Von Neumann machines

[Almo]

Von Neumann machines are machines that can make replicas of themselves. They're a favorite in Sci Fi stories where some amuk VNMs are converting everything in their path to more of themselves.

It just occurred to me last night as I was reading Stephen Baxter's Manifold Time that humans are in fact Von Neumann machines.

You might say that a metal VNM is more frightening as it is inhuman, but... humans are VNMs anyway!

So we're just VNMs that do more than the business of replicating (Art etc).

Once we get space travel working resonably, the universe had better watch out!

[Cuddles]

Technically, yes we are. However, the distinguishing feature is generally self-awareness. The term Von Neumann Machine is generally applied to dumb replicators. Sentience changes things, qualitatively if not actually quantitatively. VNMs are popular in sci-fi for the same reasons zombies are - it's the blind destruction with no apparent purpose. Aliens may be alien, but it can be assumed that they at least have motivations even if we don't understand them. Zombies and VNMs are different because there is nothing else. They reproduce because they reproduce, and that's all there is to it.

[Almo]

Interesting point. Which is exactly why I posted this though here in the first place. :D

Does sentience make that big a difference? We think of it as motivations, but how is that different from the program running a non-sentient VNM?

[Soapy Sam]

Zombies and VNMs are different because there is nothing else. They reproduce because they reproduce, and that's all there is to it.

This is as succinct a description of evolution as I've seen.

[Cuddles]

Does sentience make that big a difference? We think of it as motivations, but how is that different from the program running a non-sentient VNM?

Well, it's probably better to ask that in the philosphy forum really. The important thing is that people think there's a difference, and therefore a non-sentient VNM is viewed in a different way from actual AI, or just plain I for that matter.

This is as succinct a description of evolution as I've seen.

Yes, but we're special dammit!:)

[Skwinty]

Didn't Frank Tipler postulate that this was the way for mankind to expand into the far realms of the universe. How in the name of anything does mankind transpose his sentience into these mechanical replicators?:confused:
Or am I confusing this with something/someone else?

[blobru]

Zombies and VNMs are different because there is nothing else. They reproduce because they reproduce, and that's all there is to it.

This is as succinct a description of evolution as I've seen.


For life, yes. VNM's aren't evolving of course, they're just copying. Zombies might evolve if the pathogen which creates zombies (saliva-based one assumes as it's transmitted by bites) mutates to involve different vectors and/or symptoms.

[JoeyDonuts]

The Forge of God is a pretty good read that deals with this subject.

[Rasmus]

http://xkcd.com/387/

[BigAl]

I don't know what else VN did, but if you are referring to the work he is most closely associated with in the area of computers, "Von Neumann architecture ". you are wrong. Sorry.

http://en.wikipedia.org/wiki/Von_Neumann_architecture

[INRM]

Would you really want out of control self-replicating machines

[Skwinty]

A von Neumann machine (http://www.daviddarling.info/encyclopedia/V/von_Neumann_machine.html) able to move over interstellar (http://www.daviddarling.info/encyclopedia/I/isdist.html) or interplanetary distances and to utilize local materials to build new copies of itself. Von Neumann probes are named after the Hungarian-born American mathematician John von Neumann (http://www.daviddarling.info/encyclopedia/V/von_Neumann.html) who, among many other achievements, was the first to develop a mathematical theory of machines that can make exact copies of themselves.

[Soapy Sam]

"Johnny" von Neumann did many things.
The wikipedia article here http://en.wikipedia.org/wiki/John_von_Neumann is far from exhaustive.

ETA- Incidentally, there's no reason the copying mechanism of a vNM should be foolproof. It might even be intelligently designed not to be. Any variations will be selected by circumstance. If they are heritable...

[Elvis666]

I don't know what else VN did, but if you are referring to the work he is most closely associated with in the area of computers, "Von Neumann architecture ". you are wrong. Sorry.

http://en.wikipedia.org/wiki/Von_Neumann_architecture

Nope, not wrong.
http://en.wikipedia.org/wiki/Self-replicating_machine

Edit: Darn, three of us in under two minutes.

