Slashdot Mirror
Guidelines For Nanotech Safety
aibrahim writes "The Foresight Institute has released its guidelines on molecular nanotechnology. Background information on the dangers from Engines of Creation in the chapters Engines of Destruction and Strategies of Survival. The document describes how to deal with the dangers of the coming nanotech revolution. Among the reccomendations: making the nanodevices dependent on external factors, such as artificial "vitamins" and industry self-regulation. The guidelines were cosponsored with the Institute of Molecular Manufacturing. So, is it enough ? Is it too much ? What measures should be taken to secure our safety during the nanotech revolution?" The Foresight Institute sounds like hubris, but it's got a masthead that fairly drips with smart people, like Stewart Brand and Marvin Minsky. Remind anyone of Asimov's Three Laws of Robotics?
Yes, there may be Hubris at Foresight, but the Masthead Name you need to mention is Eric Drexler, author of "Engines Of Creation", isbn://0385199732 , the 1986 seminal book in the field, and the main promoter of nanotech for the last N years.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
I'm sure you had an excellent reason for not putting spoiler warnings in that posting. I can't wait to hear what it was.
The fact that the book is 15 years old is a pretty good reason.
--
That sudden eruption of madness from Bill Joy had a very obvious cause: he was sitting right next to John Searle! The madness of illogic is highly contagious.
(Ray Kurzweil was there as well, but his mind is at least 20 years ahead of those of the other two, so their Neanderthal pack rantings were just "Ughh, Ughh, Grunt!" to him.)
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
Oh, I see. Yes, I suppose you're right, if it's that old I'm sure everyone on Earth has read it by now.
A spoiler isn't necessary in perpetuity. I've never seen any credible guide to nettiquette that required a spoiler for any revelation of any plot in any story, no matter how old.
Do I have to put a spoiler to say Romeo and Juliet die? Or even to say that Tony dies and Maria lives?
No. That would just be silly.
It's been 15 years. If you didn't read it yet, nobody cares.
--
Of course, the dangers of this are obvious. You might mistakenly apply punk nanobots to your hair, and end up walking around with a blue mohawk instead of the yuppie look you had in mind. Or if you put punk nanobots on one side of your head, and hippie nanobots on the other, they might start a war. The subsequent arms race would presumably end when one side or the other (probably the punks, being more violent and less passive) invented nano-nukes and blew your head to nanosmithereens.
...before we know what the reproduction mechanisms are.
The solutions will have to depend on the actual machines we make. We're not 100% clear on what we can make yet, so the feasability of (and need for) any safety system can't be established.
It might not even be a problem. The energy requirements of reproduction are likely to be high, and I imagine that it would take a lot of hard design work to make ones that can reproduce without perfectly controlled conditions and pure chemical feedstocks. Building the gray goo seed might very well be one of the hardest things to do with nanotechnology.
I personally think that gray goo breakouts are likely to be as annoying as mould and computer viruses are today.
"Damn it, Tim, the little buggers have gotten into the acetone feedstock again! We'll have to plas-burn the whole batch."
"Okay, Harry, but we should get a sample. Some idiot kid probably did it on purpose, and they usually don't know enough to cover their tracks."
Nanotech safety:
No science fair for Wesley
He'll destroy the ship
and what about industry self-regulation
isn't that the same thing that was supposed to take care of our privacy ? well will they learn that there's no such thing as "industry self-regulation" ?
Allmost true. Much less the nature of humanity than the nature of things.
Things don't change to acomindate the greater good. A tree remains a tree and it dose far more than is nessisary. It cares not if it cuts off the sunlight to other trees it comes first.
Dirt is dirt it dose not yeald to more useful matereals. It may be made into other things by plants etc but in the end it all returns to dirt.
Human nature is not really to be lax or selfish but to be socal and coprative. We loath those who act out of pure selfiishness or put others in harms way for self.
But as lacking in human nature as it is. As much as we aspire to be great things we do not overcome the simple fact that we operate as singler as self and must defend self and function as self.
It is the nature of this singlaity that produces a lack of responsabilty. No matter how much we are a group we will never overcome our singularity and will allways function at what is best for self.
Anyone who thinks we can be otherwise with out being omnipotent is fooling themselfs. Anyone who thinks an omnipotent man kind could use nanites are being silly.
I don't actually exist.
This reminds me of Jurassic Park.
Remember, the dinosaurs were dependent on lysine, so they couldn't leave the park. However, they just found some chicken in the surrounding area, and had lysine feast!
