offutt@CAEN.ENGIN.UMICH.EDU (daniel m offutt) (06/01/89)
I have seen no evidence that the task of developing effective active shields is tractable, even if almost everyone on the planet were a "good guy"/active shield researcher. To build something that destroys is easy. To build an active shield that protects a complex system such as the human body or a forest from damage by replicators seems orders of magnitude more difficult. In some cases, certainly impossible. The task of an active shield is not just to stop noxious replicators, it is to do so without disrupting the protected system. And to establish that a candidate active shield does not seriously disrupt its host in any way seems difficult or impossible due to the complexity of the host and the resulting large number of unpredictable side-effects of the resident candidate active shield. Moreover, as Mr. Papiewski pointed out in an earlier article, the great variety of potential gray goo makes the problem even more difficult, since the active shield must be proof against nanoreplicators it has never seen. Animal immune systems have been evolving for hundreds of millions of years. If immune systems and active shields are to be of comparable complexity, they are not going to be developed by only five or fifty billion people in a matter of decades, or even centuries. Dan Offutt offutt@caen.engin.umich.edu [The same arguments could be brought against the possibility of building gray goo in the first place. This is one reason to hope that the two might occur with some parity of capabilities... --JoSH]
offutt@CAEN.ENGIN.UMICH.EDU (daniel m offutt) (06/10/89)
In article <8906010521.AA05561@athos.rutgers.edu> offutt@CAEN.ENGIN.UMICH.EDU (daniel m offutt) writes: > >[ various arguments against the claim that active shields can be developed > soon enough. ] > >[The same arguments could be brought against the possibility of > building gray goo in the first place. This is one reason to hope > that the two might occur with some parity of capabilities... > --JoSH] There is clear asymmetry between the difficulty of destruction and the difficulty of maintaining a complex integrated system such as a building, workstation, standard of living, or human body. To destroy is relatively easy, to protect against destruction is much more difficult. To design, construct and maintain a building requires a wide variety of specialized expertise and equipment, applied continuously from the time the building is designed to the time it wears out. But a building can be destroyed in minutes with simple explosives that the Chinese knew how to make more than a thousand years ago. You could destroy the workstation you are reading this message on in less than 60 seconds. But to fix a small problem with the workstation, much less maintain it in good working order for several years, is a much more difficult task requiring special knowledge and equipment. Maintaining the American standard of living is a complex problem demanding millions of specialized forms of knowledge and other capital. But destroying the American standard of living could be done much more easily. A few dozen nuclear explosions would do it. The erection of trade barriers around the US would do it. How does the AIDS virus compare in complexity with the minds of all the thousands of people who have been looking for a cure for the last ten years? Do they have cure for AIDS yet? No. How long have researchers been looking for a cure for the common cold? Do they have a cure yet? No. The human body can be destroyed by the disruption of just *one* life-critical process. There are numerous simple elements, chemicals, viruses, and bacteria that can kill. Killers can be simple. But an active shield (such as the human immune system) is necessarily much, much more complex. So designing killers should be easy compared to designing a broad defensive system. One reason the immune system (and artificial active shields) must be more complex is that the assaults it must defend against are largely unknown. Killers can be simple and compact, so a wide variety can be generated and pitted by Nature against the human immune system at low cost in time, energy and other resources. (Recall that bacteria are rapidly developing resistance to antibiotics. Recall how recently the Plague occurred, and how recently AIDS has arisen.) This means that the immune system is always running into novel assaults. And that means that the immune system faces substantial uncertainty about what assaults it will encounter in the future. On the other hand, the human body is relatively constant -- over the same time scale. That means that processes designing attacks on the human body face substantially less uncertainty. Uncertainty of the mode of attack implies a need for a diversity of defenses, and greater diversity implies greater complexity. This is why viruses and bacteria are so simple relative to the human immune system. If the immune system must be complex in order to cope with uncertainty of mode of assault, then artificial active shields will also have to be complex in order to cope with similar uncertainty. The human immune system is the product of hundreds of millions of years of evolution, including a like amount of time of rigorous testing against the most noxious assaults Nature has been able to devise against it. The immune system reflects, in some never-to-be-fully-understood way, a long history of mostly unknown assaults by viruses and bacteria. Clearly there are not millions of years available to evolve active shields capable of defending against gray goo. Mr. Drexler and others seem to think that cultural evolution on a time scale of decades or centuries is at least as powerful (in the sense of ability to discover solutions to problems -- such as artificial active shield design) as biological evolution is on a time scale of hundreds of millions of years. I believe this assumption is dead wrong. It needs to be considered very much more carefully by nanotechnologists. Dan Offutt offutt@caen.engin.umich.edu [Very cogently argued. The only hope I can see is that the argument may perhaps be reversed. A self-reproducing gray goo is itself a complex mechanism, of necessity. Thus countermeasures, like antiseptic swabbed on a wound, can occupy the simple, destructive role in your scenario. This only works, of course, if there is an active intelligence recognizing threats and orchestrating the application of the destructive agent. This would mean that a gray could be fought, but also that it must be fought--we could not whip up and rely on some a priori defense. --JoSH]
alan@oz.nm.paradyne.COM (Alan Lovejoy) (06/13/89)
In article <8906130723.AA21670@athos.rutgers.edu< offutt@CAEN.ENGIN.UMICH.EDU (daniel m offutt) writes:
<The human body can be destroyed by the disruption of just *one*
<life-critical process. There are numerous simple elements, chemicals,
<viruses, and bacteria that can kill. Killers can be simple. But an
<active shield (such as the human immune system) is necessarily much,
<much more complex. So designing killers should be easy compared to
<designing a broad defensive system.
And yet still we survive. Your arguments that we cannot survive are just about
as impressive as the "proof" that bees cannot fly or that rockets cannot reach
orbit. Is there a problem? Yes! Is the situation hopeless? No!
<If the immune system must be complex in order to cope with uncertainty
<of mode of assault, then artificial active shields will also have to
<be complex in order to cope with similar uncertainty. The human
<immune system is the product of hundreds of millions of years of
<evolution, including a like amount of time of rigorous testing against
<the most noxious assaults Nature has been able to devise against it.
<The immune system reflects, in some never-to-be-fully-understood way,
<a long history of mostly unknown assaults by viruses and bacteria.
<Clearly there are not millions of years available to evolve active
<shields capable of defending against gray goo.
Your argument is based on three overly-pessimistic assumptions:
1) Equal effort will be expended towards developing gragu and active shields.
2) Nanotechnology which is sufficiently advanced to create gragu will appear
before AI which is sufficiently advanced to speed up technoscientific
advancement by 6 orders of magnitude (or better).
3) The first team to make the AI/nanotechnology breakthrough will either be
inimical, or else stupid enough to freely distribute their knowledge.
The first assumption is probably not true, because most people oppose the
goals of gragu. Gragu will not be an accident. Inimical people will have
to create it. Only the highly insane will consider releasing an
indiscriminately-destructive goo on the world. Most people who put any effort
into gragu will intend to survive their creation. And most of those will
only intend to release the goo for purposes of retaliation to being attacked
by someone else's gragu. Mutual assured destruction all over again.
Military gragu will be designed to be selective and controllable by its owners.
The "control codes" will be the target of the most intense espionage campaign
the world has ever seen. And everyone will seek to subvert everyone else's
goo. Can you be SURE that your goo has not been subverted by the other side?
Remember, your neighbors have nanoagents, just like you do. Will anything
ever be truly secret and/or secure again?
The problem isn't gragu--it's inimical intelligences. Perhaps the best way
to prevent gragu is to prevent the sicknesses, abuses and depravities that
engender insanity and evil. Nanotechnology and AI may provide us with the
capability--and the imperative--to heal the world. Can we learn enough
about what makes people willing to commit mass murder to prevent it from
happeing? We'd better!
