Self-Replicating Machine: Beginning Design

JoshNH4H · 2014-12-27 21:18:43

The development of practical, industrial, autonomous, self-replicating machines (AKA Von Neumann Machine or VNM, which I will use to refer to this ideal case) would be a tremendous boon to all of humanity. Each of those adjectives is vital to its function: Practical, insofar as it's something that can actually built and used; industrial, insofar as it uses relatively low-energy physical and chemical techniques similar to the ones we are already familiar with; autonomous, insofar as no human labor (in the form of direct coordination or inputs that are the product of human labor) is required for its operation, upending economics as we now see it; and self-replicating, most obviously, describing its basic function.

These machines are great because self-replication likely also entails the ability to manufacture a wide variety of functional components that could be used in the production of goods beneficial to people. In effect, a VNM would result in a massive improvement in the physical standard of living of people all over the world at almost no cost.

A real self-replicating machine (SRM) will not be a VNM. An SRM, for example, might require the provision of a pre-programmed computer control unit to function properly, as well as maybe some chemical catalysts.

I propose that we start out in the tremendously important task of developing this technology by working out the techniques that will be needed to make it happen.

The first question: Where? Earth has a lot of different environments, but two stand out to me for the low value of their land to us and the plentiful availability of resources to a prospective self-replicating machine: Oceans and seas (Especially shallow ones) and deserts. I think deserts are better. While any SRM is going to use pretty large amounts of water, access to raw solid materials is more important IMO. I'd be willing to accept a need to collect water from the surrounding environment (There is some rain in deserts, after all, and it would be possible to program the SRM to grow a larger collection surface), or even to truck in some saltwater from the ocean.

Then there's a question of methods. How do you turn raw materials into self replicating machine?

Some things we know for sure. We need Iron. We need Carbon (to alloy, and make steel). We need to take this steel and turn it into stuff. We're gonna need a bunch of electric motors, and some sensors too, for feedback control.

We'll probably need glass. We're probably going to use solar power of some kind. Probably Concentrated Solar Thermal power, maybe in the form of a solar power tower. We're going to need some wiring, and it's probably going to be Aluminium rather than Copper since Copper is pretty rare in nature.

What else?

SpaceNut · 2014-12-28 10:47:02

So far the machines (ie 3 d printers or plastic solar cell fabs) that are capable of creating must have the supplies created in a fashion that the machine can use which means lots of processing by other machines that are not replicators....Simularly the electronics would require multiple processes to create them are combinations or processes as well as fabrications that are within the realm or printing.

The software is all different for each type of machines as well as the file that is created to make the copies from are not compatible for anything but the machine that it is intended to be run on.

So while we can send the abilities to print the desired item we will need to focus on the raw ore refinement processes for each to have the supplies that they will need to make them work.

Terraformer · 2014-12-29 14:17:51

Perhaps the development of additive manufacturing will reach a point at which robots can be fabricated straight from raw materials. Such robots could then be used to lay out solar panelling and collect raw materials, which would then be processed and used to build more robots and solar panels. It's not self replication though. Add in the ability to build machines to refine the raw materials, however, and make modular units which can be joined together to make another factory...

JoshNH4H · 2014-12-29 21:59:45

Advanced manufacturing techniques would definitely be a big boon to self-replicating machinery. Lucky for us, there's even a process that could be useful for this: The production of Carbonyl Iron. We could use something like a stereolithographic 3D Printer, except with metal, where the target isn't a photosensitive resin but Iron Pentacarbonyl, mixed with an organic compound that will decompose and produce Carbon at the relevant temperatures (Alternative method: Deposit Carbon on the surface and heat everything for long enough that it dissolves in). I don't know too much about this, but it's something that would be worth testing. So we'll add it to the list, I suppose.

Terraformer · 2014-12-30 09:44:06

Also, how much human labour are you willing to accept? Would automating the production of various components that require humans for final assembly fit your requirements? We've got something like 7 billion people available for such a job. I'm thinking of, say, machines to print the various parts of a smartphone, but requiring some soldering, screwing and clipping together by a human. Humans are good at finicky tasks like that. They shouldn't take long at all, and could be done by most people as long as they bother to follow the instructions.

Yes, GPs will lose their jobs to machines before hairdressers do.

Mark Friedenbach · 2014-12-30 11:33:38

I assume you've read Freitas et al on this subject:

http://www.islandone.org/MMSG/aasm/
http://www.molecularassembler.com/KSRM.htm

I also advise you to talk to some of the people on ##hplusroadmap on Freenode (IRC), the only place I know of where people are actually working on this.

