or the latter?
>
> Nanotech ain't what it used to be - we all started off reading Eric's
> exciting stories and the ensuing rush of speculation and imagination
> that followed from them throughout the late 80s and early 90s. The
> underlying assumption in a lot of this was that if some application
> is possible, it is nigh certain to be developed and deployed. This
> is good for creating scenarios to illustrate the full potential of
> the technology, but I don't think it's the way things will happen in
> the real world. The first commercial nanofactories will be carefully
> restricted, the question is, "in what ways?", "who will be making
> those decisions", and "can we influence them to make the right
> choices?"
Given the enormous and far reaching effects of the tech, I don't think a "commercial" approach, based on concepts of private property, is easy to justify at all. Sure, commercial interests have a stake, considering how much of their business MM can demolish, but workers and consumers have equal, if not greater stakes. A democratic civic regulation approach, seems to be the only sane way to go about it.
--
Nato Welch
Tihamer Toth-Fejel
Thanks for the thoughtful response, Michael.
>There will be a significant period of time between the debut of
>the first nanoproducts, made based on variants of the machinery
>within nanofactories themselves, and the release of much more advanced
>products such as utility fog, personal shields, corpuscle-unfreezing
>nanobots, etc.
Technically speaking, Utility Fog is very primitive, non-autonomous nanobots - give me a half-way decent 3D printer (with diamond mechanosynthesis capabilities), and I'll have foglets two weeks later (I'll just download the 3DS Max files from my NIAC project). Programming 'em is also easy-- it took only 1000 lines of code that would make a collection of 512 kinematic cellular automata pixilate to spell out EH05 (this was for the NASA/DoD Evolvable Hardware Conference) and then sprout legs and run off. ("Non-Evolvable Indirectly Replicating Nanorobots with Self-Assembling Parts", Evolvable Hardware, 2005 NASA/DoD Conference on Evolvable Hardware (EH'05) pp.183-192, http://doi.ieeecomputersociety.org/10.1109/EH.2005.36) Admittedly real foglets would have slightly more things to worry about (balance, sensory input, GPS orientation, etc.), but the control software, if the foglets are hierarchically distributed, should be identical to the code I wrote. If, OTOH, you try to control the whole thing with distributed algorithms... Well then it could be quite difficult.
Personal shields would be utility fog running the mathematical analogs of anti-viral software. Not too difficult, because it's not autonomous nanobots. In fact, I'm not sure that really small autonomous nanobots are even possible. Heck, we can't even build autonomous macro robots. It depends what you mean by autonomous, I suppose it could be something as primitive as an agent-based controlled robot.
Corpsicle-unfreezing nanobots are new science and hence unpredictable, but my guess would be 5-10 years after we have autonomous nanobots.
The big questions are: When can a nanofactory print a copy of itself? When can a nanofactory print preprocessors so that it only needs dirt and sunshine? When can a nanofactory print diamondoid?
Before these things can be done, the "nanofactory" is just a 3D printer, though replacing a billion-dollar semiconductor fab lab with it is a pretty impressive 3D printer! (that was a great example, Brian!)
>Even if the gap of time is as little as 3-5 years, it
>is crucial that a regulatory framework be in place for that period of
>time so that the international societal and political infrastructure
>is not seriously destabilized.
The transference of trade from atoms to bits is going to cause major dislocations in the world economy, and I don't see how regulation is going to change that. Once an open source flat screen print file is on the web, what will keep Samsung or Panasonic afloat? How are you going to keep people from printing more molecular RepRaps?
>Medical nanobots are small, yes, but not entirely autonomous.
>The suggested ban was on autonomous robots. The range of the
>robotics platform is a central issue here.
Maybe it would be a good idea if you defined autonomous. Is my vacuum robot autonomous? What if it could recharge itself? And do windows when it needed to? What is the tipping point? If I was devious enough, I could probably use the vacuum robot as a nasty weapon (our cat already thinks it's one).
>Keep in mind that the CRN approach, from what I can tell anyway,
>is one based on government regulation, preventive arms control, and
>building nanofactories in a certain way that they are restricted
>from building certain dangerous products.
When it comes to the cussedness of human beings, I'm a realist. After all, anthropologists call the 20th century the Genocide Age.
Anything more complex than a pencil, when multiplied by low-cost 3D molecular printing, can be used as a pretty nasty weapon.
>The plan is not to assume that nanofactories will be entirely unrestricted and that
>comprehensive nanotech defenses will have to developed to deal with every
>conceivable abuse of the technology. Is your approach more
>similar to the former or the latter?
My answer is to run. Not a very good answer, I'll admit, but Space is big. Failing that, Utility fog should make a fairly good defense shield if it's software is configured correctly.
>Nanotech ain't what it used to be - we all started off reading
>Eric's exciting stories and the ensuing rush of speculation and
>imagination that followed from them throughout the late 80s and early 90s.
>The underlying assumption in a lot of this was that if some
>application is possible, it is nigh certain to be developed and deployed.
>This is good for creating scenarios to illustrate the full potential of
>the technology, but I don't think it's the way things will happen in
>the real world.
You're right, not everything that is technologically possible will be built. But if there is an economic reason for developing something, or an ideological one, it will happen, don't you think?
>The first commercial nanofactories will be carefully
>restricted, the question is, "in what ways?", "who will be making
>those decisions", and "can we influence them to make the right choices?"
Was the printing press carefully restricted? The computer? The first web browser? Are RepRap and other 3D printers restricted?
I think we're asking the wrong questions.
Maybe the questions to ask are, "What would motivate someone to do something evil?" and "How can a transparent and accountable society stop them? Or ensure that nobody chooses to do evil? And how do we decide which 3D objects are too dangerous to tolerate?"
Keep 9/11 in mind of what can be done by combining relatively harmless stuff. I don't think prohibiting dangerous things can be done. Look what hackers do with ones and zeros.
Personally, I think that *everyone* on a plane should be carrying a box cutter. Heinlein was partially right: An armed society is a polite society--but only if the citizenry are honorable, kind, and basically good. But how can we make sure that happens? I'm not sure we can.
Tihamer T. Toth-Fejel
Nato Welch
On Friday 13 April 2007, Toth-Fejel, Tihamer T. wrote:
>
>> Even if the gap of time is as little as 3-5 years, it
>> is crucial that a regulatory framework be in place for that period of
>> time so that the international societal and political
>> infrastructure is not seriously destabilized.
>
> The transference of trade from atoms to bits is going to cause major
> dislocations in the world economy, and I don't see how regulation is
> going to change that. Once an open source flat screen print file is
> on the web, what will keep Samsung or Panasonic afloat? How are you
> going to keep people from printing more molecular RepRaps?
As above, I'm not seeing any particularly compelling reasons to release fully capable atomically-precise fabs to the public. Most of the high-demand applications could be provided in micro/nanoblock form for assembly in non-molecular factories/3dprinters/fabs.
In such a situation, those wanting to bypass regulation to create dangerous applications would either have to be licensed and vetted, or would have to create their own factories "from scratch", which would raise the bar of both education and coordination among like-minded individuals.
Even if malicious actors succeeded in creating their own recursive nanofactories, if the only practical applications for them are malicious in nature (because the non-malicious apps are all provided by the sanctioned, legal fab systems), then their popularity will be limited in proportion to the number of people after malicious uses. Chris speaks often about trying to limit the likelihood of a black market by //appeasing users// as much as possible in order to keep them hooked on the safe, controlled fab systems, and I have always appreciated with that strategy.
This is why I find it easy to see the danger in trying to lump in systems designed to protect public safety with systems that attempt to protect intellectual property and/or industrial profits. Class breaks in security systems motivated by a non-malicious (and at times healthy) disdain for industrial greed must never be allowed to enable otherwise relatively marginalized malicious actors with new tools and techniques.
A single system to prevent the production of sex toys should not also be used to prevent the production of bombs - or something deadlier. If some innocent perv in Oklahoma cracks a "Trusted Manufacturing" system to make "love", and then publishes his results, the same info is likely to be usable to make war.
The design of the fabs - and of the larger systems of deployment and regulation - need to compartmentalize not just the safety designs, but the motivations for using them.
None of this is going to be flawlessly perfect; but that's no excuse for not taking a strategy of reducing risks as much as possible.
Tihamer, I don't know if you're familiar with Tom Craver's nanoblocks proposal; I've been a fan of it ever since he first proposed it. It goes a long way to addressing the common concerns you raise, and it has served as the foundation of my own security design thinking ever since.
http://wise-nano.org/w/Craver_CTF_Essay
Nato Welch
