von Neumann probes

[Genady]

A former colleague of mine is very enthusiastic about prospects of large scale space exploration based on self-replicating von Neumann probes. I've never heard of them before. From a brief reading of several Internet articles, it seems to me not realistic. Not the self-replicating aspect of it, but what it will be replicating, i.e. the fully automated explorers. Do you know of strong arguments I could use to curb his enthusiasm?

[beecee]

7 hours ago, Genady said:

A former colleague of mine is very enthusiastic about prospects of large scale space exploration based on self-replicating von Neumann probes. I've never heard of them before. From a brief reading of several Internet articles, it seems to me not realistic. Not the self-replicating aspect of it, but what it will be replicating, i.e. the fully automated explorers. Do you know of strong arguments I could use to curb his enthusiasm?

Despite automation and its inevitable improvement, The human need for adventure, exploration and boots on the ground? Because its there?

https://www.rei.com/blog/snowsports/the-nature-fix-why-were-hardwired-for-adventure#:~:text=The thirst for new%2C exciting,are meant to seek adventure.&text=Studies have shown that adventure,emotional resiliency and social bonding.

https://thepsychologist.bps.org.uk/volume-2018/january-2018/adventure-important-part-being-human

https://www.sciencedaily.com/releases/2008/06/080625122945.htm

Edited February 4, 2022 by beecee

[Endy0816]

I look at them as simply an advanced form of automation, so don't see them as being particularly unrealistic.

Should note,  making more of themselves from available materials to fulfill some purpose would probably be a better definition. Exploration may be that purpose, but something more realistic like automated colony building, resource extraction, and megaprojects is more likely IMO.

[Genady]

33 minutes ago, Endy0816 said:

I look at them as simply an advanced form of automation, so don't see them as being particularly unrealistic.

I agree with this. Moreover, I don't see what is so special about being self-replicating. A machine X can be programmed to build a machine Y. This is a robot. As a particular case, it can be that Y=X. This makes it self-replicating. Big deal!

[Endy0816]

2 hours ago, Genady said:

I agree with this. Moreover, I don't see what is so special about being self-replicating. A machine X can be programmed to build a machine Y. This is a robot. As a particular case, it can be that Y=X. This makes it self-replicating. Big deal!

Practically speaking their potential for exponential growth would be incredible.

Concept has been around since the 40's now though. It probably won't seem quite as futuristic as we could just about slap together something.

Edited February 5, 2022 by Endy0816

[Genady]

How / where from do they get materials needed to make another machine?

[Endy0816]

8 hours ago, Genady said:

How / where from do they get materials needed to make another machine?

From anywhere in theory. Probably asteroids and moons more realistically.

I have to admit miniaturized-automated smelting and refining still seems implausible to me, but I also thought the same about 3D printing, artificial diamonds and the like.

 

[Genady]

1 hour ago, Endy0816 said:

From anywhere in theory. Probably asteroids and moons more realistically.

I have to admit miniaturized-automated smelting and refining still seems implausible to me, but I also thought the same about 3D printing, artificial diamonds and the like.

 

Well, this is where "unrealistic" starts IMO. To discover the sources of materials, first. We - not robots - cannot yet figure out water on Mars, after a long time trying, observing, experimenting, sending multiple probes, etc. On / in a next door neighbor. Can we program a machine to do something that we don't know ourselves how to do? I don't think so. That's even before smelting and refining you've mentioned. 

[Endy0816]

1 hour ago, Genady said:

Well, this is where "unrealistic" starts IMO. To discover the sources of materials, first. We - not robots - cannot yet figure out water on Mars, after a long time trying, observing, experimenting, sending multiple probes, etc. On / in a next door neighbor. Can we program a machine to do something that we don't know ourselves how to do? I don't think so. That's even before smelting and refining you've mentioned. 

 

Water is more problematic. It's likely to only exist as sublimation prone ice or brine. We've been playing it safe so far with landing sites and have barely scratched the surface of Mars digging.

They may need some kind of AI along to deal with more complex situations, but I'm not convinced it'll need to be too advanced. We should have regular contact with them anyways and can provide them with directions if needed.

[Genady]

15 minutes ago, Endy0816 said:

 

Water is more problematic. It's likely to only exist as sublimation prone ice or brine. We've been playing it safe so far with landing sites and have barely scratched the surface of Mars digging.

They may need some kind of AI along to deal with more complex situations, but I'm not convinced it'll need to be too advanced. We should have regular contact with them anyways and can provide them with directions if needed.