Tom Kalbfus · 2015-01-12 10:15:19

Terraformer wrote:

Also, how much human labour are you willing to accept? Would automating the production of various components that require humans for final assembly fit your requirements? We've got something like 7 billion people available for such a job. I'm thinking of, say, machines to print the various parts of a smartphone, but requiring some soldering, screwing and clipping together by a human. Humans are good at finicky tasks like that. They shouldn't take long at all, and could be done by most people as long as they bother to follow the instructions.
Yes, GPs will lose their jobs to machines before hairdressers do.

if we eliminate the requirement for human labor, we can also eliminate poverty on this planet. So long as there are jobs which require human labor to do them, there will be some humans who can do this labor and some humans who can't, thus we have both rich and poor.

The rest of this post has been deleted for a violation of the Politics Rule as well as for being off topic.

-Josh

Terraformer · 2015-01-12 16:27:28

Political discussion continued at http://newmars.com/forums/viewtopic.php … 19#p123019

JoshNH4H · 2015-01-12 16:32:58

Mark, I've ordered that book and can't wait for it to arrive.

Terraformer, surprisingly enough I actually agree with Tom on this one. The amount of labor that we can eliminate from self-replication is the most important thing of all. Our economies are already massive self-replicating machines with human components; We should try to design a smaller one with no human components.

Terraformer · 2015-01-12 16:39:16

Sure, no human components would be the ideal system. But what's that phrase about the last 10% taking 90% of the effort? It might be easiest to pick all the low hanging fruit first, producing massive wealth from only a few hours of work each week, *then* closing in on getting that number down to zero.

JoshNH4H · 2015-01-12 19:06:45

Well, IMO it's just a matter of how much human labor you're willing to invest in the development of the technology. It's not as if VNM technology (even in its purest form) will completely eliminate human labor. What it will do is change the character of human labor so as to make it much more voluntary.

No matter how much the resources to develop VNM tech cost, any return will come back in perpetuity, which makes it worthwhile, no matter how small.

Mark Friedenbach · 2015-01-12 22:31:50

Josh, which book? The KSRM or the NASA technical report? KSRM is available to read online at the link I posted, although I'm glad your purchase will help fund Freitas' research. The chapters of the NASA report which have to do with the self-replicating lunar factory are available to read at the other link I posted, although I'm not sure where you'd go to get the full report. There's some other chapters I'd love to read if I had the time to devote to this stuff. Unfortunately last time I looked at the NTRS I couldn't find the report. Stupid chinese hackers ruined it for the rest of us :\

Terraformer · 2015-01-13 03:42:45

However, if you get labour down to a few hours week in a few years time, you're going to be able to reap the benefit quite soon - as well as free up a lot of human time to work on reducing labour requirements even further. Whereas you could spend decades striving for the perfect before achieving full self replication.

So, my suggested path involves moving forward with the self-replicating FabLab (which can make all it's own tools, save for a few high tech items like lasers and microprocessors), work on developing something like a solar cell printer to gain very cheap power, develop the means to extract raw elements from dirt, and keep developing desktop manufacturing. Work towards producing electronics in a desktop foundry, Perhaps we could get it to the point where humans are mainly assembling stuff after the parts have been made, and then it's a case of automating that to close the loop.

JoshNH4H · 2015-01-14 00:01:15

Mark- I'm more a fan of physical books so I'd rather order it and then have the paper.

Terraformer- It would be really helpful in this case if there were a general measure of closure for a system. Does anyone know of one?

Once we have one, the obvious criterion for what we should do first is whatever has the highest value of dC/dL, or whatever increases closure the most per unit value of research labor.

I don't think that assembly is necessarily going to be the last thing we automate. However, we are going to have to design to a more challenging standard for assembly. This means, firstly, standardizing methods of attachment so that it would be easier for smaller numbers of machines to do the attaching. It means designing standardized methods of manipulating pieces, and it probably means that assemblies are going to take up more space. It also probably means that they're going to be harder to modify with peoples' hands since they won't be designed for that.

It's a lot of very small-scale design and definitely takes a lot of design effort, but I think it's actually one of the more doable things that needs to be accomplished.

Mark Friedenbach · 2015-01-14 11:02:27

I have a dead-tree copy on my shelf, so I'd hardly be the one to talk you out of it

Just wasn't sure if you knew it was available online.

