Thoughts on space colonisation

Longtermist EAs tend to believe that space colonisation will become feasible at some point,  which would result in humanity spreading throughout the cosmos. In this post, I’d like to offer some thoughts on how likely this is.

The degree to which humans colonise depends on the costs and the benefits of such an endeavour. So we should distinguish between the ability to spread to the cosmos, and the willingness to do so since one does not imply the other. But before looking at space colonisation, let’s consider other examples of colonisation.

One plausible reference point is the European colonisation of the Americas:

• The cost of a transatlantic voyage was significant but not insurmountable.
• The New World was broadly comparable to the Old World in terms of hospitality and resource endowment.
• Historians generally recognise three motives for colonisation: God, gold, and glory. Ideologies such as colonialism and imperialism also played a role. On an individual level, many Europeans emigrated because they reckoned that they’d have a better chance to make their fortune in the New World. 
• As a result, the Americas were fully colonised. But it is worth noting that this did not happen very quickly and the process was somewhat contingent, with interest in colonisation rising and falling. ¹ Also, some colonies failed. 

Compare this to Antarctica, the last continent on Earth to be discovered (in 1820):

• The cost of travel to Antarctica is similar to the Americas (and significantly lowered by modern airplanes).
• There isn’t much of a reason to colonise Antarctica due to its inhospitable climate. It is perfectly possible to live there, but it’s just not very pleasant. The main reason to go to Antarctica is research.
• As a result, human activity in Antarctica is limited to a few research stations.

It is worth noting that humanity is currently not at all constrained by available land. You’ll find unused land if you start in Central London and go 30 miles in any direction; there’s no need to go to distant continents, let alone outer space. The mere existence of reachable land is no strong reason to colonise it; nor does there currently seem to be a strong drive to expand for its own sake or to maximise human population size. Humanity is, however, constrained by useful space that’s close to valuable resources or existing physical and social infrastructure.² (Colonies, by definition, are not close to existing infrastructure.)

With all this in mind, let’s look at space colonisation in particular. 

The costs of (manned) space travel is currently prohibitive but will come down as technology advances. It seems plausible that future costs will not be prohibitive anymore but still significant, similar to how transatlantic voyages were not prohibitively expensive but also not trivially cheap.

From what we know about space so far, other planets are unlikely to be as hospitable as Earth. Space is like Antarctica, not like America. If anything, Antarctica seems a more attractive colonisation target; it’s just cold but still has the same atmosphere and gravity. (Robin Hanson makes similar points here, here, and here.)

It is possible that such constraints will cease to apply in an age of advanced artificial intelligence or other extremely powerful technology which allows the terraforming of other planets at low costs. However, I suspect that imagining a godlike AI is an overly idealised and abstract way of thinking about the future, and it seems more likely to me that economic and technological constraints will continue to be relevant – albeit less prohibitive than they are now.³

Of course, even if living on other planets is unappealing, people may still venture into space for other motives. Many are fascinated by the idea of exploring outer space, and it is clear that vast quantities of valuable materials can be found in space. But without major social change, it seems unlikely to me that people would be motivated to pay significant costs to set up large-scale space colonies. It seems more plausible that future agents (in particular AGIs) will want to access these vast amounts of matter and energy (which doesn’t necessarily require many people living in space); although it is worth noting that mining and energy production is a small fraction of GDP and arguably not the most important economic constraint at the moment.

One possibility of a social change leading to space colonisation is that future actors will strive to create as many human or nonhuman beings as possible (particularly if those beings enjoy a very good life). This is currently not a common value system, but is endorsed by some effective altruists and might become more widespread. It is also conceivable that the future will be shaped by competitive “Malthusian” dynamics, i.e. increased economic pressure to create as many offspring⁴ as possible. This could mean that humanity is at some point constrained by available land after all.

The following are possible implications for s-risk reduction:

• Mining, space exploration or even small colonies are not s-risks. We are most interested in the possibility of large-scale expansion to space. 
• One possible implication is that we should try to prevent either major social change or Malthusian dynamics that result in increased pressure to colonise space.
• Even if humanity has no strong interest in living in space, people could still use this matter and energy to create more sentient beings (that might experience suffering). Indeed, this is more plausible if there is no counterfactual use for the necessary resources. What matters is the technological capacity to create large amounts of suffering, not colonisation or human population size.
• However, some sources of suffering plausibly scale with human population size. For example, this is true for factory farming: more humans means more meat consumption.  
• Spreading wild animal suffering or even nonbiological “life forms” to space may be much easier and cheaper than colonisation for human life, so this might happen even if the latter is infeasible or unattractive.

1. For example, the number of European emigrants was far higher in the 19th century compared to earlier centuries.
2. This is why rents are high in big cities, particularly near transport hubs or town centres.
3. A detailed discussion of the future of artificial (general) intelligence is beyond the scope of this post. Cf. Summary of my views on AI risk.
4. This doesn’t necessarily refer to biological offspring, it could also be about digital copies of oneself or economically useful “worker robots”.