Over in this fascinating thread: http://boards.straightdope.com/sdmb/showthread.php?t=798052, much seriousness is given to the notion of self replicating probes. One poster even suggesting we are only a hundred years away from the technology.
In theory I get it. But as a practical matter I’m not so sure. Can someone explain how they would work, exactly? As hard as it to build a space capsule it seems impossible that one sent into space could then carry out the complex task of assembling a clone of itself without human assistance. That’s setting aside the impossible (in my mind) job of mining the raw materials let alone taking into account the problem of building a propulsion system and finding the fuel to power it.
The SF stories I’ve read with self replicating probes generally assume that we have both strong AI and nano tech as solved problems. In theory all the elements you need are out there, if you have the ability to do 3D printing at the atomic scale then you can pretty much build anything, including a copy of yourself.
So is 3D printing possible at that level? Maybe, see here:
Of course there’s also the issue of building something that will still function correctly after potentially thousands of years travelling through space to reach the next place it can build more copies. It would need to have incredible self repair ability.
To make it easy assume you use fusion as an energy source. Hydrogen is abundant in the universe so you don’t worry about running out of fuel. As your ion drive propels you through space you are gathering free particles of all sorts of elements, but you don’t need that many. Size is not a constraint, you don’t need light weight materials, you don’t need very efficient machinery. The elements you need the most of are found in big chunks around stars. Your giant space roaming machine is made mainly of some hard material like iron, a conductive material like copper, silver, or gold, and insulator like silicon, some carbon to form polymers for flexible pieces though you may not need many. Maybe you make lots of semiconductors for electronic logic, or maybe you don’t need that much logic, the machine only has to replicate itself and send back information about what it finds. A lot of the control logic could be performed electromechanically or using vacuum tubes without needing highly advanced semiconductor manufacturing capability. Your starship is a giant automated machine shop that can make all the pieces that it is constructed from, and the additional pieces of a fusion reactor, ion drive, and material collection devices. It only needs to operate long enough to make it to another star and build a copy of itself. If it lasts long enough to create two or more copies of itself the number of probes multiplies until they’ve used up all the materials needed to copy themselves in readily available form for all stars within range of travel.
We already have the example of self-replicating machines, and in fact, you are one (and are composed of billions more on a cellular level). No, we won’t me making Voyager-type probes that smelt metals, fabricate microprocessors, produce plutonium, and assemble it all together into a copy of itself, bit it is not so fanciful that we make take a ‘machine’ made of simple elements on a very small scale and combine them together into a device complex enough to male more of itself from elemental and mineral raw materials. Call it a Space Chicken (with apologies to Freeman Dyson), laying eggs as it puts around the galaxy, and indeed, starting with actual biological material, sufficiently hardened to withstand the radiation and thermal conditions of interstellar space may be the very way we proceed.
I’ll be very suprised if we don’t have something like this capability by the end of this century if not sooner. Whether we use it to explore space, and what we might do with such knowledge, is another question.
In the interests of completeness, I’ll point out that there’s a strain of SF with self-replicating AI spacecraft that replicate on a macro scale*, rather than a nano-assembler scale. Case in point: Berserkers, which create more berserkers in planetary systems using automated factories.
*Presumably macro-scale. The guts and operating principles of a berserker factory aren’t described in any of the works I’ve read, so I can’t be sure, but it’s certainly not grey goo growing in orbit.
Robert Freitas has written a quite speculative article about one possible form a self-replicating probe might take;
…this turns out be a pretty large spacecraft, about 1.07x10e10 kg in mass (ten million tonnes). To carry a payload that is big and complex enough to survey the system/mine the objects found there/refine the mined materials and produce a copy of itself, you need to start with a fairly large and complex payload.
Stranger’s idea of a biological self-repping system looks a lot more attractive, except that you have to keep the material alive during transit, then find somewhere for it to live that won’t immediately kill it. Note that all the self-repping biological systems on Earth survive and thrive precisely because they are introduced into an environment that is favourable to them, with free oxygen, liquid water and carbon dioxide available for the taking. Most interstellar destinations won’t be like that.
And then we have to consider the ethical dimension; if one of our biological self-repping systems found its way to a suitable world, that world is quite likely to have indigenous life. We would have introduced an artificially-designed replicator into a biosphere that might not be able to compete with it.
We could probably do better than this, but even in the worst case, you could load an initial probe with a reservoir of millions of IC chips - each the size of a speck of dust - that serve as the brains for new probes. That part would be non-reproducible, but every probe would split its reservoir in half when it created a child, and the child would go a direction that no previous probe had gone, then bear a child with another half.
The probe itself could be very rugged and simple - not really having to care how hospitable any one planet is, because it’s built to survive almost anything - and able to scavenge the materials it needs, because it’s built to use whatever is easy to get - maybe having a few different designs depending on what planetary surfaces are most likely to be composed of.
If the probes have enough information about their target destinations, they would simply not send themselves to places that wouldn’t allow more probes to be built. With low-level information about their targets, they could split their IC load based on the probability that the target destinations are good targets. But otherwise, if you send out enough initial probes, so long as even a few are able to get through a few reproduction cycles, they would end up being able to go pretty far.
I mentioned in that other thread that humans are pretty good self-replicating probes already. DNA is an amazing self-replicating nano-machine to beat all nano-machines. It just lacks error correction to prevent mutations and evolution that might lead to something disastrous. Presumably we’ll develop that ability before long so humans themselves can be our Von Neumann probes. We already are pretty fantastic nano-machines.
Technically, p53 is an error correcting molecule that scans and fixes DNA.
And, of course, the evolutionary process allows for the “correction” of DNA by the process of profligate multiplication mixed with the weeding out of poor specimens.
I think a lot of the issues with mining and building are solved if you skip planets and think asteroids or comets. These are rich in the elements you’d need for a space ship and their lower gravity (compared to planets) dramatically reduces the amount of energy you’d invest in extracting materials and building a duplicate probe.
Looking at self-replicating, we’re not very far away from that right now. Artificial intelligence will be the biggest hurdle, I think, but we do need to improve on machines that have the dexterity and sensory input of a human. If you had a human-like robot, you could put those to work with 20th Century machines and get a simple self-replication scheme up and running. Obviously, that would leave lots of room for improvement…
I tend to assume nanotech will be necessary to achieve interstellar probes - if not for the self-replication task, then at least for the self-repair task. It’s not feasible with current technology to keep equipment operational for hundreds of years at a time, but nanotech should provide a way to build and repair the system at a molecular level. Comparing nanotech to biological life, there are life forms (like trees) that can maintain themselves over hundreds or even thousands of years. In fact, by the time we have the capacity for an interstellar self-replicating probe, I think the line between “biology” and “technology” will be a very fuzzy one.
I do assume that an interstellar self-replicating probe will be quite large by today’s standards. It will probably have more in common with an aircraft carrier or a city than a Voyager-style probe. But if we’re thinking asteroids, this isn’t so hard to build. The biggest hurdle right now is having to fight Earth for every single trip into space.
On the issue of a propulsion system: we have some theoretical systems on the drawing board that would be dramatically more efficient than modern rockets. Ion engines produce tiny accelerations - perhaps just a few percent of Earth gravity. These engines will not get you into orbit in the first place. Once in orbit, you just maintain that continuous acceleration over a long period, and nearby stars are not really all that far away. In some ways, propulsion is one of the simpler problems to solve.
If it can build the parts it’s made out of it and assemble a clone it doesn’t need to repair anything, it will just replace whole major systems. It’s going to be building those systems anyway to replicate itself, it just doesn’t split into separate entities until it has at least two of everything.