OK. Thank you.

[Ken Fabian]

It is interesting as a thought experiment but I am doubtful of the actual viability of such "machines".

On 2/5/2022 at 12:24 PM, Genady said:

How / where from do they get materials needed to make another machine?

The obvious answer is asteroids and comets but the practicalities of actually mining and refining them are pretty much entirely hypothetical. Without the active geology including liquid water of a planet (and often biology too) essential materials may not exist as usable ores. Complex machinery tends to require a wide variety of materials and we currently draw on a multitude of specialties to produce them - multiple specialties for just one material. Requiring mining on planets to obtain them greatly increases the difficulties. 

And what kind of energy sources will they use, especially at any great distance from a star? There will be high energy requirements and low grade ores are likely to require ever greater energy inputs to get the required amounts - with potential diminishing returns. I think it possible that energy requirements outside an inner solar system can exceed what is needed for the essential infrastructure, including energy production capability - ie there may be physical limitation (which would also apply to space colonies). Are these machines going to be building and running fusion power plants? Fission would be easier (but not easy) but whilst fissionable materials almost certainly exist there they are going to be at very low concentrations except on geologically active planets.

Seems like these won't be merely machines somehow reproducing themselves - the requirements for interstellar space probes being anything but simple - it will be advanced machine economies made of multiple industries reproducing themselves. I think so unlikely as to be indistinguishable in practice from impossible.

I am not sure what is required can be pared down sufficiently to work in the absence of a large and advanced industrial economy - where essential but difficult to refine and produce materials have a sufficient variety of uses and levels of demand that industries can be commercially viable providing them.

When it comes to manufacturing machines to make machines that can manufacture themselves it seems we might be able to bypass those limitations but I suspect in reality there are physical limitations that cannot be readily overcome.

Edited February 16, 2022 by Ken Fabian

[Genady]

3 hours ago, Ken Fabian said:

It is interesting as a thought experiment but I am doubtful of the actual viability of such "machines".

The obvious answer is asteroids and comets but the practicalities of actually mining and refining them are pretty much entirely hypothetical. Without the active geology including liquid water of a planet (and often biology too) essential materials may not exist as usable ores. Complex machinery tends to require a wide variety of materials and we currently draw on a multitude of specialties to produce them - multiple specialties for just one material. Requiring mining on planets to obtain them greatly increases the difficulties. 

And what kind of energy sources will they use, especially at any great distance from a star? There will be high energy requirements and low grade ores are likely to require ever greater energy inputs to get the required amounts - with potential diminishing returns. I think it possible that energy requirements outside an inner solar system can exceed what is needed for the essential infrastructure, including energy production capability - ie there may be physical limitation (which would also apply to space colonies). Are these machines going to be building and running fusion power plants? Fission would be easier (but not easy) but whilst fissionable materials almost certainly exist there they are going to be at very low concentrations except on geologically active planets.

Seems like these won't be merely machines somehow reproducing themselves - the requirements for interstellar space probes being anything but simple - it will be advanced machine economies made of multiple industries reproducing themselves. I think so unlikely as to be indistinguishable in practice from impossible.

I am not sure what is required can be pared down sufficiently to work in the absence of a large and advanced industrial economy - where essential but difficult to refine and produce materials have a sufficient variety of uses and levels of demand that industries can be commercially viable providing them.

When it comes to manufacturing machines to make machines that can manufacture themselves it seems we might be able to bypass those limitations but I suspect in reality there are physical limitations that cannot be readily overcome.

This list of physical and technical difficulties and potential points of failure can go on and on. It seems that self-replication algorithm is an easy part - it is all the rest that is difficult or impossible. Self-replication doesn't help in solving other issues. 

Edited February 16, 2022 by Genady

[Danijel Gorupec]

I first heard about 'von Neumann' machines when I was a kid. Since then, I occasionally think about it.

As Ken Fabian, I also think that it is easier to build an ecosystem of machines (or, better, a 'super-organism' of machines) than a single self-replicating machine. I also think that it could be easier to accomplish this in nanotechnology.

For the difference, I don't think it is impossible. In fact, I think it is inevitable (if we manage not to destroy ourselves sooner). Once our artificial intelligence machines become sufficiently advanced and self-motivated, it seem logical that they will start experimenting with self replication.

Or, humans themselves might start experimenting with from-scratch designed bio-organisms using already-proven self replicating 'technology'. Maybe it is possible to intelligently design organisms that have properties completely out of reach to the good old evolution.

Although, I must admit, if it is that easy... where are they?