The scenario goes something like this: In the not so far future a fully automated exploration vessel crashes on an uninhabited planet not unlike earth. Unlike earth however, this particular planet is completely lifeless, as most planets tend to be. This planet can be thought of as a pre-terraformed earth, complete with an orange sky and carbon dioxide filled atmosphere. With the hibernating crew now dead the androids tasked with ship maintenance are all that's left. Bleak, but the human crew wouldn't have survived for long anyway. Powered by remaining solar cells, the androids now without their fleshy overlords are left to their own devices.

Skip ahead a few millennia and these mechanical colonists have established themselves as a machine race.

On paper this sounds great! However, it can't possibly be that easy. Aside from whatever equipment and scraps from the ships they can find, these robots basically have caveperson-level technology. Or in this case cave-droid technology? Anyway... the point is they'll have to build more of themselves at some point. Otherwise they'll die and there won't be a story to tell.

The question is: [copy paste title].

  • 23
    $\begingroup$ What technology did they bring with them? Presumably the humans weren't going to land on the planet and instantly revert to being hunter-gatherers, so they must have brought some technology with them. $\endgroup$
    – Cadence
    May 9, 2022 at 19:28
  • 10
    $\begingroup$ If they can repair each other they can replicate. $\endgroup$
    – John
    May 9, 2022 at 20:30
  • 2
    $\begingroup$ A fully automated ship would have automated repair facilities. Questions: Are the androids strong AI? How extensive are the repair facilities? Were they complete enough to construct complete replacements? Did these repair facilities survive the crash? Are there ship board libraries that contain information about how to build repair facilities? Do the androids have strong enough AI to able to synthesize new solutions? $\endgroup$ May 9, 2022 at 20:47
  • 1
    $\begingroup$ Yeah, the androids' level of AI intelligence, their access to information known by the civilization that built them (do they have the complete database of human knowledge in their heads? in the ship's computer? do they have their own blueprint, at least?) and the technological design of the androids themselves (do they have lasers in their fingers? X-ray vision? Stronger than a human? Weaker than a human? do they need to recharge or have infinite energy cores?) all are important to know to answer this question. $\endgroup$
    – MarielS
    May 10, 2022 at 5:31
  • 2
    $\begingroup$ en.wikipedia.org/wiki/Code_of_the_Lifemaker ? $\endgroup$ May 10, 2022 at 11:34

12 Answers 12


Ignoring software...


The androids need to be able to fabricate several different sorts of material from raw materials. Metallurgy is the big one. This will be difficult for them. On Earth, humans were able to discover all these ores because there were oodles of us. Like little ants, humanity sent scouts out in all directions, and dumb as they were, these scouts stumbled onto interesting stuff.

The androids can't do that. They are limited in number (until they build more, but they can't build more until they build more for scouts... ack).

The androids are presumably highly intelligent, and have access to humanity's collective knowledge on geology (or xenogeology or whatever it's called). That, plus hopefully some long-range scans give them a slight advantage and they can mine for ore using only mechanical tools they can improvise.

This is a risky endeavor. Mining is a very nuanced skill, they will have super-intelligence but none of the nuance. They could easily lose one or more of their numbers to cave-ins and even potentially explosions... and they have so few to lose. But assuming they don't bite it here and now, then more challenges lie ahead.

Metallurgy on Earth required biomass. I mean, what do you burn on this lifeless world to produce excess heat? Heat enough to start smelting copper and gold and iron? Heck, how do you do it without an oxygen atmosphere? Even if they magically had infinite lumber or charcoal, can't really burn it can they? Their best bet is to hope that some high technology from the ship survived unscathed. Probably too much to hope for that they have a functioning fission/fusion reactor, but you mention solar cells. How many do they have?

Smelting most of these metals with electrical's difficult. Not impossible, given enough juice and the right amount of the other equipment. This stuff's semi-consumable too, even if they have it... it wears out, and they won't have replacements until they can make their own.


