Self-Replicating Technology is Key

JoshNH4H · 2018-03-05 19:02:00

Prior to now, I've been a big proponent of what could be called the "Export Model (XM)" of Martian and Space development. I discussed my thoughts on the matter in this post as well as in other posts in that thread.

The basic idea for the Export Model is to treat space colonization as a private-sector/public sector infrastructure investment, and to expect that it will return a profit over time. In order to generate a profit, you need exports. This makes the question of exports a deeply vital one for settling Mars. It means that a viable plan to generate substantial profits through exports is vital not just for the future economic viability of the settlement but for the settlement to exist at all.

According to the Export Model, the economies of Mars, the Moon, and the various other places in the solar system will be deeply and inextricably tied to that of Earth because of their dependence on imports and their need to provide a return on investment through exports.

In this post, I want to try to develop an alternative model, which I will call the Self-Replicating Economic Model (SREM). The SREM is not a refutation of the XM, because it doesn't counter any of the underlying assumptions, but it does look at the same ideas in a different light and come to different conclusions about what the important things are to get from here to there. I think SREM accords better with a lot of the talk you see on this forum and in the space enthusiast community.

Anyway, fundamentally the idea behind the SREM is this: That if self-replicating machines (or some close approximation thereof) are available to Martian settlers, after the initial establishment of a settlement exponential growth will pretty much guarantee that it will be prosperous and able to accommodate many new immigrants.

In the ideal case, you would have a machine that was able to extract resources from the environment, refine them into usable materials, form them into components, and assemble these components into a new machine. Because this necessarily entails the capacity to build a whole bunch of different things. In other words, an ideal Self-Replicating Machine is also pretty close to being a universal constructor.

Any real self-replicating machine (or self-replicating system, because resources are not always located in the same spot) is going to require some amount of labor input and also will probably need some input of finished products.

Depending on the size of the gap, it could potentially be closed by emigrants selling off all of their earthbound assets in order to buy things to bring to Mars and sell them there.

The key is that in this model, the offworld settlements are almost totally self-sufficient, and population growth is driven by the availability of higher standards of living and by people's desire to try something new.

There's a few highly nontrivial technological developments that need to happen before this is a viable model:

Self replicating machines (or a very close approximation thereof) need to exist, and need to have relatively short duplication times, probably less than 10 years and ideally closer to 1-2 years
We need to develop life support technology to the point where we have it licked. Air controls need to be as simple as I described in the Rickety Skiff thread. Food, water, air, pressure, temp., all need to be developed down to being a science with lists of predictive rules and procedures.
The cost of a ticket needs to go down. Way down. In the first few decades of their lives people need to be able to save up enough money to afford it and then some.

What do you all think?

Terraformer · 2018-03-06 04:26:43

Self replicating technology =/= self replicating machine. It's possible to conceive of a highly automated machine shop that can (almost) fabricate all it's own tools, requiring humans only to do final assembly and install certain imported components (e.g. computer chips). That would mean that the limit on expansion would be the rate of human population expansion, but you're assuming that that will be quite high.

JoshNH4H · 2018-03-06 08:10:48

That work in theory, but I don't think assembly jobs are productive enough to generate the high standards of living you would hope for on Mars. That's not to say it kills if if a few people are involved in certain assembly tasks (such as construction) but it would be best if they weren't

elderflower · 2018-03-06 12:36:27

A major issue that will almost certainly require a lot of human intervention is that of fixing broken equipment and adapting functioning equipment to new tasks. As the story goes "That's 50 bucks for hitting it with a hammer and 500 for knowing where to hit it and how hard!"
There is no mechanised production if the machines don't work.

louis · 2018-03-06 15:10:17

I think actual self-replication with no human intervention is unnecessary and would be a diversion. What is required is a highly level of automation, robotics, CNC machines and 3D printing and in the initial stage, importation of prepared materials that can be used easily in automated processes.

