Note 1: This is an old post I wrote in 2019, I have moved it here because I find the need to send it to people.
Note 2: I rediscovered this post on the EA forum which is probably where I got most of my inspiration from, the comments are better than this post
TL;DR:In the short term, colonising Mars is a very expensive way of building a secure bunker. Secure bunkers can be made on Earth for a fraction of the cost and with few downsides. Long term, the construction of a self-sustaining colony on Mars is likely a matter of waiting for technology that makes it cheap and easy.
Why consider this?
People (e.g. Elon Musk) often think Mars colonisation is a good route to reducing existential risk. If something bad happens to people on Earth, we can "reboot" humanity from Mars instead of going extinct.
This would mitigate risks that targeted Earth specifically, like an asteroid strike or nuclear war. It would most likely mitigate pandemic risk. It would not mitigate risks from malicious AI, or just evil humans on Earth that come to Mars.
I think that in the short term this is a bad plan. To see why, I'll go through a couple of scenarios that I think people conflate when talking about Mars colonisation (for x-risk reduction).
We could conceivably just ignore the fact that we are on Mars and think of "colonisation" as an expensive bunker in an extremely secure location. The goal would be to maintain a human population until humans could return to Earth after humanity is wiped out or much reduced there.
In this strategy, we would ship finished materials and people to Mars to build a small colony. We would continue shipping this stuff until there were enough supplies to last ~50 years and then just send maintenance. We would not be trying to build an industrial base, or go much beyond a single group of people.
The colony could have some facilities for local production of useful stuff (e.g. water extraction and cracking for oxygen, greenhouses for plants, energy via solar panels) - but self-suffiency would never be the goal. Instead people would get by on regular shipments from Earth, plus stored supplies of finished goods in the event of an interruption in shipments.
One would need enough humans to produce a breeding population on Earth. This could be accomplished by maintaining the required population on Mars, but this would require a lot of people to be safe. Estimates of a minimum viable population go from roughly 160 people to over 1000. With frozen sperm/eggs/embyros, the number would likely go a lot lower - so that would probably be the best route.
The main advantage of this approach is that it would be significantly cheaper than a self-sustaining colony and much easier to setup with present day technology.
Also, after most disasters, Martians could return to Earth and not _need_ present day technology to survive. Regressing to a lower level of development would be acceptable and expected, humanity could rebuild in an easy climate to which we are adapted.
The disadvantage is that it doesn't protect against long term catastrophes on Earth - for example an asteroid strike that destroyed the biosphere for a long period of time would make it harder to return within a reasonable timeframe. A single small bunker is also a single point of failure.
Here, we would be pursuing a completely self-sufficient Mars. The aim would be for Martians to be able to colonise the universe without ever going back to Earth.
In comparison to the bunker strategy, we would need to ship much larger numbers of people/materials there, and then use those to build a much larger colony (or colonies). These colonists would have to develop an industrial base that was capable of running and growing until it could support further space travel. We would need to ensure that a colony had all the specialised skills and material production facilities it would need to support continued growth. We cannot allow technology to regress much past the present day, as we could in a bunker, due to the harsh Martian climate and lack of atmosphere requiring advanced technology to survive.
This approach is extremely expensive to achieve with near-future technology - transporting the required tens of thousands (or even millions) of people is just the beginning. You need to transport most of the required infrastructure as finished goods and setup the industries you need to produce all manner of basic goods (electricity, oxygen, food and water for starters). You also need to produce finished goods (from paint to pencils to semiconductors) of an extraordinary variety. You need to set this all up to be able to run without input from Earth's supply chains, which are notoriously vast.
With advanced technology it is relatively easy - self-replicating robotic factories could produce and maintain the required infrastructure rapidly and cheaply, so the cost would mostly be in shipping humans and an initial factory. But with near-term technology, using humans as the main source of labour, things would have to proceed much slower. To have the best chance of long term survival, we might even need to terraform the planet, which is most likely another few orders of magnitude in cost.
This approach defends against the majority of threats that could affect Earth, and doesn't rely on Earth being habitable after the threat has passed.
Stay on Earth for now
A self sustaining colony in the near future seems like a non starter to me due to the ruinous expense.
What about a bunker? It is more feasible, but less useful. Earth based bunkers (under mountains or down mine shafts, in remote locations, under the sea) seem vastly easier to build, maintain and return from. They are less secure, as wars on Earth could threaten them and threats to Earth as a whole (e.g. asteroids) could damage them. However, this doesn't seem like a vast problem to me - go deep enough, keep the location secret, don't rotate people often (to guard against pandemics). The bunkers could even be somewhat self-funding - but in any case they would be so much cheaper than Martian bunkers that we would be able to build many more of them.
If the economics of space travel and construction change, a bunker on Mars might be worth it. But as it is, I think it's strictly dominated by Earth based bunkers.
Possibly someone should continue looking into and trying this though. If it turned out to be cheap, it might be a good fit in our portfolio of options to reduce the risk of extinction.