Slashdot Mirror
Drexler Clarifies Grey Goo Scenario
b00le writes "The BBC says that the scientist many regard as the father of nanotechnology has backed away from his famous claim that runaway nanomachines could turn the planet into 'grey goo'. Eric Drexler now says nanomachines that self-replicate exponentially are unlikely ever to enter widespread use. So that's all right, then, but he also said 'tiny machines would need close control' - which not everyone would agree with. I always imagined some kind of emergent behaviour would, er, emerge." Bill Joy is still suitably pessimistic.
Straight from the Outer Limits episode. These "nanobots" turned a man into something of a jellyfish and he had gills as well. Of course as in any good Outer Limits episode, the "abort" command issued to the nanobots failed. But then, thats just a television show, right? These nanomachines couldn't REALLY churn through every nanogram of matter on our planet, RIGHT? IHMO, the Martian Sand Kings episode was way cooler, I mean they ate a dog for christs sake. Those beasts would mangle some nanobots. Thats it...we just need a bunch of sand-dwelling cockroaches with fangs on methamphetamine to regulate the reproduction of nanobots.
"Quis custodiet ipsos custodes?"
... whatever you do, don't let director Roland Emmerich get ahold of this article!
RS
Shoes for Industry. Shoes for the Dead.
When you outlaw exponentially self-replicating nanomachines, only outlaws will have exponentially self-replicating nanomachines. That's just not a world I want to live in.
We are borg, resistance is futile, you will be turned into grey goo! or not... well, we realy don't know!
The biggest problem with the grey-goo scenario is that it requires an astonishing amount of work (tearing apart molecular bonds and using the resulting material to make an extremely complex machine) without taking power consumption into account. Getting energy to a machine that small is extremely difficult (your body has to basically immerse it's cells in fuel to keep them going). A machine that small recieves an absolutely puny amount of sunlight, and Tesla style distributed power doesn't work over long distances. Worse, the energy potental of almost every material on the planet is far too low to be useful in powering a tiny machine (you can't power a robot with dirt).
This problem, coupled with the fact that the nanotech people have barely demonstrated anything even remotely close to grey-goo yet, lets me sleep easy at night. There's no need to get so worked up over vapor.
I read the internet for the articles.
If they could turn the world to grey goo, bacteria would have already? Well, I suppose it's multicoloured goo really. But wouldn't anything that can reproduce uncontrollably be just as affecte by the pressures of the environment as any other living organism?
folks did nobody read PREY, by Michael Crichton... little nanorobots, evolving and becoming WAY too smart for our own, good... thank goodness for parallel processing
...damn, there is *always* a tone change in the front page stories when Michael is up to bat. This is not a troll; it is an observation. When he is at the wheel, it's all end-of-the-world, privacy, government related stuff. Go ahead, check his history.
As for nanobots, honestly, we had this discussion and i hold the same view: tread lightly. You and i both know that if something were to become easily synthesizeable by the layman, nanoweapons in this case, and were to be exponentially self-reproductive, then...well, the human race would not survive it. Think about that, no one person in the human race could have "a bad day". Most are not intelligent enough to have a healthy respect for the miracle that is human life.
When I was a kid we were obsessed with large robot machines. Now these few short years later we are concerned with the tiniest of machines.
I'm going with the big ass machines. I'll always win the mine is bigger than your contest.
Evolution or ID?
We're all just human. 50 years ago, they predicted that we'd be zipping around in flying cars-- and no one at all predicted the huge impact of the Internet. We don't know if self-replicating nanobots will ever enter the market. For that matter, we don't know if the grey goo scenario is possible or not. When they first tested the atom bomb, there were those who feared that the blast would ignite the atmosphere itself-- and until we tried it, we couldn't be sure if it would or not. Today's particle accelerators are creating heretofore-unknown forms of matter, and for all we know, they could create a new sort of matter that would destroy the world. We're just people-- we aren't gods. How can we say "This will happen" or "this won't happen"? All we can say is "We don't think this will happen"-- but that is no guarantee.
Honey, I shrunk the Cygwin
The primary limitation on even arbitrarily sophisticated nanotechnology which could prevent a runaway grey goo reaction is the lack of a sufficient source of energy. A nanomachine wouldn't be able to get much energy out of eating inorganic matter such as rocks because, aside from a few exceptions (coal, for example) it's mostly well-oxidized and sitting in a free-energy minimum.
Wikipedia
It would seem that nature's methods of self-replication work best.
Prey had a really dumb ending anyway :(
building requires consumption of raw materials which is what the grey goo scenario was refering to... the self replicating machines need raw materials to replicate, and the point was that the machines would exponentially reproduce, (doubling at the rate it takes to build a new nanobot) they need material to produce that, they take that material out of whatever is nearby... turning the earth into grey goo...
... whatever you do, don't let writer Stephen King get a hold of this post! I can just see an unwatchably painful miniseries coming out of this...
i am the drexler. i speak for the nanobots.
-ninjaneer
just want some stargate and see what trouble replicating robots/nano machines could get us in..
We do not have to build something smart enough to take over the world.. We don't even have to build something smart enough to learn..
A single machine programmed to take over another machine ( A nice tech to be developed for the military ) is all it would take.
Machine A, Trys to hack machine B. In the combined code has the abilitys of both.. Repeat over and over again and in time it might be able to think and act on its own.
Its sort of kin to programming and various other human tasks..
Take 2 people with 2 diffrent skill sets. Together they could build something that neither could build apart. There tech together might make a doomsday weapon, Apart they are useless.
Personal Website
this image is frightening.
the potential for error with something like this is huge: whoops, programmed the little bugger wrong! sorry, you don't need that hemoglobin, anyway.
It only takes one.
"Asleep at the switch? I wasn't asleep, I was drunk!" -- Homer
(1) Machines only do what you design them to.
Mind you, people often design them wrong, and then the fail to function, but that isn't going to spontaneously create self-replicating machines. Besides, if the raw materials are not available in the right form, they cannot replicate.
(2) Self-replicating machines are prohibitively complex.
Have you had a look at the genome of a simple bacteria lately? How about the support machinery in the bacteria? Trust me, an evil mad scientist would not have the funding or resources to develop a self-replicating machine.
(3) The real problem with nano machines would be simple design flaws, not replication.
If your nano machines are supposed to identify cancer cells and kill them, but they mistake healthy cells for cancer cells, THEN you have a problem. That is a lot more realistic. But a decade of testing on any given design would happen before it was used in humans.
These nanomachines couldn't REALLY churn through every nanogram of matter on our planet, RIGHT?
The whole grey goo scenario is pure alchemy. Except instead of turning lead into gold, we're turning it into grey goo. We've got people inventing perpetual motion, too. Are the 1800s back? Can't we invent new scams?
After a few million years of evolution, we have enzymes. They are generally very large molecules, bigger than what some claim for nano-machines, and they are also very specialized. They do one thing. You don't get anything general-purpose or intelligent at the molecular level, there just isn't room for it.
Organic life has already covered the planet, in green stuff.
I doubt that any man-made gray goo could compete with the Green Goo God made without a LOT of help. By the time we were good enough to make the gray goo beat the God's Green Goo, we would have already made safeguards such as Gray Goo Cops, little nanites whose sole job it is to rome the world looking for rogue nanites and eat them and reproduce more Gray Cops.
Organic based reproducers beat metals based ones before, and they will do it again if the silly puny little machines try to take over.
excitingthingstodo.blogspot.com
"Eric Drexler now says nanomachines that self-replicate exponentially are unlikely ever to enter widespread use."
Why is that still not particularly comforting? Just one tragically (intentional or otherwise) bad design is all it could take, theoretically. Not to turn the earth to "goo", but to seriously screw the conditions we humans deem useful to our existence.
Not a few decades from now, but a century or so down the road when this stuff really picks up and the tools are more accessible. With every step of our advance, we seem to merely reinforce the reality that we're really just fancy homonids with an ever-increasing number of dangerous gadgets, mashing the buttons on the controls.
Humans are so convinced we're a required part of the fabric of the universe. But *poof* Gone. Nobody would care beyond the occasional underpaid archeological student of the next dominant sentient life form.
Maybe I should start planning what kind of confusing fossil record to leave behind. Time to find some cooling lava and a pair of Godzilla shoes.
Bill Joy, while clearly a genius, is (like any good genius) a nutcase. Seriously, the man is paranoid! He's a compulsive risk-mitigator:
This told to the reporter during the interview about nanotech risk-mitigation. Sure, it's a perfectly rational way to choose your movie library, but it's almost too rational. Most people don't consider watching a bad movie an outcome to be avoided at all costs. Mainstream critical consensus is a very conservative method of choosing movies. I've watched a lot of bad movies, but I've found a few that I really liked that were panned by critics. Is Mr. Joy so risk-averse that he needs his movies to be guaranteed satisfactory?
If a job's not worth doing, it's not worth doing right.
Maybe you should have ended that post with a IANAS (scientist) disclaimer. In both high school and college debate, nano was my primary and favorite topic for years, and I frequently debated on both sides of the issue. The one argument that I could never win against nano was an attack on Drexler's qualifications.
Perhaps he should not be called the father of nano. The real father of nano is Richard Fayman. In his lecture entitled "There Is Plenty of Room At the Bottom" he basically invented the concept. Drexler, however brought it forward. He has a Ph.D. in Molecular Nanotechnology from MIT (a degree that did not exist before Drexler was awarded it). His S.M. and S.B. are both from MIT as well. He was a research affiliate for two departments at MIT and a visiting scholar at Stanford, where he taught a doctorate level class. As recently as 1993 he won the Kilby Yound Innovator Award. He has testified before Congress, written dozens of articles and books, even winning the 1992 Oustanding Computer Science Book for Nanosystems, a VERY technical book almost impossible to understand for anyone without at least a M.S. in Chem or Engineering (or both!). He holds numerous patents, and has lectured everywhere from Apple and Bell Labs to TI and the Xerox PARC.
Disbelieve if you want, but please do not be so foolish as to challange the credentials of Dr. Drexler.
It is terribly hard to build your first few nanites. Then you have to look at the replication ratio. How many more of itself can a self-replicator build before it fails? You've got to get the ratio above one.
The likely scenario is that the self-replicators are not robust and we never develop the technology to the point at which the ratio is solidly above one. So civilisation potters along quite wealthy for 50 years, then problems with contanimation, vibration, temperature, something, result in the nanites dying off. It could take decades to recover the lost art of building the first few, decades of great hardship for a society that has come to depend on nano-technology.
These tribbles are everywhere!
- Thomas;
___ This sig is in boldface to emphasize its importance!
... is that it is the ultimate straw man.
It's the most ridiculous possible argument against nanotech. It's like being afraid that a nuclear reactor will turn the Earth into the Sun. And once everyone dispels the straw man argument, we go happily about our merry way, la la la, it's nothing to worry about.
Let me give you some scary: nanobots that go down to the bottom of the ocean and mess with the clathrates, spilling all of that methane out into the atmosphere - there's enough energy there to get that done. Nanobots burrowing into the Earth's crust along fault lines in a long chain, using the temperature gradient to make a heat engine in order to drive any kind of mucking about with tectonic plates. Nanobots carefully and quietly sabotaging subtle but key parts of the ecology.
It's easier to destroy than to create. And nanobots would be able to replicate, with probably greater (but not much) efficiency, anything you could dream up via genetic engineering, because nanotech is going to look a lot like biology plus some nifty physics. And it will be a biology freed from some of the constraints and old hacks Nature imposed.. They could also use physical properties perhaps not accessible to mere biology - how about something really wacky, like point fusion? Nature did a lot of clever tricks, but there's much optimization that could be done.
Yes, grey goo can't be taken seriously, due to physical constraints. The bad guys don't need to destroy the entire planet - they just need to make it unlivable so that $MESSIAH can come. And it wouldn't be tough to spend a few afternoons dreaming up doomsday applications that are not energy intensive.
On a related note, consider this readable account of how genetic engineering to insert IL-4 into an otherwise fairly innocuous mousepox transformed this disease to where it would effectively kill all the mice, even those mice that had been previously vaccinated to protect them against mousepox.
"Provided by the management for your protection."
just wait until robin williams begins to roll out his nanunanutechnology...
Sacred cows make the best burgers.
Drexler *never* said that "grey goo" would consume the biosphere. What he actually said was "Dangerous replicators could easily be too tough, small, and rapidly spreading to stop - at least if we made no preparation." (emphasis mine, see Engines of Creation Chapter 11). It has been known for more than a decade that there are easy solutions to the problem of designing "safe" replicators that do not grow exponentially using strategies such as the "broadcast architecture" (in computer science terms -- you never give a replicator a copy of its own source code). [See Merkle, R. C., "Self Replicating Systems and Molecular Manufacturing", JBIS 45:407-413 (1992)].
Nor is the idea that assembly lines produce better manufacturing systems than self-replicating systems new. [See Hall, J. S., "Architectural considerations for self-replicating manufacturing systems", Nanotechnology 10(3):323-330 (September, 1999).] It is obvious that the ability to self-replicate is extra overhead when compared with assembly systems optimized for specific assembly tasks.
Finally, it was shown several years ago that we have the technology to detect out-of-control self-replicating systems (nanorobots generate heat which can be detected by existing satellite systems). [For a discussion of various scenarios read: Freitas, R. A., "Some Limits to Global Ecophagy by Biovorous Nanoreplicators with Public Policy Recommendations" (May, 2000).]
Drexler alludes to the fact that we are already in the midst of a "green goo" ("We have trouble enough controlling viruses and fruit flies.") Most people are unaware of the fact that they have more copies of foreign genomes (in the form of self-replicating bacteria) on or in their body than they have copies of their own genome. Some of these bacteria actually produce vitamins that humans use. So "goo" scenarios should not be viewed as completely negative. It is worth noting that the same methods that can be used to stop the "green goo" (e.g. heat or radiation) can be used to stop the "gray goo" if we are prepared to detect and eliminate it. One sees examples of this today as government agents circulate through the crowd waiting to view President Regan's body in Washington with biological and chemical weapons detectors. It simply comes down to understanding the hazards and being prepared to deal with them.
It is also worth noting that the design of fully self-replicating nanorobots is *not* a simple or inexpensive task. (Look at how long it took Nature to get it started...) So it is highly improbable that such abilities could be developed by rogue groups before civilized nations developed robust detection and elimination methods.
For people who want to read more details, the IOP press release is here and points to the actual paper (registration probably required).
Also, I would respectfully request before you post any responses to this note that you "go do your homework" (that will put you one up on the reporters reporting on this and allow for an informed discussion).
Eric Drexler now says nanomachines that self-replicate exponentially are unlikely ever to enter widespread use
No, that's not what he said; that statement is an oxymoron. If something self-replicates, its numbers necessarily grow exponentially until it hits resource constraints in the environment. There are no "nanomachines that self-replicate sub-exponentially".
What Drexler said that nanomachines that self-replicate are unlikely to ever enter widespread use, and therefore nanomachines will not replicate exponentially. Instead, they will be manufactured by desktop machines, according to him.
Nobody knows how to make molecular assemblers anyway, yet alone self-replicating nano-bots. Many scientists say Drexler's ideas would not work in any case.
... There are much more important technological threats to the environment to worry about in the real world.
Look at it this way - we have self-replicating nano-bots right now - they are called bacteria. Have they turned the world into gray goo in runaway exponential growth? Are we going to be able to make more efficient nano-bots than mother nature has done in the last 4 billion years?
Bill Joy's worries about nano-bots are like saying we should stop all research into magic because we could set off a chain reaction that would turn us all into frogs. Nano-bots are FANTASY
Our planet already has 'nano-scale' machines which self replicate. Bacteria have been breaking down complex molecules in order to exponentially self replicate for, well, about as long as life has existed on this planet. What has stopped a single celled organism turning everything into 'grey goo' already?
I expect it something to do with the amount of energy required to do the job. Although there's a lot of energy around, it's distribution is fairly sparse. Evolution has already made some pretty damn good systems for capturing, storing and using stored energy. Unless nanobots happen to be an order of magnitude more efficient than any possible thing evolution has ever produced, I doubt that it would be possible to achieve any high-impact 'grey goo' scenario.
Well, first off, that's the 'fat fingers' nanotech problem. The tool on a nanobot would have to be made out of atoms, making it tough to manipulate things on a sub-atomic level. Secondly, you're missing the point. The grey goo isn't waste, it's the nanomachines themselves. If they replicate exponentially without end, you get this flood of lil' bots consuming everything, and the grey goo eats the Earth. You can see how that'd be a problem.
I'm surprised CleverNickName hasn't chimed in, he being our resident expert on runaway nanites. :)
I am Sartre of the Borg. Existence is futile.
However, the totality of life in its present form is actually quite vulnerable to being taken over by a distinctly different and new form of life (in fact this already happened once, to a lesser degree, with photosynthesis). The reason is that, although the current totality of life appears incredibly diverse in one sense, at the most fundamental level there is an extraordinary unity. This unity is found in the method by which the principle components of all living organisms are assembled: the linkage of amino acids on the ribosome as directed by DNA sequence.
This unity makes us (and ALL other extant life) vulnerable to outcompetition by a new type of assembly system. But if such a system emerges, it will NOT resemble the industrial kinds of nanoassemblers proposed by Drexler et. al. Instead, this kind of system would have the flexibility and compositional variability of existing living chemical systems, which would enable it to evolve through mutation and mechanisms of selection.
Second, such a system would have machines capable of genetically-directed molecular assembly, but the components of such a system would not be limited to existing biological building blocks such as amino acids, nucleic acids, carbohydrates and lipids. Indeed, the advantages of a wider material repertoire have been pointed by Drexler.
Of course, a new kind of self-replicating system such as this would have to be initially created by pre-existing life (presumably us), but since it is evolvable, its subsequent nature could easily grow out of our control.
Now, to the final question of whether a new self-replicating system could outcompete ALL existing life. I assert that this is unlikely, but for a very different reason than that given by Drexler or others. The reason is NOT because it would be limited by energy utilization, or because that current life forms are already optimally evolved in the use of energy and materials.
Current living organisms do NOT come close to achieving the theoretical optimums of efficiency. This is only achieveable by the industrial kinds of nanomachines mentioned above, which are not a threat because of their brittle and specialized nature. In addition, the criteria for what is optimal depends on the conditions of the local environment, so that control of the nature of the local environment is a critical factor in determining who can best survive in that environment.
The real reason that the threat is limited is that any self-replicating system, no matter how optimized at the molecular level, would also need to compete for resources and control of the environment at the macroscopic scale. To compete at the macroscopic scale requires macroscopic sensor and effectors, and some kind of control system to integrate them. That is, any new form of life that hopes to take over will have to acquire something akin to a macroscopic nervous system.
While such a scenario is certainly possible, this is a whole new requirement that must be met, and I don't believe that it has been sufficiently addressed when considering the likelihood of the 'grey goo' scenario.
mhack
Building a better ribosome since 1997
This has gone way off topic, but if your parents set up a voting system where they said they would do whatever the family voted, but then they made sure that they outvoted you 2-to-1 on anything that you cared about, would you bother adhearing to (and therefore validating) your parents silly system?
Prince Charles is next in line to the British throne and more than likely will be the next King of England. The British Monarch has three essential rights:
Granted, Charles is not yet king, but his contribution on this issue falls more or less within his future remit (and would indicate the advice he would offer to the British Prime Minister of the day)
Whether one wants to have a (future) monarch around to give such advice is another question entirely.
(Obviously, they don't care because his system still leaves them the 1% or whatever percent advantage on every bet, so they'll be fine)
When he tried to talk me into playing the system, I explained that I don't gamble for the odds and winning. I can't, since I know the odds are against me. There's no joy in it that way. However, if I sit, visit with people, make sure I get the maximum number of free drinks and other comps, and keep my burn rate down below a bearable level, I can have fun. Trying to gamble based on a system would take enough concentration that it'd lose me every one of those advantages, so I don't hack gambling. In fact, what I really appear to myself to be hacking is the chance to practice my social skills. Some of us nerds do need practice there, after all.
A few days later, he brought the subject up again. Two additional sources had taught him about system that depended on a limited level of multiple-deck card counting. Now, this is a system that works. It gets you past the 1% house advantage, and if you're good at it you'll probably get banned from casinos that catch you at it.
At that point, I realized this guy was hacking blackjack. He was simply applying hacker principles to gain maximum advantage in a situation. It wasn't about any deeper obsession or nutjob personality quirk... it's just something every hacker does. In fact, every hacker I've ever met does this. One saves a few cents a day by bringing his own soda to work rather than use the vending machine. He'd make a year's worth of those savings up by working another 10 minutes. Go figure. Another spends untold hours cracking DirectTV smartcards, but then scrupulously guards the info so DirecTV isn't harmed beyond his own single larceny. Again, his hourly rate makes this time worth about 10x the cost of just buying the services. It's the challenge, not the money. Another optimizes driving routes until he's got the fastest routes home at any time of day... oh, wait... that's me.
So what that Bill Joy optimized his video buying. It isn't necessarily obsessive. He probably JUST GOT THE IDEA and followed thru out of curiosity.
Saying Bill's a nutcase for this and that it somehow invalidates his opinion on the risks of nanotech is as wrong as somehow coming to a conclusion about Richard Stallman's politics based on the fact that he has some ragged personal hygiene issues. They're so unrelated that you're a nutcase for even thinking they're proof of anything.
Not to be unfair to your well-taken larger point, but your premise is only true in theory.
Exceptions to the military following the orders of politicians come in various ways, from self-protection to obstinance. Let's take just one. Sometimes orders are nebulous or ambivalent. Sometimes military engagements are ill-defined. And sometimes deniability goes all the way to the top. Case in point: Tiger Force in Vietnam.
As the Toledo Blade's Pulitzer-winning investigative series established last year, Tiger Force was a law unto itself. Ostensibly performing recon, the truth was much more complex and sinister. In fact, the squad was just raping and murdering whomever they pleased, as surviving members told the Blade's reporters. Nobody specifically ordered them to do what they did. Nor--and this is the key point--did anyone tell them not to do it. (Note for the conspiracy-minded: the Blade is far from being a leftwing publication. It's a family-owned daily newspaper--one of the last--serving steak-and-potatoes Ohio. It doesn't get much more staid than this in journalism.)
Fast-forward to this year's atrocities in Abu Ghraib. The soldiers say they were told to commit torture. Their commanders deny it. The politicians deny it. The truth is probably somewhere in between. We only need look at the souvenir photos of US soldiers committing evil to know it didn't take a politician anywhere to tell them to enjoy it.
We cannot excuse military malfeasance and free-lancing. The answer is oversight, constant and vigilant, and punishment for abuses. And we must be very cautious about what technologies that barely-governable institution is allowed to play with.