JoshNH4H · 2015-01-16 14:17:12

It's only a matter of time until Copyright holders object to paper copies because they infringe on their DMCA property rights >:-(

SpaceNut · 2015-01-16 22:11:31

Terraformer wrote:

However, if you get labour down to a few hours week in a few years time, you're going to be able to reap the benefit quite soon - as well as free up a lot of human time to work on reducing labour requirements even further. Whereas you could spend decades striving for the perfect before achieving full self replication.
So, my suggested path involves moving forward with the self-replicating FabLab (which can make all it's own tools, save for a few high tech items like lasers and microprocessors), work on developing something like a solar cell printer to gain very cheap power, develop the means to extract raw elements from dirt, and keep developing desktop manufacturing. Work towards producing electronics in a desktop foundry, Perhaps we could get it to the point where humans are mainly assembling stuff after the parts have been made, and then it's a case of automating that to close the loop.

links for the solar cell

http://inhabitat.com/printable-solar-ce … onstrated/

Inkjet printing solar panels: cheap and almost green

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

Tom Kalbfus · 2015-02-09 15:20:34

You could use them for making a giant orbiting laser and laser sail 1000 km wide.

Void · 2015-02-14 14:54:55

Part of worm brain "Uploaded" to a robot.

http://www.cnn.com/2015/01/21/tech/mci- … index.html

So, I am thinking that some time from now, a 3D printer/Robot serving as the "Queen".

Printing out little bee or ant robots. Workers only I would think.

Then the workers go out and get grains of materials that can be "Eaten" by the "Queen" to make more workers.

Dead bees/ants recycled I presume.

But when the "Queen" malfunctioned I suppose it would have to be repaired/replaced by humans.

But even this is still far fetched. However I might suggest that by "Digesting" a grain of soil material at a time it might be possible to find ways to purify the materials at a low energy cost.

Still a long way off though I think.

But such robots would not scare me very much, since their replication machines would be dependent on humans, and at most they would have a hive mind, and no stingers.

I suppose Venus and Titan might be a good place for it. In the case of Venus, perhaps they would not so much gather materials but construct a floating hive. The 3D printer would have to absorb materials from the atmosphere. In the case of Titan, I am supposing that the energy source would have to be chemical, I recall mention of an existing one, involving Titan might use hydrogen and acetylene as an energy source

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

But that does not seem to me to be a very vigorous energy source.

Spaniard · 2015-02-17 13:59:44

Void wrote:

Then the workers go out and get grains of materials that can be "Eaten" by the "Queen" to make more workers.
Dead bees/ants recycled I presume.

I think this model is very interessant.
http://www.philipmetzger.com/blog/affor … ilization/

It goes far beyond the hive model.
Instead, we could talk about a "robotsphere" in analogy to a biosphere or a ecosystem.
A group of robots that, as a group, it could replicate and "evolve" (with a code uploaded by humans).

It is based on minimize payload. Sometimes it is better to send complex "made on Earth" machines, with enough replacements.
Other parts, bulky, are maded from materials of the destination.
IRSU at great level.
And making more a more infrastructure, less and less "made of Earth" parts or raw material are needed, and different space sources are used.

Void · 2015-02-18 11:21:07

Well I will wish them success. I expect that if they do go for it, a swarm of humans and there robotic offspring could augment that nations and businesses are doing in space. Providing another aspect unique to non-national, and not necessarily profit harvesting (Until success).

While I was coming up with the above response, it occurred to me that factory automation followed anchored robotics is a bit like sponges and other anchored sea organisms which wait for their raw materials to float/swim by.

We are not quite to the level of paying self actuated robots which alter/manufacture objects, but it is likely to come when it does come to Earth, it would not be that hard to transfer it to other worlds.

Spaniard · 2015-02-18 13:52:17

I think that the numbers of Philip Metzger are too optimistics, but it is true that robotics advance a lot faster than any space related activity (if it grows at all).

This ideas are only on paper, but there is more real and mature projects on space manufacturing like http://www.tethers.com/SpiderFab.html

Tom Kalbfus · 2015-04-01 06:27:12

Void wrote:

Well I will wish them success. I expect that if they do go for it, a swarm of humans and there robotic offspring could augment that nations and businesses are doing in space. Providing another aspect unique to non-national, and not necessarily profit harvesting (Until success).
While I was coming up with the above response, it occurred to me that factory automation followed anchored robotics is a bit like sponges and other anchored sea organisms which wait for their raw materials to float/swim by.
We are not quite to the level of paying self actuated robots which alter/manufacture objects, but it is likely to come when it does come to Earth, it would not be that hard to transfer it to other worlds.

And we could build other worlds.
th?&id=HN.608046053982736308&w=300&h=300&c=0&pid=1.9&rs=0&p=0&r=0
th?&id=HN.608017050071207696&w=300&h=300&c=0&pid=1.9&rs=0&p=0&r=0
this is the largest "O'Neill" colony we might yet build using material such as carbon nanotubes, same stuff we might make space elevators out of.

Bishop Ring (habitat)

A Bishop Ring is a type of hypothetical rotating space habitat originally proposed in 1997 by Forrest Bishop. Like other space habitat designs, the Bishop Ring would spin to produce artificial gravity by way of centripetal force. The design differs from the classical designs produced in the 1970s by Gerard K. O'Neill and NASA in that it would use carbon nanotubes instead of steel, allowing the habitat to be built much larger. In the original proposal, the habitat would be approximately 1,000 km (620 mi) in radius and 500 km (310 mi) in width, containing 3 million square kilometers (1.2 million square miles) of living space, comparable to the area of Argentina or India.
Because of its enormous scale, the Bishop Ring would not need to be enclosed like the Stanford torus: it could be built without a "roof", with the atmosphere retained by artificial gravity and atmosphere retention walls some 200 km (120 mi) in height. The habitat would be oriented with its axis of rotation perpendicular to the plane of its orbit, with either an arrangement of mirrors to reflect sunlight onto the inner rim or an artificial light source in the middle, powered by a combination of solar panels on the outer rim and solar power satellites.
Also unlike the 1970s NASA proposals, where habitats would be placed in cislunar space or the Earth-Moon L₄/L₅ Lagrangian points, Forrest Bishop proposed the much more distant Sun-Earth L₄/L₅ Lagrangian points as the sites for the habitats. more from Wikipedia

Last edited by Tom Kalbfus (2015-04-01 06:39:09)

Mars_B4_Moon · 2023-12-30 05:38:58

New DNA nanobots can replicate themselves using UV light
https://interestingengineering.com/inno … themselves
An international team of scientists has collaborated to develop a new DNA-based nanobot that can self-replicate indefinitely under the right conditions.

New Mars Forums

Announcement

#1 2014-12-27 21:18:43

Self-Replicating Machine: Beginning Design

#2 2014-12-28 10:47:02

Re: Self-Replicating Machine: Beginning Design

#3 2014-12-29 14:17:51

Re: Self-Replicating Machine: Beginning Design

#4 2014-12-29 21:59:45

Re: Self-Replicating Machine: Beginning Design

#5 2014-12-30 09:44:06

Re: Self-Replicating Machine: Beginning Design

#6 2014-12-30 11:33:38

Re: Self-Replicating Machine: Beginning Design

#7 2015-01-12 10:15:19

Re: Self-Replicating Machine: Beginning Design

#8 2015-01-12 16:27:28

Re: Self-Replicating Machine: Beginning Design

#9 2015-01-12 16:32:58

Re: Self-Replicating Machine: Beginning Design

#10 2015-01-12 16:39:16

Re: Self-Replicating Machine: Beginning Design

#11 2015-01-12 19:06:45

Re: Self-Replicating Machine: Beginning Design

#12 2015-01-12 22:31:50

Re: Self-Replicating Machine: Beginning Design

#13 2015-01-13 03:42:45

Re: Self-Replicating Machine: Beginning Design

#14 2015-01-14 00:01:15

Re: Self-Replicating Machine: Beginning Design

#15 2015-01-14 11:02:27

Re: Self-Replicating Machine: Beginning Design

#16 2015-01-16 14:17:12

Re: Self-Replicating Machine: Beginning Design

#17 2015-01-16 22:11:31

Re: Self-Replicating Machine: Beginning Design

#18 2015-02-09 15:20:34

Re: Self-Replicating Machine: Beginning Design

#19 2015-02-14 14:54:55

Re: Self-Replicating Machine: Beginning Design

#20 2015-02-17 13:59:44

Re: Self-Replicating Machine: Beginning Design

#21 2015-02-18 11:21:07

Re: Self-Replicating Machine: Beginning Design

#22 2015-02-18 13:52:17

Re: Self-Replicating Machine: Beginning Design

#23 2015-04-01 06:27:12

Re: Self-Replicating Machine: Beginning Design

#24 2023-12-30 05:38:58

Re: Self-Replicating Machine: Beginning Design

Board footer