It almost seems that crypto is the only way to ensure they don't go haywire; you could have nanotech-antibodies to go around checking the MD5SUM of a characteristic of the nano robots.
It's an interesting idea, and certainly one that must be addressed.
|/usr/games/fortune
Sure, it limits a lot of the practical use of nanotechs, but since this is a new technology proceed carefully. Give them 20 years testing and using nanotechs in inert gas before you think about deploying them in environments containing oxygene, that way they have real world tests of how well nanotech's work and how likely they are to run away.
It seems like a good compromise to me.....
When I was able to do my own spam-armoring, you got a chance to email me. Now you can only hope I see your reply.
Remember that in Asimov's universe, a bug in the implementation of single robot's Laws resulted in him slowly rendering the entire Earth unfit for human habitation because he thought it was "for their own good."
The fact that he was right in the story matters not a whit in this cautionary tale.
The same care should be exercised in working with nanotechnolgy as with any other potentially dangerous technology.
And, as with every other potentially dangerous technology, that shouldn't prevent us from working with it.
--
Nanotech is an incredible thing. If you read Age of Spiritual Machines, by Kurzwiel, you'll get an awesome insight into how nano will change our world.
The real problem is with nano-terrorism. Think about it: Will governments *not* make self-replicating (non-restricted-growth) nano-bots, when terrorists will likely be able to for near no-cost?
And... if you trust governments to handle this tech, you should better trust the scientists researching it in the more open public sector.
With nanotech, the human experience from work, health, death, knowledge... can all change for the better. (not just from Kurzwiel's influence).
With terrorists & a trend of open technology, I can see that nanotech will never be able to be controlled like the US has done such a secure job with, say drugs. Terrorists will have access to nanotech, and the question is will they use it, and will the masses have any defence from it.
Now: back to science. While it will be important to wire in some of the " precautions " from the guidelines published, they are mostly obvious, and most likely useless for any real university lab with actual scientists (ie: not just a bunch of techies reading slashdot...)
All that said - even with infinite computation power, the largest problem in creating useful nanobots will be programming them to do anything *useful*.
Assuming we do indeed get nanobots, they must have some sort of way to kill them that cannot be altered without there destruction. If a nanotech does ever gain sentience that could be very bad unless its programmed not to do harm to people. The atificial 'vitamins' is a good idea, provided that you make it impossible for the nanobots to reproduce it, or supplement it with something else. Anyone remember that one Outer Limits episode where the guy gets nanobots to kill his cancer? and they end up evolving him, the Scientist guy tries to kill them with a kill switch, but they modified themselves so it didn't work. Then the guy tried jamming their radio signals so they couldn't communicate, they switched to chemical communication in the body. A general rule of thumb "Never give anything a brain, that you wouldn't want an enemy to have". I figure if we do create robots that can think indepentedly, we might have a little problem as they could become our masters. Or, another example is that Star Trek voyager episode with the robots designed to fight the other robots. They killed their masters because of self preservation, and when they tried to replicate themselves they failed because everything was the same, except for the power supplies which were unique to each robot.
--
Insert Witty Sig Here
I read their rules. Too complicated. This is better:
The goo shall not be gray. "Goo of colour" is acceptable. Red, orange, purple... but not gray.
No gray goo. Problem solved.
Back at the dawn of Modern Sifi Isac Assimov wrote several stories in which he mentions the 3 laws of Robotics. These were rools hardwierd deap into every subsystem of "inteligent" robuts that govorn it's actions even in the most dier sercomstances. This set of guidlines looks like an attempt to impliment the same concept for a difernt kind of technology that once build won't be easily controled or halted ( the way a pulled key can halt a car ). BTW: Those infamus laws ran something like this. 1. Do not harm a human or throgh inaction alow one to be harmed. 2. Obay orders from a human. 3. Prevent your own destruction. Nanorobots *need* this kind of guidline more than other stuff because once they are deployed there realy is no way to disable them, except maybe by using an EMP blast. That however is not the Matrix conection. The Matrix comes into the picture when you think of why the machines wold keep humans around for power suply when they have Fusion ? Could it be because somewhere deap inside they can't bring themselvs to render the spiceis extinct ?
--= Isn't it surprising how badly I spell ?
The agro genmodders promised that their stuff would never get into the wild, either. Ooopsies! And then they came up with the brill idea of a killer gene to prevent self-replication, er, well, that was so the farmers couldn't use the seeds without paying first. Hmmm. What happens when *that* one gets into the wrong gene code? And what happens when the standby security nanopups run wild? Let's just hope they've been paper-trained. We'll make 'em all snap to at the first crinkle of a restraining order.