[BigAl]

Nope, not wrong.
http://en.wikipedia.org/wiki/Self-replicating_machine

Edit: Darn, three of us in under two minutes.
I stand corrected. Thank you.

[Third Eye Open]

For life, yes. VNM's aren't evolving of course, they're just copying. Zombies might evolve if the pathogen which creates zombies (saliva-based one assumes as it's transmitted by bites) mutates to involve different vectors and/or symptoms.

Why wouldn't the VNM's evolve? There is no guarantee that the copies are going to be exactly perfect every time. Maybe one will have a glitch in its program that causes it to make two copies of it's self instead of one? Or maybe a glitch that tells it that its OK to use other VNM's for parts in making its copies?

Also, if you like Stephen Baxter, read his Evolution.

[Almo]

Would you really want out of control self-replicating machines

That's actually one of my points about humanity. If we get past the solar system, you will have out of control self-replicating machines. :)

Why wouldn't the VNM's evolve? There is no guarantee that the copies are going to be exactly perfect every time. Maybe one will have a glitch in its program that causes it to make two copies of it's self instead of one? Or maybe a glitch that tells it that its OK to use other VNM's for parts in making its copies?

Also, if you like Stephen Baxter, read his Evolution.

Good point about the glitches. And thanks for the tip on the SB book. If I dig this first Manifold book, I'll do the others then get onto Evolution.

[blobru]

... ETA- Incidentally, there's no reason the copying mechanism of a vNM should be foolproof. It might even be intelligently designed not to be. Any variations will be selected by circumstance. If they are heritable...

Why wouldn't the VNM's evolve? There is no guarantee that the copies are going to be exactly perfect every time. Maybe one will have a glitch in its program that causes it to make two copies of it's self instead of one? Or maybe a glitch that tells it that its OK to use other VNM's for parts in making its copies? ...


Well ok, I'm used to thinking of vNM's in the abstract, like Turing machines. As SS mentioned, a lot would depend on whether the glitches are heritable, or meant to be. If the designers are intent on exact copies, they'd have some correction routines built-in to make "mutation" virtually impossible (bit-check against dozen nearest neighboring vNM's, e.g.); if they want the machines to evolve, however, maybe allow glitches within parameters -- the glitch code must still be executable, etc. -- to simulate biological evolution. But even for the former, unforeseen random glitches, which are also beneficial, although extremely unlikely are always possible, so yeah you're right, a real vNM could evolve.

[ElMondoHummus]

Von Neumann machines are machines that can make replicas of themselves. They're a favorite in Sci Fi stories where some amuk VNMs are converting everything in their path to more of themselves.

It just occurred to me last night as I was reading Stephen Baxter's Manifold Time that humans are in fact Von Neumann machines.

You might say that a metal VNM is more frightening as it is inhuman, but... humans are VNMs anyway!

So we're just VNMs that do more than the business of replicating (Art etc).

Once we get space travel working resonably, the universe had better watch out!

Wait a minute... maybe I'm imposing too much of my own personal sense of the definition on this, but I thought Von Neumann machines were supposed to make exact copies of themselves. Humans don't really do that; what we do is reproduce (offspring are not exact genetic templates, but rather a mix of traits from both parental lines), and do so imperfectly (i.e. genetic drift leading to trait evolution, etc.). I thought that a more proper application of the Von Neumann concept would be to cloning, not reproduction. Am I misinterpreting things here?

[Ashles]

Wait a minute... maybe I'm imposing too much of my own personal sense of the definition on this, but I thought Von Neumann machines were supposed to make exact copies of themselves. Humans don't really do that; what we do is reproduce (offspring are not exact genetic templates, but rather a mix of traits from both parental lines), and do so imperfectly (i.e. genetic drift leading to trait evolution, etc.). I thought that a more proper application of the Von Neumann concept would be to cloning, not reproduction. Am I misinterpreting things here?
I second that question.

[Almo]

Wait a minute... maybe I'm imposing too much of my own personal sense of the definition on this, but I thought Von Neumann machines were supposed to make exact copies of themselves. Humans don't really do that; what we do is reproduce (offspring are not exact genetic templates, but rather a mix of traits from both parental lines), and do so imperfectly (i.e. genetic drift leading to trait evolution, etc.). I thought that a more proper application of the Von Neumann concept would be to cloning, not reproduction. Am I misinterpreting things here?

In principle I think you're right. But in terms of a replicating thing sucking up all available resources and energy in the universe, we fit the bill should we get outside the solar system and develop means of harvesting what's out there.

[Ziggurat]

Technically, yes we are. However, the distinguishing feature is generally self-awareness. The term Von Neumann Machine is generally applied to dumb replicators.

I disagree. Bacteria are dumb replicators, but they don't really match the phrase. I think more important than self awareness is actually mortality: machines are sort of immortal (they don't have to die on their own, though they can be killed), but bacteria are not.

[geni]

ETA- Incidentally, there's no reason the copying mechanism of a vNM should be foolproof. It might even be intelligently designed not to be. Any variations will be selected by circumstance. If they are heritable...

Any halfway sane species building one would put a lot of effort into preventing that happening since sooner or later you would either run into a varation that reproduced too fast or one that decided to use you as raw materials.

[ImaginalDisc]

I disagree. Bacteria are dumb replicators, but they don't really match the phrase. I think more important than self awareness is actually mortality: machines are sort of immortal (they don't have to die on their own, though they can be killed), but bacteria are not.

Um, no they don't. A bacteria will either die of disease, damage, starvation, or what have you, or it will reproduce which amounts to splitting in half. The difference isn't apparent to me.

[GodMark2]

Wait a minute... maybe I'm imposing too much of my own personal sense of the definition on this, but I thought Von Neumann machines were supposed to make exact copies of themselves. Humans don't really do that; what we do is reproduce (offspring are not exact genetic templates, but rather a mix of traits from both parental lines), and do so imperfectly (i.e. genetic drift leading to trait evolution, etc.). I thought that a more proper application of the Von Neumann concept would be to cloning, not reproduction. Am I misinterpreting things here?

In theory, yes, vNMs are perfect replicators. But as soon as you get one into the real world, some bloke opens his hand and lets a delicate butterfly's wings (http://xkcd.com/378/) flap once. The disturbance ripples outward, changing the flow of the eddy currents in the upper atmosphere. These cause momentary pockets of higher-pressure air to form, which acts as lenses that deflect incoming cosmic rays, causing them to strike the drive platter and flip a bit.

Now you no longer have that 'perfect' vNM, and it may have a survival advantage over the rest, and that leads to all sorts of interesting possibilities.

[Cuddles]

Wait a minute... maybe I'm imposing too much of my own personal sense of the definition on this, but I thought Von Neumann machines were supposed to make exact copies of themselves. Humans don't really do that; what we do is reproduce (offspring are not exact genetic templates, but rather a mix of traits from both parental lines), and do so imperfectly (i.e. genetic drift leading to trait evolution, etc.). I thought that a more proper application of the Von Neumann concept would be to cloning, not reproduction. Am I misinterpreting things here?

I don't think you're misinterpreting, you're just taking too narrow of our reproduction by focussing on humans and not just life in general. Sexual reproduction is relatively late to the table, the original replicators we evolved from did replicate by cloning. And of course, many things still do. If we allow VNMs to be imperfect replicators, then they will be subject to evolution just as any other imperfect replicator would. Which could in fact lead to sexually reproducing VNMs. At which point I'm going to have to invoke rule 34 on VNM porn.

For life, yes. VNM's aren't evolving of course, they're just copying. Zombies might evolve if the pathogen which creates zombies (saliva-based one assumes as it's transmitted by bites) mutates to involve different vectors and/or symptoms.

Except that everyone knows zombies are actually caused by radiation from Venus, or possibly sunspots.:)

[blobru]

... Except that everyone knows zombies are actually caused by radiation from Venus, or possibly sunspots.:)


Plan 9 from outer space, originally... unless it's evolved into plan 10. :eusa_eh: