What would happen if Grey Goo (self-replicating nanobots) got out of control?

[KAL 9000]

Just wondering.

[legoclone09]

Really, really bad things. They would eat everything on Earth until it was only Grey Goo. Or run out of IP addresses and stop.

[akron]

9 minutes ago, legoclone09 said:

Really, really bad things. They would eat everything on Earth until it was only Grey Goo. Or run out of IP addresses and stop.

Our best weapon is IPv4

[peadar1987]

Not much. Grey goo is extremely thermodynamically unfavourable, there's basically no small enough power source for it.

[Buster Charlie]

5 minutes ago, peadar1987 said:

Not much. Grey goo is extremely thermodynamically unfavourable, there's basically no small enough power source for it.

Not even the power of Love?

[Dman979]

2 minutes ago, Buster Charlie said:

Not even the power of Love?

I hear that it's all you need.

[peadar1987]

3 minutes ago, Buster Charlie said:

Not even the power of Love?

Oh please. Have you ever tried harnessing the power of love? You would need ridiculous quantities of Huey Lewis to even make it remotely possible.

[GeneralVeers]

The grey goo wouldn't get ten feet out of the test tube, because the nanobots would eat each other.

(srsly, why has nobody ever thought of that before?)

[Raven Industries]

I would assume they would take over part of the earth, and then we would do one of 3 things: 

1. Nuke it. Nuclear weapons would be able to destroy all grey goo in the immediate area, and/or fry a bunch of them with the resulting electromagnetic pulse.

2. Destroy them with fire. Lots of it. Napalm seems like it would be pretty effective, since you can deploy it out of range of the goo, and it is self-spreading. In addition, it could be used to impede the path of the goo.

3. Use the kill switch any sane and rational engineer or scientist would include in the design.

[KAL 9000]

16 minutes ago, KerbalSaver said:

I would assume they would take over part of the earth, and then we would do one of 3 things: 

1. Nuke it. Nuclear weapons would be able to destroy all grey goo in the immediate area, and/or fry a bunch of them with the resulting electromagnetic pulse.

2. Destroy them with fire. Lots of it. Napalm seems like it would be pretty effective, since you can deploy it out of range of the goo, and it is self-spreading. In addition, it could be used to impede the path of the goo.

3. Use the kill switch any sane and rational engineer or scientist would include in the design.

Good point.

16 minutes ago, GeneralVeers said:

The grey goo wouldn't get ten feet out of the test tube, because the nanobots would eat each other.

(srsly, why has nobody ever thought of that before?)

I think they might program them not to eat each other?

31 minutes ago, peadar1987 said:

Not much. Grey goo is extremely thermodynamically unfavourable, there's basically no small enough power source for it.

Never thought of that! Hey, you never know, someday we might discover how to harness the zero-point field (not impossible).

[HebaruSan]

24 minutes ago, peadar1987 said:

Not much. Grey goo is extremely thermodynamically unfavourable, there's basically no small enough power source for it.

Are you referring to an argument along the lines of the "Thermodynamics" section here (from a nanotech chemist)?

http://berkeleysciencereview.com/grey-goo-end/

I'm awful at chemistry, but when I read that part, it sounded like he was assuming no external energy inputs. Isn't this graph effectively what plants do? Take minerals from the soil and use solar energy to assemble them into organic compounds?

Quote

For the Grey Goo apocalypse vision to appear, what ever this nanorobot is, it will have to outcompete DNA, which is a tall order.

I found this the real "sleep soundly at night" quote. If gray goo nanobots escape from a lab, they'll get swarmed and dismembered by the run of the mill bacteria that we swim in every day.

[KAL 9000]

5 minutes ago, HebaruSan said:

Are you referring to an argument along the lines of the "Thermodynamics" section here (from a nanotech chemist)?

http://berkeleysciencereview.com/grey-goo-end/

I'm awful at chemistry, but when I read that part, it sounded like he was assuming no external energy inputs. Isn't this graph effectively what plants do? Take minerals from the soil and use solar energy to assemble them into organic compounds?