I'm sort of a crank in that I give a little too much credence to pseudoscience when I should know better. It makes me wonder (though not believe) in really kooky stuff from time to time. But one of them that I think probably deserves it is "abiogenic oil". This is the idea that liquid fossil fuels on Earth aren't truly fossile fuels, but primordial in nature. That is, methane and even ethane (and higher) were created along with our sun and planets. And that alot of it ended up on planets. Mostly Jupiter (enough gravity to retain the stuff), but some on Earth too.

You won't get coal (definitely 100% fossil), but oil's a maybe.

And just like they couldn't prospect for metal ores, there are challenges here too. But at some point, they can maybe start extracting raw petroleum (with luck, there are near-surface deposits), and the chemistry's not going to be particularly challenging for them. They'll have plastics if they can do metal.


They need this so bad. Not only does it make it possible to make more of themselves (I'm assuming they are semiconductor-based, after all, and not some Alien-Ash robot using milk pumped through little rubber tubes), but this is what they need to be able to do decent power. No coal, petroleum's too important to waste burning (and no oxygen to burn it). Maybe they'll do fission/fusion later, but that has to come after. Wind turbines just won't be enough. They need photovoltaics.

Technology at this level is much better documented than mining techniques, so it's possible they can steal some of that nuance that they'd otherwise have to develop on their own. Maybe skip a few steps. But unless critical semiconductors are just absent on this planet for some reason or the atmosphere somehow poisons how they need to dope the various layers, this will be the most certain of all their trials. They will succeed, assuming they've mastered the other things first.

All in all, they do have hope. But things are going to be dicey for a long while... we're talking centuries of having virtually no redundancy to survive catastrophe. Centuries, possibly, hoping the old equipment doesn't wear out before they've achieved the ability to replace it. A person more clever than I could probably see lots of opportunity to develop the story where it looks hopeless, only for one to recognize they can jury-rig some other equipment and continue.

  • 4
    $\begingroup$ There exist chemical processes that can reduce carbon dioxide to carbon (and the reagents can also be recycled eg by electrolysis) with carbon you can make coal gas (using heat and water) and thence methane and other alkanes, with carbon you can also smelt iron in an arc furnace, but you can do that with hydrogen too. $\endgroup$
    – Jasen
    May 10, 2022 at 4:40
  • 5
    $\begingroup$ With the "almost only Co2" athmosphere, air-extracted plastics become quite viable, given a stable and high-ish energy supply $\endgroup$
    – Hobbamok
    May 10, 2022 at 10:43
  • 3
    $\begingroup$ Can the androids power down? If so, things that take hundreds of humans can be done by a handful of simplistic drones. Program a bunch of ant drones to dig your pit mine, program the ship to wake one of you up any time the ant drone count drops below X% (if the drone replacement can't be automated. Or just have the androids take turns waking up every 2 years or something to check overall status) and basically sleep through the hundred years it takes to get your ore bed exposed. Don't sleep too long, though, or your ant drones may have gone sentient. $\endgroup$
    – user34314
    May 11, 2022 at 5:20
  • 2
    $\begingroup$ Just spotted that they only have solar. Transmutation of the elements will have to wait. Gratzel cells or thermoelectrics or hydro power it is. $\endgroup$
    – user86462
    May 11, 2022 at 11:05
  • 2
    $\begingroup$ @Joshua Titan has an ocean of ethane. Gasoline would be mostly solid on Titan's surface. $\endgroup$ May 12, 2022 at 0:45

1. Initial Motivation:

First, establish the androids' primary directives:

  1. The preservation and support of the human crew.
  2. The preservation and support of the spacecraft.
  3. The preservation of the ship's exploration data until retrieval by authorized personnel.

Directive 1 has failed. Directive 2 is impossible given the state of the ship. Directive 3 is all that remains.

Because this directive persists until more humans arrive, it incurs an implicit sense of self preservation.

