At 80,000 Hours, we think it’s really important to find the causes in which you can make the most difference. One important consideration in evaluating causes is how much we should care about their impact on future generations.
Important new research by a trustee of CEA (our parent charity) Nick Beckstead, argues that the impact on the long-term direction of civilization is likely to be the most important consideration in working out the importance of a cause.
Briefly: Barring a disaster, the Earth will remain habitable for about a billion years, and many millions of future generations could come after us. Most of humanity’s potential lies in the future. There are actions we might take today that could have a significant impact on these future generations. For instance, if we cause a nuclear war that ends civilization, then this future will never happen at all. So, from the perspective of making a difference, the expected impact of our actions today on the far future might be the most important thing about them.
In the rest of this post, we summarise the argument and its consequences, concluding:
The argument seems plausible, so it’s plausible that the expected impact of different causes on future civilization is the most important thing about them.
We know little about the effects of different causes on the long term future
Thus, if the argument is plausible, it’s very unclear which cause is most effective
Since prioritization research that considers the far future is also extremely neglected, we think the argument implies it’s highly important to invest more in research to prioritise among causes (which we argue in a future post)
We’ll also outline in a future post our current best guesses at how best to help the far future
Why is the far future important?
Here’s the argument stated more formally:
The future of civilization might be extremely long, which means its value might be extremely great (in comparison with the present)
Some of our actions today could shape the path that civilization takes in not-ridiculously-small ways
Due to (1), the importance of long-term changes to the path of civilization outweigh merely short term changes
So, from the perspective of making a difference, what matters most about our actions today is their effect on the general path that civilization takes
Let’s go over these points, starting with (1). If civilization isn’t ended by a disaster, how long could it go on for? Our best estimates from astronomy say that the Earth will remain habitable for around a billion years. In addition, it’s likely there are other planets that will last much longer. If our descendants can ever colonize these planets, then civilization could last hundreds of trillions of years. Our civilization has only lasted several thousand years so far, so its future potential is vastly larger than our history to date.
Whether we ever fulfill this potential is extremely uncertain, but we think its value would be so great that it’s worth caring about even if there’s only a small chance of it coming about.
Imagine an asteroid hit the Earth in Roman times and ended civilization. In Roman times, there were around 100 million people living on the Earth. How bad would that event have been? It seems to us that the impact would have been much worse than 100 million deaths. It would have also included the loss of two thousand years of civilization from then until today, and probably beyond. Similarly, if civilization ended today, the cost wouldn’t just be 7 billion deaths, but it would also be the chance of an extremely long and flourishing future, and we think that would be really bad. So, we think the far future matters.
This is in part a value judgement, which people can reasonably disagree about (see some of the responses in the next section.) On balance, however, we think the reasons in favor of caring about future generations outweigh those against.
Looking now at (2), our actions can have consequences that ripple forward and change this future. I’ve already mentioned one trivial example. If we caused a disaster that ended civilization today, then this future would never happen. But there are other ways we might shape the future. For instance, if we pass down certain values to our descendants, then it might move civilization onto a slightly better path, which could alter what happens to our descendants over millions, billions or trillions of years.
The value of actions that ripple forward is likely to be many, many times larger than the value of actions that only affect the present. This comes out more clearly if we look at some very rough numbers. If there’s a 1% chance humanity survives the next billion years, and a 1% chance of then colonizing other planets, we can expect civilization to last for 10 billion years (and this seems conservative). One way we can change the path of future civilization is by reducing the risk of the end of civilization. There seems to be some actions we can take today that would reduce this risk by more than a 1 in a million chance; for instance, tracking more asteroids, improving disease surveillance programs, reducing greenhouse emissions and strengthening global governance. If that’s true, then the expected value of these actions is adding an expected 10,000 years to civilization. We think 10,000 years of future civilization is worth far more than even very large improvements over the next couple of decades. So, this example of a path change matters much more than the realistic near term impacts of our actions.
Why might this argument fail?
There are plenty of ways to reject this argument. First, a clarification. The argument only concerns what we should do insofar as we care about making a difference. Other factors can matter morally. For instance, we might think that we have a greater obligation to present generations than future generations. I think, however, that everyone should care about making a difference to some extent. As Rawls said, “All ethical doctrines worth our attention take consequences into account in judging rightness.” And since in this case the stakes are so high – there’s a potentially huge future ahead – what happens to the far future should matter whatever your general moral beliefs.
In addition, you can object in three ways: 1. Morally, we can’t do good to people who don’t exist yet – this is called the ‘person-affecting view’ in moral philosophy. 2. People in the future matter much less 3. Theoretical objections to the formalisation of the argument – for instance, you can point out that it looks a bit similar to Pascal’s mugging and argue that accepting this way of reasoning leads to paradox.
You can find responses to these arguments from Chapter 3 onwards in the thesis, which we think are reasonably compelling.
Another response is that the conclusion seems very unintuitive. How could the morally right thing to do be so weird? This makes us cautious in accepting the argument, since we normally seek to bring our views in line with what informed experts would say about the issue. Nevertheless, in this case there are some explanations for why common sense might be wrong, as explored in the second chapter of the thesis.
Finally, note that we don’t have to be certain that this argument is valid in order for it to matter a great deal. Both Nick and we are uncertain that the argument holds, but because if it does hold it’s extremely important, even if we just think there’s a small chance that it holds, it still matters.
What about speed-ups?
There’s an important type of near-term benefit that is useful to treat separately from the others, which slightly complicates the situation. Our actions can also speed-up civilization’s progress along a path. In total, this gives us three ways to categorise the effects of actions:
Near term benefits
Speed-ups – increasing the rate at which we move along a path
Path changes – the extent to which the action puts the future of civilization on a different path. These sub-divide into broad and targeted ways to cause path changes.
We think that speed-ups are likely to matter far less than path changes. If everything is speeded up at the same rate, the impact consists of bringing forward future civilization. However, it seems very hard to advance progress by more than a couple of decades. So, the value of these speed-ups is normally going to be much less than the value of improving the path of future civilization, which we have shown could plausibly have a value of over tens of thousands of years of future civilization. This is also covered in more detail in the thesis.
Note that in practice most speed-ups to general development are also likely to cause path changes. For instance, increasing economic development could cause a path change because it might increase (or decrease) the chance of developing a dangerous technology faster (or slower) than our ability to mitigate the risks of technology. A mere speed-up is relatively unimportant, but in practice speed-ups could cause important path changes.
What does this mean for people looking to make a difference?
What about our current attempts to do good?
The argument implies that path changes matter far more than the near term benefits of our actions. If the argument holds, what does this imply about our current attempts to do good?
Most attempts at making a difference today seem to be aimed at having near term benefits. If the effect of our actions on the far future is what matters most, then our current attempts might be misguided.
On the other hand, near term benefits have ripple effects that could cause path changes. For instance, if you save a child of malaria, then that child has a chance to receive an education and work, which helps the community, who have further descendants, who might contribute to broader progress and so on.
Nick makes a helpful distinction between broad and targeted ways of trying to improve the future. Broad interventions are things like improving education. They improve our general capacity to deal with future problems. Targeted interventions involve working on specific risks to the future, e.g. trying to reduce the risk of an asteroid strike. In between lies a spectrum. An intervention in the middle could be something like ‘teach our children to be good stewards of the future.’ Broad interventions are promising because they don’t depend on the details of what happens in the future, which are highly unpredictable. Targeted interventions are promising because they are more focused and often more neglected.
If the best ways to help the future turn out to be broad, then our current attempts at making a difference, which are often focused on near term benefits which have positive ripple effects, might turn out to be pretty good. In this way, common sense beliefs about the best ways to do good could turn out to be largely right, even if the reasons behind them were wrong. On the other hand, even if this turns out to be true, there would likely still be room for improvement to common sense ways to doing good. This is because some interventions with large near term benefits will likely have much stronger ripple effects than others, it seems like it might be possible to identify these, and very little effort has been put into identifying them so far.
On the other hand, if the best interventions turn out to be relatively targeted, then our current attempts at doing good are likely to be highly sub-optimal.
What should we do going forward?
Some in the effective altruism community have previously thought that targeted interventions are likely to be best at helping the far future, in particular interventions that aim to directly reduce the risk of extinction. This position has been advanced by Nick Bostrom and MIRI.
In this new research, Nick (Beckstead) points out that Bostrom never argues that reducing the risk of extinction is more important than other forms of path change i.e. making the whole of the future a little bit better. Which is better is currently unsettled. If we care about all types of path change, then it becomes pretty unclear whether targeted interventions are better than broad. Moreover, very little work has been done on this question. (See a summary of this argument here)
Our take on the situation is that since the degree of uncertainty is very large, there’s a huge amount at stake. Since very little work has been done on these questions, the most important thing to do now is more prioritisation research. This research would examine (i) whether the argument for focusing on the far future is correct (ii) if it is, then what are the best ways to help the far future? Even without the far future consideration, we think further prioritisation research is highly important, so this extra consideration just adds to the case. 80,000 Hours is preparing a report on why we think prioritisation is one of the most important causes.
In terms of our own research, we want to investigate how people can best contribute to prioritisation research in their careers. For evaluating other causes, we’ll aim to identify causes that seem plausibly good for the far future and look good by ordinary standards, and flag both, then leave it up to our members to decide how much weight they want to put on the far future.
If forced to guess now, what look like the best ways to help the far future? We’re also preparing a report on that, which will be out shortly.