I found this the real "sleep soundly at night" quote. If gray goo nanobots escape from a lab, they'll get swarmed and dismembered by the run of the mill bacteria that we swim in every day.

Go bacteria! Saving our lives from Grey Goo!

[GeneralVeers]

32 minutes ago, KAL 9000 said:

I think they might program them not to eat each other?

And there will inevitably be a programming glitch somewhere. The thing about non-sexual reproduction is that each duplicate is an exact duplicate, including all the errors. One glitch in one nanobot. That's all it would take. That one glitchy nanobot will look around for the ideal food source, and see a whole lot of exactly the right materials it needs to replicate itself. And it will proceed to eat its neighbors, replicate itself, thereby replicating the glitch, and very quickly the entire mass of grey goo will self-destruct.

[HebaruSan]

14 minutes ago, GeneralVeers said:

And there will inevitably be a programming glitch somewhere. The thing about non-sexual reproduction is that each duplicate is an exact duplicate, including all the errors. One glitch in one nanobot. That's all it would take. That one glitchy nanobot will look around for the ideal food source, and see a whole lot of exactly the right materials it needs to replicate itself. And it will proceed to eat its neighbors, replicate itself, thereby replicating the glitch, and very quickly the entire mass of grey goo will self-destruct.

Hmm, organic life eats itself and still survives, which suggests a few ways around this problem.

I wonder if the nanobots' general danger-avoidance mechanisms might serve as a defense against rogues. E.g., if some nearby food looks like a dead nanobot, then flee. At that point the two populations might develop along familiar predator/prey lines.

[Pixel of Life]

They would form a new race called Asuran and start destroying human worlds in an attempt to starve the Wraith to death.

[Darnok]

1 hour ago, GeneralVeers said:

The grey goo wouldn't get ten feet out of the test tube, because the nanobots would eat each other.

(srsly, why has nobody ever thought of that before?)

Even simple bacteria doesn't eat each other why would nanobots do that? It would be useful to "eat" broken units, but other than that they should process raw materials or anything except their own "species".

[fredinno]

1 hour ago, HebaruSan said:

Hmm, organic life eats itself and still survives, which suggests a few ways around this problem.

I wonder if the nanobots' general danger-avoidance mechanisms might serve as a defense against rogues. E.g., if some nearby food looks like a dead nanobot, then flee. At that point the two populations might develop along familiar predator/prey lines.

Then the nanobots should also be programmed to destroy cannibalistic colleagues.

[HebaruSan]

10 minutes ago, fredinno said:

Then the nanobots should also be programmed to destroy cannibalistic colleagues.

Well, that assumes they're programmed with this scenario in mind, which would open up all sorts of options. I was trying to build off of behavioral principles that you'd have to include whether or not you anticipated that they might go cannibal.

[kunok]

14 minutes ago, fredinno said:

Then the nanobots should also be programmed to destroy cannibalistic colleagues.

That nano-bots will be cannibalistic too how can you differentiate between a noncannibal robot eating a cannibal and a cannibal eating other cannibal?

And a noncannibal eating a noncannibal eating a cannibal? And a noncannibal eating a noncannibal eating a noncannibal eating a cannibal?  And a noncannibal eating a noncannibal eating a noncannibal eating a noncannibal eating a cannibal?  And a noncannibal eating a noncannibal eating a noncannibal eating a noncannibal eating a noncannibal eating a cannibal?

Edited March 21, 2016 by kunok
typos

[cantab]

2 hours ago, GeneralVeers said:

And there will inevitably be a programming glitch somewhere. The thing about non-sexual reproduction is that each duplicate is an exact duplicate, including all the errors. One glitch in one nanobot. That's all it would take. That one glitchy nanobot will look around for the ideal food source, and see a whole lot of exactly the right materials it needs to replicate itself. And it will proceed to eat its neighbors, replicate itself, thereby replicating the glitch, and very quickly the entire mass of grey goo will self-destruct.

My view is that a "kill switch" won't work for the exact opposite reason - that the nanobot replication inevitably will occasionally make new mutations. For the nanobot to make its killswitch necessarily requires some resources, and so a mutant nanobot that loses the killswitch will have a competitive advantage and reproduce more successfully that the original design. It could be that long before you ever try using the killswitch, the nanobot population already evolved away.