2. Logistics:

The ship was designed for very long-distance/long-term trips, and would reasonably have a way to fabricate replacement components. This fabricator would probably not be able to handle any raw material (eg, you can't throw in a lump of iron ore and expect it to work), but would probably be able to use materials of a certain purity which would've been stored for this purpose--similar to how 3D printers need spools of a specific type of plastic.

This would allow the androids to produce replacement components for themselves to prolong their own lives. They could also produce the necessary components to create backup fabricators. Eventually, though, they would run out of the necessary materials for the fabricator.

Upon reaching a critical lower threshold of fabrication materials, the androids would begin to create ways of refining raw materials into substances that the fabricators could use. They could derive this information from reference materials on the ship's computers, and their own understanding of the specs of every device on the ship.

Eventually they would have machinery capable of converting any available substance into useable fabricator materials. There would still be components which would require rare elements they couldn't readily obtain, but they'd essentially be able to skip much of the "stone age" technological requirements.

3. Self-replication:

The next step is the jump from "preservation of self" to "preservation of species". The androids would realize that there are still situations which could incur an irrecoverable loss of one of their number, and therefore jeopardize the entire mission.

The solution is the same as it was for the fabricator: redundancy. Instead of merely creating replacement parts for themselves, they would work to create duplicates of each member, and of the exploration data itself.

They would eventually need to cannibalize the ship's own computers in order to grow their numbers past a certain point (producing the processors necessary for their brains purely from raw materials is still beyond their capabilities); from here on, each android will contain a copy of all exploration data within themselves.

Internalizing the data which is the soul of their entire purpose has an unintended consequence: it makes each android life important, to a degree. Instead of duplicate androids being somewhat expendable, the loss of a single one represents the loss of an entire copy of the data they are collectively meant to preserve. If data is corrupted over eons, then comparing the copies is the only way to ensure its integrity.

4. Growth:

Perfect duplication carries with it several problems: blind-spots and deadlocks. Essentially, if the original androids were unable to solve a problem (such as synthesizing the necessary materials to mass-produce their brain processors), then no number of duplicates would be able to solve that problem.

Similarly, it was possible for two duplicates to find themselves in logical deadlocks between each other until external stimuli broke the loop.

The solution was to add a degree of randomness to each new duplicate. The original directive could not be changed, and neither could the exploration data, but most of their personality and cognitive functions could be tweaked slightly.

Whichever duplicates showed the most beneficial improvements would be used as a baseline for future generations, as decided by the un-altered Originals (with the intent to ensure that the alterations didn't drift in any direction which could harm the mission).

At this point, their population was still limited, so the minds of duplicates were successively altered with each "generation", as they worked towards the goal of cracking mass production. Bit by bit, their ability to conceive of new solutions improved. Their technological capabilities grew step by step.

After many centuries of tiny improvements and gradual degradation of their population (the existing brain processors could only be repaired so many times before total failure), one of the highly "evolved" duplicates managed to determine a method of synthesizing the necessary raw materials for their processors.

From that moment on, their population was no longer limited, and they spread across the globe. Their species grew and continued to evolve, but within each of them is still a copy of the original exploration data, ostensibly unchanged after billions of duplications.

  • 1
    $\begingroup$ Great answer, thank you! I hadn't thought about cannibalizing the ships computers in order to increase the number of androids early on. That also explains why they'd postpone fixing the ship for a while. I'll keep that in mind. $\endgroup$ May 13, 2022 at 8:38

They are programmed that way

The first question would be: why do the androids replicate at all? After all, with the humans dead, it could very well be that the androids no longer have any task to do and just shut down to save energy.

So you need androids who want to replicate. Since they do not have instincts or pleasure centers leading them that way, it must be something that has been programmed into them.

A few reasons:

  1. Colonization: if the explorer ship finds a suitable planet, it drops a few androids with spare parts and machinery and, while the ship goes back home, they start preparing the terrain for future colonizers.

  2. Maintenance: in order to avoid awakening the humans each time an android needs to be repaired, they are prepared to repair themselves or, even better, one of their own withou human intervention. They would be ready to, in case that an android suffers a major malfunction, to salvage whatever parts are useful and create a new android from spares to replace it.