The second assumption is probably false because a gragu agent would have to
be much more sophisticated than a virus or bacterium--and we aren't even
close to being able to design such a thing [a gragu agent] without advanced AI
to help us. The human brain, and hence the human mind, is a machine. It
operates in accordance with natural law just like everthing else. It was
designed and built by a process of evolution over billions of years. Note
that most of the design time was taken up to create the most primitive
aspects of the human brain--the really advanced capabilities appeared with
relative rapidity. Consider what that means in light of the current state
of the art in AI. The brain is not magic. If it can evolve, it can be
purposely designed. There can be no credible refutation of this logic.
The rate of progress in machine-intelligence technology is such that artificial
human intelligence will almost certainly appear before 2050. Idiot savant
machines that are very good molecular engineers will probably be available
by 2020--or sooner. It is the capabilities of these AI machines that will
determine the success--or failure--of active shields.
The third assumption is probably false because most scientific researchers
are not inimical--nor are they stupid (if they are, they're in the wrong
profession).
<Mr. Drexler and others seem to think that cultural evolution on a time
<scale of decades or centuries is at least as powerful (in the sense of
<ability to discover solutions to problems -- such as artificial active
<shield design) as biological evolution is on a time scale of hundreds
<of millions of years. I believe this assumption is dead wrong. It
<needs to be considered very much more carefully by nanotechnologists.
The idea that cultural evolution proceeding at the rate it does today can
produce effective and reliable active shields quickly enough to counter
gragu is almost certainly wrong. That is NOT Drexler's argument.
Drexler argues that AI--and other advancements--will drastically accelerate
the rate of progress by many orders of magnitude. The first team to use
nanotechology to create a "super computer" will probably be able to achieve
and maintain an unassailable technological superiority over everyone else,
if they so choose.
Alan Lovejoy; alan@pdn; 813-530-2211; AT&T Paradyne: 8550 Ulmerton, Largo, FL.
Disclaimer: I do not speak for AT&T Paradyne. They do not speak for me.
______________________________Down with Li Peng!________________________________
Motto: If nanomachines will be able to reconstruct you, YOU AREN'T DEAD YET.harry@moncam.UUCP (Jangling Neck Nipper) (06/15/89)
> There is clear asymmetry between the difficulty of destruction and the > difficulty of maintaining a complex integrated system such as a > building, workstation, standard of living, or human body. To destroy > is relatively easy, to protect against destruction is much more > difficult. This is not true; the sophistication of the attack determines the sophistication of the defence. Something like a physical blow simply needs to be avoided, but something like a complicated toxin, or a virus need more sophisticated defences. Also, I think you are confusing the defender and the defence response here; any system is the sum of all its attacking modes and its defending modes. The reason that the immune system is so complex is because it has to deal with lots of >different< kinds of attack, but on a one-to-one confrontation of, say, toxin to anti-toxin, the complexity is similar, and, moreover, the complexity of >producing< both the toxin and the anti-toxin are also similar. Anything capable of producing >all< the toxins of >all< the bacteria &c. would be as complex as the immune system; the `simplicity' of a toxin is because it >is< just one toxin, and not all of them. In fact, if you look at warfare, the aggressor's technology is much more complex and costly than the defender's, mainly because attacking invloves lots of moving around, which is more diffult to do than staying still, eg, tanks are more complicated than mines.
nagle@lll-crg.llnl.gov (John Nagle) (06/16/89)
One approach that would work, although it would be a rather dull
society, would be a society of space habitats with very limited physical
interaction, each equipped with defenses capable of vaporizing anything
incoming down to the molecule level. This is a last-ditch solution, though.
It may turn out that nanotechnology is power-limited. In a vat, with
power and suitable working fluids supplied, you may be able to achieve
high-speed assembly. In the outside world, the available power may limit
how fast an attacker could propagate. It may well limit how effective
nanomachines can be without macro-scale support systems. But this remains
to be seen.
John Nagle
John Nagle