We don't attempt to prevent murder through a system of self-regulation, we make it illegal. Knowing that it is always easier for an institution to act unethically than for an individual, why place fewer restrictions on institutions?
Proud member of the Weirdo-American community.
". . .Any replication information should be error free. . ." hmmmmmmmm.
Every living organism is dependent on resources that can only be found in a specific region. Organisms living today are dependent on resources, such as oxygen, that weren't available in earth's past. I hope no nanotech researcher honestly believes that replication information can be transmitted without error. We have to assume that there will be error; with that assumption any self replication will result in evolution. If errors are one in a million, then there will be a mutation a thousand times every time a billion nonobots are produced. If all but one in a million are deleterious, it will still only take a trillion nanobots before one has a happy (for it) accident.
I never agreed with my Biology texts, when they made the arbitrary decision to consider only cellular organisms as living. When we create a self replicating system, the best model we have to predict how they will behave is organic life. And the life form most simular to the proposed nanobots, is the virus. Nonobots and virii will share a number of common attributes:
1. Biologists don't consider them to be living.
2. They are precocious molecules.
3. They can reproduce.
Well, I still think we should take the risks, but this document is a little to optimistic in my opinion. We need stiffer controls, and I DON'T like the idea of purposefully making nanobots evolve, even in "controlled" conditions.
People theoretically see the need for lots of nice protections. Then they go ahead and cut corners unless someone has been burned and the memory is fresh.
I cannot think of any area of technology from automobile design to nuclear power plants to office suites where this principle of human nature has not been operational. I can personally list examples from NASA to genetics research to the SNMP spec. (It was nicknamed Security - Not My Problem for a reason!)
IMNSHO anyone who thinks that nano has the potential to be any different is just kidding themselves about human nature...
Cheers,
Ben
My usual seat in the cluetrain is at A HREF="http://pub4.ezboard.com/biwethey.ht
Actually it points out a different problem, the "leak" was initiated on purpose by a person who wanted them to be leaked and had designed them specifically to cause the damage they caused.
Thus it wasn't the nanites themselves which were the problem, but the madman who used them as assassins.
Also they did have a kill switch (Hard UV) the people affected just didn't have access to the "kill switch" when they were being eaten. Later in the book they do go out and clean up the nano-mess and destroy the responsible nanites.
-------- This space intentionally left blank --------
It looks like everyone has already brought up the point that the danger in putting a "self-destruct" mechanism in a nanite. With millions or billions of nanites, even if the odds of one of them surviving that self-destruction are one in a million, those odds are too high. And if that nanite is designed to construct other nanites (or, worst case, copies of itself) then you have a problem on your hands.
If nanotechnology ever reaches the total control of matter, self-replicating machine, Diamond Age "Seed" level (I don't have enough information to argue either way, but it seems to me that it'd be easier to create macroscopic Von Neumann machines than microscopic ones, and we haven't even done that yet) we're going to need more protection than a self destruct mechanism.
What I'd like to see, in a world swarming with potential nanotech viruses, is an analogous nanotech immune system to take care of them, nanites which can be set to recognize and rip apart other nanites which meet certain parameters. Got a rogue oil-spill cleaning nanite ripping up asphalt in San Francisco? Get the standby security nanites in Oakland to kill it.
There was an interview with a somewhat apocalyptic tech giant (a veep at Sun? I forget) who believed that the ever increasing technological power available to humanity (nanotech, biotech, and AI being three examples I remember) would cause the world to be ripped apart by terrorism in the coming century. He likened it to an airplane in which every passenger had a "Crash" button in front of their seat, and only one psycho was necessary to bring everyone down with him.
I don't think it will be that way. With nanotechnology specifically, if our available defenses are kept up to the level that our potential offenses would require, then having a small set of nanites go rogue wouldn't be a concern; they would be overwhelmed by their surroundings. Going back to that analogy, if everybody had a "Crash" button in front of their airplane seat, but the plane was guaranteed to survive unless 50% of the passengers voted to crash, that would be the safest flight in history.
The vitamin dependent nanobot is a good idea, but what about the flipside? What if, for example, Kool Aid Corp. decides to put nanobots in their packages. The bots are dependent on Kool Aid and if they don't get any, the person gets headaches/feels bad/whatever until they drink more Kool Aid. I'm glad that the Foresight Institute is looking ahead into regulating these things, because being addicted to something truly horrible, like health food, would be very, very bad =).
-Antipop