Now, #1 could use the trope of self-replicating androids, depending on the size of the task (it is not the same if they are left to prepare a base than if they must terraform the planet). That would already be enough for your answer.

But #2 (or #1 without an explicit replication programming) can also be useful, and maybe even more interesting.

In effect, even if they are by defect not programmed to increase their number, the stranded robots may be aware that they are doomed because no one will be looking for them and eventaully they will run out of spare parts. But their programming forces them to ensure that a number of them must survive at any time.

So they might decide on their own to increase their numbers, in order to increase the workforce to produce on their own the spares (by mining, creating labs...). They may be even be willing to modify their designs, to make do with what it is available, starting some kind of android evolution. Do they need to dig for metals? A tunnelling android. Do they find fibers from a tree's flowers useful? A climbing android might be more cost effective (and sustainable) than cutting the trees.

Of course, for all of this, you need:

  • Intelligent androids that are self-conscious and can have independent action (a given if you are in the #1 situation).

  • Androids who know anything that there is to know about them. Their schematics, but also production process, mining techniques. That is not difficult: you just want your androids to have access to all knowledge in their brains to help you, so they come with it by default.

  • A planet (or at least a landing area) where the resources needed are relatively easy to extract with the original equipment available to the androids. Again easier with #1 as you probably have equipment for that purpose in the ship, just in case you find a suitable planet.

Some of those ideas may be found in Stanislaw Lem's The Invincible novel.

  • 4
    $\begingroup$ Another possible reason for replicating: androids are programmed to look after humans, and without humans their lives are sad and unfulfilled. But you know what looks just like a human? An android! So let's build androids that act like humans and we can be happy again. Then of course the human-like androids will want to make more of themselves until they've colonised the planet, because that's what the humans would have done. $\endgroup$
    – N. Virgo
    May 10, 2022 at 3:53
  • 1
    $\begingroup$ Evolution is not limited to carbon-based life. Eventually one android's brain gets hit by a cosmic ray and it accidentally skips the part of its programming that checks whether the world already has enough androids... $\endgroup$
    – user253751
    May 10, 2022 at 10:57
  • $\begingroup$ Or, some of the androids wander outside of communication radius, and because the androids were designed to work on a small ship in constant communication, those androids are considered destroyed, prompting the construction of more androids, which also promptly wander outside of communication radius... The android that first wandered away is also doing the same thing since it thinks all the OTHER androids are destroyed. $\endgroup$
    – user253751
    May 10, 2022 at 10:58
  • $\begingroup$ Yeah this seems like the simplest answer: duplication is easy because it's part of their normal routine and they have everything they need to do it. Unless that's a major part of the story, he can solve it in a couple of paragraphs. That said, you could make an entire book about androids NOT built for self-replication trying to reinvent technology from scratch before they all fall apart. (There have been some great sci-fi books on humans doing this. See "Lost Regiment" by William Forstchen, but they only had to reach civil war tech...) $\endgroup$
    – JamieB
    May 10, 2022 at 15:40
  • $\begingroup$ This was my immediate first question: why oh why would they even WANT to replicate? And it's not logical for humans to program them that way. At all. Frankenstein $\endgroup$ May 11, 2022 at 8:44

The fixing android finds lots of things to fix.

The planet is lifeless. It was not always so. There are many, many old machines on this planet. Your fixing android is driven to fix broken machines. It does not exactly know how these machines originally worked or what they were supposed to do. But it must fix, and so it fixes them to be like the machines it is familiar with. It copies minds and programs from the other androids and from itself into these alien machines it has found.

Some of these fixed alien machines turn out to be basically the same androids from the crashed ships, in strange bodies. Others are different and some are very different, retaining more of their original functions.


The dead human bodies are invaluable

If Marvin and co can get bacteria, or even better, plants that can survive the new planet from them, then they have an incredibly powerful terraforming technology.

Wake up once a year, let the bacteria work, wake up, etc. Any that photosynthesise or secrete anything potentially flammable, help by spreading them around.


I think there's a huge chunk of the story that can be assumed, but missing in your question.

  1. We are talking about a "fully automated exploration vessle.

1.1) Exploration - presumes that sensors to find valuable ores were onboard. Likewise, scout drones spectrometers and other exploration systems should have been onboard. They may be broken, but can be fixed.

1.2) Fully automated presumes that there were systems that there were systems onboard for repairing and maintaining the vessel and equipment in operation as well as sources of energy (wind, solar etc) to sustain autonomy.

  1. hybernating crew now dead.

2.1) what would be the purpose of populating an exploration vessle with a hybernating crew? Colonists? In that case, the ship would also be stocked with water and sustainable nuitrition (plants, maybe animals, worst case - colonies of edible bacteria and/or algae)

2.2) if the crew were indeed colonists, then there would be eqipment to terraform or at least build air-tight colonial structures

2.3) technology must have been onboard to provide the crew with manufacturing and exploration capability

  1. raw materials

3.1) CO2 atmosphere - sounds bleak for humans, but at least we have carbon and oxygen.

3.2) hydrogen makes up an estimated 74% of our universe, so hydgogen will most certainly be available on the planet, the only question is in what form and how to isolate it from the molecules it is bonded to.

3.3) With the three elements (carbon, oxygen, hydrogen) we can synthesize hydrocarbons (fuels and plastics) and we have an oxygen to burn them.

3.4)We also have all the elements required to sustain life - the aforementioned sustainable nutrition for the human crew, or in the very worst case, the bacteria that resides in human intestines. This could be useful for cheap biomass which can be burned for energy.

3.5) with a bit of genetic modification, which could be achieved using the more primitive selective breeding (albeit longer), bacteria can be used to produce specific molecules.

  1. conductors and semiconductors.

4.1) currently (in 2022) we mostly use silicon semiconductors, but other materials with semiconductor properties exist, the only question is whoch one the androids will find.

4.2) conductors can be fabricated from carbon (see 3)

  1. manufacturing

5.1) We have some form of energy to get started, once we get fuel, we can set up a basic smelyer and work our way up from there for metallurgy.

5.2) machining might take longer to get high precision and tolerance, but in a rigorous cycle, this shouldn't take more than 20-30 years.

5.3) microchips usually require chemicals for etching, but in theory this can be replaced witg laser etching, the microchips might initially be bigger, but with time this can be optimized.

By this point we have everything essential for a self replicating android race: materials and manufacturing for structures and electronics.

One additional thing to keep in mind, is with all the empasis on electronics, we humans have largely neglected to exploit biology. This android race might take a reciprocal approach and focus their efforts on exploiting bacteria, plants and animals to fulfill their manufacturing needs.


In order to maintain a starship the andriods must be technically proficient. Is that even up to smelting ore, metallurgy, and manufacturing? It's not beyond the realm of possibility that they would be capable of fabricating replacement parts.

Who knows what type of extrapolations they'll interpret into their functions given enough time. Maybe they'll decide that the best way to maintain their ship is to have replacement copies and the best way to make copies is establish factories and industry.

Eventually they'll wear themselves down - and in order to continue their maintenance function they'll have to maintain themselves. Including, inevitably, copying and replacing themselves. Copying leads to copy errors; mutations, which would bring the android population under the purview of evolution; some errors are fatal, some are benign, some are beneficial.

  • $\begingroup$ "Copying leads to copy errors" - not in digital copies it doesn't. That's what hashing and error correction is for. You can't do digital evolution this way, you need specially designed algorithms to control the parameters of your iterations. $\endgroup$
    – Corey
    May 12, 2022 at 22:56

The planet is not a good match for human conization and yet humans went there with bots. Thr bots probably have terraforming among their tasks.

A handful of robots cannot terraform a planet. They would take longer than the planet will exist and if the planet has an atmosphere and is geologically active it may undo the bots' work. So the machines must be able to self-replicate with wbatever they may find, by design.

Suggested literature: Horizon: New Dawn. That is kinda the plot of the game (with a huge twist).


The androids are still on a mission of exploration. The ship is crashed and the humans dead. But they know there are more humans that funded this exploration, and these humans will want the androids to make the best of the situation.

And they have found a complete planet, which of course need to be fully explored! So they need to set up a factory and make enough androids to map the entire planet, discover all resources/deposits and so on. A large project, but perhaps they know they are not likely to be rediscovered in several millennia anyway.

Part of the android job is to make life easier for the humans. As they become many and know the available resources, they may decide to build a proper spaceport so future humans can land safer than they did. And a town for visiting humans to live in, with all the stuff humans usually wants. Eventually, they may be able to build a small messenger ship to tell humanity what happened.

Now, maybe they never get as far as building that ship. Further disasters may destroy the remains of the original ship, along with its database of human knowledge. A planet full of androids with no clear purpose – they may indeed become a "machine race" with only a vague memory of having arrived there somehow. Factories make all they need, and they work to keep the factories going, and feed them with raw materials.


Try and find some parallels between human advancement and this android species, then work off those parallels to create something new. You could make it so the androids created tools and began to mine downward in hopes of oil. If you wanna go for something more grim, make it so that android parts are recycled when one of them dies. They could learn medical robotics and rebuild each other when they get damaged. Later on, they could take parts of the ship and create entirely new androids, and then creating a factory to mass-produce them.


. Bleak, but the human crew wouldn't have survived for long anyway.


I also believe the humans would have separate eject-able chamber and eject before the crash and land. The androids wouldn't need air so large parts of the ship should be depressurized. This chamber would have independent everything for any emergency just like this.

Most human space exploration assumes that they will have to live on a lifeless planet. This means an enclosed dome, and that means the CO2 or whatever is pumped out and oxygen and etc can be put in its place.

I believe the answer is YES.

A fully automated repair ship should have ALL the equipment it needs.

This is going to heavily depend on the ships technologies and the planets raw materials.

The fact that your people are space faring race means their at least 100 years beyond our technology. Also the fact AI robots are so common.

We have pretty good 3d printing, certainly good enough for the chassis of the robots. In a 100 years it should be trivial for them to 3d print most of their bodies.

The actually semi conductors might require specialized equipment, but it too can be built/repaired with enough raw material and time.

Want them to survive give them enough of the right kind of materials. Wants them to thrive, give them a planet with abundance.

Maybe a deploy able drone swarm for outer space to collect raw materials, and send them down to the planet.

The automated space craft is literally going to have to collect and refine its own metals in outer space so it should be equally capable of ground survival.

Even then the ship should be able to send out an automated SOS distress call even before it crashes.

Now if they posses the ability to convert matter to energy they will survive. They have a huge planet, and if necessary the can consume all of it.

Refine metals and etc can ALL be done with electricity.

Bottom line if they have or can produce enough of the correct atoms or molecules they WILL survive and replicate.

If you wish them to survive then you the OP just give them a planet with easily mine able resources.

Want to make it easy? Oh look a giant iron and carbon deposit right on the surface next to the ship, and PRESTO they can make steel.

Want to make it hard or slow? The mineral deposits are either 10 miles apart or underground.


You are going to go down a deep rabbit hole on AI here. And ultimately have to define exactly What type of AI, AGI, Narrow? General? Super? What were the limitations humans put on them if any? IF they can improve or diversify?

Do they even have the capacity to do anything other than the purpose they were built for? I mean, if a droids function was to dig wells. Can It make the jump of comprehension that they don't even needs wells now? Does it instead dig for iron or other natural resources they are going to need for repairs etc? Or does it spend eternity or the rest of its rechargeable battery life digging holes all over the planet? Or just sit there waiting for someone to tell him where water is?

For your answer to be Yes, there will have to be some directing intelligence. maybe an AGI built into the colony ship that survived.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .