How could a civilization create a dangerous autonomous mechanical ecosystem that will not be overly dangerous to its creators?

Militaries are expensive; military research doubly so. So what's a small colony from a small civilization supposed to do to defend itself? It needs all the resources it can get to bootstrap its own industrial base, so it can't create much more than a token defensive force unless it wants to cripple its own growth. Nanofabrication and robotic mining can help, but everything you produce is going to be inferior to the high-end military technology of the people who might invade you.

So why not evolve new threats? Organisms can reproduce and evolve without your direct involvement, crating new and completely unexpected threats. It's hard to engage a conventional military when strange monsters are swarming your position. Biological organisms probably won't do: they're too vulnerable and unsuited for war. Claws and teeth have limited use against combat armor. You need something that can't be easily handled with fire or chemical weapons, and can do real damage.

So let's make the organisms artificial. A mechanical ecosystem of autonomous (https://en.wikipedia.org/wiki/Self-replicating_machine) robots that uses evolutionary robotics to become ever-better killing machines.

Now we're getting somewhere: armor-piercing guns and diamond-toothed saws beat keratin shards and teeth any day, and robots evolve much faster and with more purpose than organic life ever could. Programmed evolutionary imperatives and and behaviors mean that this 'mechology' can devote itself to the destruction of intruders without endangering the colonists or engaging in wasteful perpetual war against itself.

A simple ecology will suffice to start. Plant-equivalent robots mine for materials and convert them into refined forms and power sources. Animal-equivalent robots prey on them and each other. Their AIs are simple neural networks. Their genetically-specified controllers ensure that they are evaluated based on military suitability, and ensure that the least fit don't get to breed. All are designed with an artificial version of chromosomes which are merged and mutated during reproduction. The resulting template is then assembled into a robot and set loose. All elements of the robots are subject to mutation, to ensure that preconceived notions don't limit their evolution too much. The only thing that's immutable is instructions not to attack colonists or their infrastructure.

Now comes the tough part: making sure it stays safe to you. Sure, you can enforce behaviors and try to direct the evolution as best you can, but evolution has a nasty habit of doing the truly unexpected, so sooner or later you'll discover that your safeguards aren't as safe as you think. And while an uncontrolled militarized mechology that indiscriminately kills everything in sight is great if you want to make an uninhabitable death world, it's significantly less great if you're trying to live there.

So what can you do to prevent this dangerous mechanical ecosystem from evolving out of your control and becoming a danger to you? How can you create safeguards that are strong enough to not be subverted while still not being intrusive enough to force your new army to evolve predictable traits and tactics? Or is there no way to reduce the risk to an acceptable level?

  • 1
    $\begingroup$ If you've seen some of the insanely clever things single celled organisms do with their mechanical linkages (proteins), you'd probably argue that they are already militarized. You just don't have their unwavering support! $\endgroup$
    – Cort Ammon
    Commented Dec 19, 2016 at 15:20
  • $\begingroup$ After reading the bolded title I immediately thought of Mass Effect, in which the main enemies were created in that way $\endgroup$
    – Zxyrra
    Commented Dec 20, 2016 at 0:29

2 Answers 2


It will come down to how to decide 'Friend or Foe?'

After handwaving away the complexities of evolving machines and the supply chain required to enable the kind of evolution the OP describes, you're still stuck with a fundamental and difficult question "How do you figure out who's a bad guy?"

On Earth, this problem has been "solved" a number of different ways. Most applicable to spaceships are the Friend or Foe Identification systems used by militaries and civil air traffic control to identify friendly ships. Note that:

Despite the name, IFF can only positively identify friendly targets, not hostile ones.

This makes sense since you know that you can trust ships that have strongly authenticated themselves as friendly. A ship that does not or cannot identify themselves as friendly is either neutral or an enemy.

Fire Control Policy

Not Blue, Shoot It (NBSI) and Not Red Don't Shoot (NRDS) are two approaches to handling neutral or enemy contacts. For NBSI, any unknown ship is considered hostile and will be fired upon. NRDS takes a more generous approach to neutrals by allowing them free passage as long as they take no hostile actions. Which policy you pick will depend on the capabilities and paranoia of the star base occupants.

Making Evolution Safe(r)

You can't ever be absolutely sure that this weapons system is capital-s Safe. Evolution is a long series of adaptation to pressure. Many of the pressures on this weapons system will be beyond the colonist's control such as local resource constraints, frequency of attacks, weapons used in attacks, tactics used in attacks, attempts at spoofing colonist identification measures, etc, etc.

Further, there may be any number of system characteristics that are beneficial at small scales that completely fail at large scales or cause er, unhelpful feedback loops. For example, it's handy to have the system mine its own resources and this works as long as the material mined doesn't impinge on the colony's general economy. However, what happens when the military grows to consume a significant portion of available mining capacity and crowds out the colonists ability to make money off mining?

Making this safe for the colonists

Depending on the hostility policy, the requirements to make this mechanical immune system safe for colonists will depend on proper identification of colonists. It's up to the author to determine which mechanisms to use.


You're trying to treat machines as you would living things, but they're not. Machines require maintenance depots, spare parts, and trained technicians, not to mention fuel and ammunition.

Furthermore, machines are built, not birthed. That implies having all the advanced raw materials to do so on hand. Nano bots are awesome, but they still can't turn dirt into a tank.

And so, if your technology is able to overcome all that anyway (aka you're hand-waving it), then it's relatively trivial to assume that the bots are also programming their creations to stay far away from your faction/species.


You must log in to answer this question.

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