People often argue that we should focus on shaping the long-term future because the number of individuals in (possible) future populations is much larger if space colonisation happens. So the impact on the long-term future dominates in expected value.
I think this argument is too simple. In a larger world, your actions can indeed affect more individuals. However, it also means that a larger number of agents tries to shape the world, resulting in a dilution of your influence.1 Prima facie, it seems plausible that the two effects cancel each other out, so it’s not clear whether you can have more impact in a larger world.
As an analogy, it is not clear whether it’s more impactful to try and influence government policy in a single state or on the federal level. The latter potentially affects more individuals, but your marginal influence over federal policy is (proportionately?) smaller.
In light of this, the mere fact that the future is (very) large is not, as is often suggested, enough to establish that we should focus on improving the long-term future. (For purposes of this discussion, I assume that future individuals matter equally.)
This can be turned into an argument against longtermism. Suppose we buy into this theoretical reason why the marginal impact of short-term and long-term interventions is prima facie of comparable magnitude. But then we should factor in that it is hard to accurately predict and influence the long-term future, which is generally considered a strong argument against focusing on the long-term; this, so one could argue, tips the balance (so the argument goes) in favor of short-term interventions.
Of course, it’s not that simple. This argument operates on a very abstract level, and there is a plethora of additional considerations.
First, the future is big in both time and space. The above argument can in principle also be applied to the temporal dimension, but this raises complicated questions about how power is shared between generations. The crux for longtermism, then, is whether or not we have reason to believe that our generation is in a good position to influence the long-run future, i.e. that it has more power (per person) than average.
Some people believe that our time is special, e.g. because an AI takeoff will likely happen soon, or because we live through a “Time of Perils” of unusually high extinction. I am sceptical about such claims (see e.g. 1,2,3), and I’m fairly agnostic about whether 21st century people are more or less influential compared to the 20th or 22nd.2
However, I still think that contemporary people are plausibly far more influential than a model of “all generations are born equal” would suggest. This is because almost all agents will live in the future (assuming that space colonisation will happen); so they are our successors, which means that we can influence them but they can’t influence us. (Another way to put it is that our generation constitutes a population bottleneck relative to a future spacefaring civilisation.)
Even so, as long as the values and power structure of future generations will drift over time, our potential impact is limited. We’d be able to have a vastly larger impact on the long-term future if civilisation reaches a steady state at some point, after which values and their relative influence are locked in. It’s not clear whether this will happen, but it is at least plausible, and I think that’s enough for us to have (in expectation) significant influence over the long-term future. (That said, I agree with Robin Hanson that value drift is unavoidable for the foreseeable future.)
All told, I think the case for longtermism is still good enough. However, it does require additional arguments about a) the feasibility of a steady state, and b) current humans being a population bottleneck. It’s not a no-brainer just because the numbers are so large.
Footnotes
- It can be the case that there are more agents but all agents still have all the impact. E.g. in a majority vote your (expected) impact does depend only on how close the vote is (in absolute number), not on how many people are voting. However, for purposes of this discussion, I will breacket this complication.
- I’m bracketing some complications like the fact that person-years or subjective time would arguably be a better framework than calendar time.