Worse, this argument applies to *any* attempted method of controlling the things.

You also might think that some sort of cryptography could prevent the mutants from operating, but the cryptographic 'lock' will itself evolve away.

[mikegarrison]

I'm not worried. We'll just use our Ice-9 to stop them.

[GeneralVeers]

1 hour ago, Darnok said:

Even simple bacteria doesn't eat each other why would nanobots do that?

I told you--programming glitch.

 

11 minutes ago, cantab said:

My view is that a "kill switch" won't work for the exact opposite reason - that the nanobot replication inevitably will occasionally make new mutations. For the nanobot to make its killswitch necessarily requires some resources

Actually, in this particular case, the figurative kill switch saves resources. As KAL 9000 said: why not program the nanobots not to eat other nanobots? That programming (the "don't eat other nanobots" command) takes extra code and memory and whatnot--it's the don't kill switch that costs extra.

And, of course, any nanobot that doesn't have that extra code (or suffers a software glitch......) will have a competitive advantage. Actually two competitive advantages: less energy to replicate, because it doesn't have to replicate the extra software, and a much larger food supply--the other nanobots.

[NuclearNut]

It depends, but assuming it is the classic, nonobot sized variety, I would give them a few weeks to go extinct.

Why you may ask, well the answer is quite simple: in order to be that small they would likely end up being quite fragile, and if they are to replicate quickly, they would need to be made of hydrocarbons.  In other words, they would be a new type of bacteria.  This means that they would suffer the same problems as other bacteria, possibly more due to human errors that do not become apparent until widespread release, something that evolution has dealt with in actual bacteria.  Problems such as an inability to live in unfavorable environments, consumption by other animals, and so on, would likely exist for them due to their carbon based, small and fragile nature.

And then we get to the whole problem of eating the earth.  You see, the earth is composed of a lot of diverse materials, and nanobots are not going to be able to have a massive memory capacity, all of it would have to be preserved for the data regarding their own replication and self repair.  Thus it is unlikely that they would be able to store the data for building nanobots that can say... be built of iron in exclusivity, or some similar situation.  You also run in to the question of power supply, what can power these millions of nanobots in their quest to destroy the earth?  Solar will quite rapidly become worthless if all goes "well" form the replication standpoint (plus, night and clouds would stop the grey goo death army), nuclear fission is too bulky, same goes for fusion.  Chemical energy is limited in the extreme.  So power is also a huge difficulty.

Then we go on to the whole eating part.  Even if you somehow empowered these death bots with the ability to consume everything, massive memory, and so on, you would have the trouble of them consuming the still living flesh of your enemies.  As you probably know, bacteria is not the fastest at consuming the still living flesh of your enemies or anything else that can have the prefix still living applied.  This is due to the immune system, which again would be hard to completely defend against with this nanobot menace.  So you would have to deal with that problem as well as the problem of other bacteria.  So there is fun to be had there.

 

So in conclusion, if it gets released, it is unlikely to survive that long.

[magnemoe]

9 hours ago, peadar1987 said:

Not much. Grey goo is extremely thermodynamically unfavourable, there's basically no small enough power source for it.

This, in the wild its limited to solar. Chemical or electrical will be to hard to get. 
Worse as its out of control the bots are unlikely to cooperate well so the top layer will not share solar energy with the ones deeper down while the bottom layer will not share raw material. 

[pincushionman]

8 hours ago, Darnok said:

Even simple bacteria doesn't eat each other why would nanobots do that? It would be useful to "eat" broken units, but other than that they should process raw materials or anything except their own "species".

Yes they do. The whole food chain is just an extreme version of "eat whatever is enough like itself to work."

And the energy problem really is super important. Even the author who first introduced the gray goo concept andmitted the whole thing became rather silly once that was pointed out to him. It would be at best (worst?) just another organism in the competing-for-resources pool, and I think we can all agree that "ordinary" life is already very, very good at that game.