The ticket price per se is not going to be a huge issue for the first couple of decades in my view because I don't think we are going to see much "homesteader" colonisation in the first couple of decades. I think we will see Mars develop firstly as a research station on the Antarctic model, serving also as a base for exploration and media enterprise; secondly as a formal University Institute. Other activities such as life support, construction, agriculture, mining, industrial production and so on will be developed to serve the research, exploration and media activities. People who go to planet will not need to raise funds for the ticket themselves.

JoshNH4H · 2018-03-07 08:34:09

Elderflower-

I agree that maintenance and repair are going to be some of the trickiest parts of the whole process to automate. I work in a somewhat related field (I repair 3D Printers) so you could definitely say that this particular kind of labor input is something I spend time thinking about

In my experience, the best way to "automate" repairs is to build machines that don't break in the first place. Easier said than done, sure, but not impossible. So far as I can tell, what you need to do is to iterate your design process (the second generation is going to be much more reliable than the first), to build solid machines well inside their failure tolerances, and to spend money up-front on engineering/design/testing and machine construction rather than on repairs later.

The second-most important thing to do is to prevent user error as much as possible. The best way to do this is to keep the user away from the machine by automating everything you can. It's also vital that the user be well-trained and familiar with basic functions so that when they do have to do something they don't mess it up. To the extent that there is maintenance required, it should be a short list of items which are simple to perform, performed on a regular schedule, and effective at improving performance and lifetime.

The third-most important thing to do is to put sensors everywhere, record everything, and provide good documentation. If it's good enough, the machine might even be able to pinpoint problems to individual parts.

The fourth-most important thing to do is to make the machine completely self-calibrating from end to end.

Fifthly, you need to standardize the way that individual parts and systems connect to the rest of the machine so that they can be removed and new ones installed by robots when one wears out or breaks.

So that's what you could do if you wanted to automate my job 100%. It's hard, and it's something that you would probably do piecemeal, but possible.

Louis-

I think actual self-replication with no human intervention is unnecessary and would be a diversion. What is required is a highly level of automation, robotics, CNC machines and 3D printing and in the initial stage, importation of prepared materials that can be used easily in automated processes.

I agree that complete self-replication is not really possible (this would necessarily include self-designing machines, which is interesting but not necessarily a good idea or possible), but I have in my head a murky notion of closure, where as you get closer and closer to a fully self-replicating machine you rapidly approach a situation where people can meet their needs with very little work of any kind. This frees up time and resources for innovation and investment/expansion, which on an empty world can create the kind of advancement, freedom, and prosperity that has only existed at rare moments in human history. KSR does a pretty fantastic job of showing this in the Mars Trilogy and his other novels, which I think is part what makes them so compelling.

The ticket price per se is not going to be a huge issue for the first couple of decades in my view because I don't think we are going to see much "homesteader" colonisation in the first couple of decades. I think we will see Mars develop firstly as a research station on the Antarctic model, serving also as a base for exploration and media enterprise; secondly as a formal University Institute. Other activities such as life support, construction, agriculture, mining, industrial production and so on will be developed to serve the research, exploration and media activities. People who go to planet will not need to raise funds for the ticket themselves.

We've talked about this before of course. As in the past I don't agree with your funding streams (I think they're fanciful), but more importantly I would say that what you're talking about is basically the export model, adjusted for what is in effect exports of prestige and media, and perhaps mixed with the public subsidy model.

And I want to make it clear that I do not "disagree" with the Export Model. I don't think it's "wrong" or that it wouldn't work, provided you develop products worth exporting (the nature of these products is where we differ, but we've covered that elsewhere).

I'm trying to develop a different model for thinking about Martian settlement and how it could be made to happen. Just as there are several different ways to support a news website (subscription, ad revenue, or money from government/a donor/one's own pocket), there could also be different economic models for settling a planet. I'm not going to oppose any settlement attempt that I think is viable (or even one that I think is not viable, so long as I thought something good like new technology or infrastructure will come from it).

As I see it, the export model is one that could be pursued sooner but will result in slower overall growth. The Self-Replicating Model will take more of an investment in technology development (one that will benefit Earth as well as Offworld Settlements) and therefore would likely kick off later, but progress faster once it does.

These two models aren't mutually exclusive either, but I feel that there's something to the SREM and I want to try to develop it in this thread.

Terraformer · 2018-03-07 09:05:19

People are going to have to clean toilets on Mars. Are janitorial jobs really the highly productive ones we want for Mars? If manufacturing is made even more productive than it is on Terra, each worker will be very productive - and it's likely manufacturing will be a small share of the overall economy as a result.

Due to diminishing returns, I think we're very far off 100% automation. But I can easily see a situation where human input is limited to monitoring, programming, and final assembly.

JoshNH4H · 2018-03-07 10:40:09

Self-replication does get exponentially more complicated the more you want to automate, but advanced manufacturing can get you a lot closer than we were able to get in the past. A lot of the customers I work with are jewellers, and their process (at a production scale) looks like this:

Design metal pieces in a CAD software such as Rhino
3D Print the parts out of wax
Post-process the parts to remove support material
Combine several different waxes into one larger piece connected with sprues (this is called a "casting tree")
Surround the casting tree with a mold (made of plaster or something like plaster)
Allow the mold to dry and heat it to bake out all the water
Melt out the wax and cast into the mold with gold or silver. The wax is generally not recycled because the machine will not allow you to do that but it probably could be.
Remove the pieces, separate the sprues, polish as necessary, and recycle the extra metal

With some modifications (for example, a heated mold and possibly injecting the metal under pressure), this process could work for Steel and Aluminium too. For parts that you need a lot of, you could investment-cast a die and then die cast the parts (This saves wax and plaster).

Tools like the CNC, mill, lathe, drill press, etc. are also options for sure, but this seems like a nice, very versatile process that could form the core of the self-replicating manufacturing process.

At some point we may get better at 3D printing metals too. In the ideal world you'd have an omni-printer that could print whole machines, fully formed. Were not there yet.

I suppose what I really mean by a "self-replicating economic model" is a self-sufficient economic model. One key predicate of the Export Model is that you need exports to fund your imports. If you have no imports (or if your imports are so small as to be insignificant). This becomes mostly irrelevant.

Autarky has not worked too well historically, but Mars is sort of a special case where you have lots of room to expand and high transportation costs serve as a big disincentive to trade. Self-Replicating machines ease this substantially by serving as an infinite productivity multiplier, in a certain sense. In the real world they would not be an infinite productivity multiplier but in the material goods sector they could be a very large one.

louis · 2018-03-07 19:08:02

Yes, that's pretty much the way I see things going.

I do recall a few years ago someone claiming that steel needed to be rolled to impart strength. There may have been some discussion about that, not sure. Put it another way: do you need to have a steel rolling mill to produce quality steel before you can use steel in 3D printing processes. Or could steel be produced on a small scale as well.

JoshNH4H wrote:

Self-replication does get exponentially more complicated the more you want to automate, but advanced manufacturing can get you a lot closer than we were able to get in the past. A lot of the customers I work with are jewellers, and their process (at a production scale) looks like this:
Design metal pieces in a CAD software such as Rhino
3D Print the parts out of wax
Post-process the parts to remove support material
Combine several different waxes into one larger piece connected with sprues (this is called a "casting tree")
Surround the casting tree with a mold (made of plaster or something like plaster)
Allow the mold to dry and heat it to bake out all the water
Melt out the wax and cast into the mold with gold or silver. The wax is generally not recycled because the machine will not allow you to do that but it probably could be.
Remove the pieces, separate the sprues, polish as necessary, and recycle the extra metal
With some modifications (for example, a heated mold and possibly injecting the metal under pressure), this process could work for Steel and Aluminium too. For parts that you need a lot of, you could investment-cast a die and then die cast the parts (This saves wax and plaster).
Tools like the CNC, mill, lathe, drill press, etc. are also options for sure, but this seems like a nice, very versatile process that could form the core of the self-replicating manufacturing process.
At some point we may get better at 3D printing metals too. In the ideal world you'd have an omni-printer that could print whole machines, fully formed. Were not there yet.
I suppose what I really mean by a "self-replicating economic model" is a self-sufficient economic model. One key predicate of the Export Model is that you need exports to fund your imports. If you have no imports (or if your imports are so small as to be insignificant). This becomes mostly irrelevant.
Autarky has not worked too well historically, but Mars is sort of a special case where you have lots of room to expand and high transportation costs serve as a big disincentive to trade. Self-Replicating machines ease this substantially by serving as an infinite productivity multiplier, in a certain sense. In the real world they would not be an infinite productivity multiplier but in the material goods sector they could be a very large one.

SpaceNut · 2018-03-07 20:37:16

Doing a little data mining I come up with another topic started a while ago... Self-Replicating Machine: Beginning Design

JoshNH4H · 2018-03-08 09:40:24

Hey SpaceNut,

I'd forgotten about that thread! I do think the subject matter here is different, because that thread was focused on engineering design and this one (at least originally) on economics.

louis wrote:

Yes, that's pretty much the way I see things going.
I do recall a few years ago someone claiming that steel needed to be rolled to impart strength. There may have been some discussion about that, not sure. Put it another way: do you need to have a steel rolling mill to produce quality steel before you can use steel in 3D printing processes. Or could steel be produced on a small scale as well.

The tl;dr is that yes, rolling does improve the strength of steel by a large but variable amount, and that any cast piece is necessarily not rolled and therefore will necessarily be weaker, but also less brittle. The benefits from rolling the steel go away once you melt it to do the casting, and flattening your cast pieces into pancakes is not really what you want to do most of the time.

The reasons why are a bit technical, but Steel (most metals, but especially Steel) has two figures that could be called its "Strength": There's the Yield Point, and there's the Ultimate Strength.

If you imagine a piece of metal, it's a bit like a spring. When you exert a force on it, it will stretch a bit. If it's only a little bit of force, the material will return to its original shape when the force is released. If it's a lot of force, the material will permanently deform (or "Yield"). If you exert even more force, the material will break. The point at which it yields is known as the "Yield Point", and the point at which it breaks is known as the "Ultimate Strength".

The key thing to know about the yield point is that it is not fixed, even for the same alloy, or for the same object. Once a piece of metal has deformed, the yield point will be higher than it was prior to yielding. The reason for this is that the crystals within the metal become stretched and more resistant to further yielding.

This is the key to rolling steel: By forcing it to yield a lot you can increase its yield point up almost to the ultimate strength of the metal (which is fixed). The opposite of rolling is annealing, in which you heat the metal to high temperatures in order to allow recrystallization. Rolled steel is strong, but brittle; annealed steel is weak but ductile. Due to the way materials harden as they solidify, cast pieces are in between the two.

New Mars Forums

Announcement

#1 2018-03-05 19:02:00

Self-Replicating Technology is Key

#2 2018-03-06 04:26:43

Re: Self-Replicating Technology is Key

#3 2018-03-06 08:10:48

Re: Self-Replicating Technology is Key

#4 2018-03-06 12:36:27

Re: Self-Replicating Technology is Key

#5 2018-03-06 15:10:17

Re: Self-Replicating Technology is Key

#6 2018-03-07 08:34:09

Re: Self-Replicating Technology is Key

#7 2018-03-07 09:05:19

Re: Self-Replicating Technology is Key

#8 2018-03-07 10:40:09

Re: Self-Replicating Technology is Key

#9 2018-03-07 19:08:02

Re: Self-Replicating Technology is Key

#10 2018-03-07 20:37:16

Re: Self-Replicating Technology is Key

#11 2018-03-08 09:40:24

Re: Self-Replicating Technology is Key

Board footer