Forethought just launched one a few hours ago!
Neither here nor there, but while we’re counting possible biases, it may also be worth considering the possibilities that
people who conclude that farm animals’ lives are good may select into farming, and people who conclude that they’re bad may select out, making farmers “more optimistic than others” even before the self-serving bias; and, pointing the other way,
people who enter animal advocacy on grounds other than total utilitarianism could then have some bias against concluding that farm animals have lives above hedonic zero, since it could render their past moral efforts counterproductive (and maybe even kind of embarrassing).
Thanks so much for putting this together! I hadn’t thought of the cross-price elasticity effects across types of animal products, but of course it’s an important thing to incorporate.
Two extensions of this sort of analysis that I would be interested to see:
Are there any important cross-price elasticity effects between animal and non-animal (including non-food) products? For instance, if the worst type of meat is beef, as you estimate, then it could be good to buy products that use the same inputs as beef—a type of grain that grows best on the types of land suitable for cattle, say—because that will push up the price of beef and push people into less harmful meat products. (It makes sense that cross-price elasticity effects would tend to be largest within kinds of meat, but other products may still worth considering, if this hasn’t already been done.)
Just as the substitution effects across kinds of meat are presumably stronger than between meat and other things, the effects are presumably strongest within brands of a particular animal product. That is, maybe buying (less in-)humanely raised chicken or environmentally (less un-)friendly beef pushes up the price of that product in general, which causes people to consume less of it, leading to an improvement overall, even though the purchased product itself still does net damage. How much would these within-product considerations change things?
Obviously there’s no end to the possible extensions, until we have a complete model of the entire economy that lets us estimate the general equilibrium impact of switching from one product to another. But maybe there are a few more elasticities that would be relatively important and tractable to consider.
This all strikes me as a good argument against putting much stock in the particular application I sketch out; maybe preventing a near-term nuclear war doesn’t actually bode so badly for the subsequent future, because “human nature” is so malleable.
Just to be clear, though: I only brought up that example in order to illustrate the more general point about the conditional value of the future potentially depending on whether we have marginally averted some x-risk. The dependency could be mediated by one’s beliefs about human psychology, but it could also be mediated by one’s beliefs about technological development or many other things.
Just to be clear: my rough simplification of the “Pinker hypothesis” isn’t that people have an all-around-peaceful psychology. It is, as you say, a hypothesis about how far we expect recent trends toward peace to continue. And in particular, it’s the hypothesis that there’s no hard lower bound to the “violence level” we can reach, so that, as we make technological and social progress, we will ultimately approach a state of being perfectly peaceful. The alternative hypothesis I’m contrasting this with is a future in which can we can only ever get things down to, say, one world war per century. If the former hypothesis isn’t actually Pinker’s, then my sincere apologies! But I really just mean to outline two hypotheses one might be uncertain between, in order to illustrate the qualitative point about the conditional value of the future.
That said, I certainly agree that moral circle expansion seems like a good thing to do, for making the world better conditional on survival, without running the risk of “saving a bad world”. And I’m excited by Sentience’s work on it. Also, I think it might have the benefit of lowering x-risk in the long run (if it really succeeds, we’ll have fewer wars and such). And, come to think of it, it has the nice feature that, since it will only lower x-risk if it succeeds in other ways, it disproportionately saves “good worlds” in the end.
About the two objections: What I’m saying is that, as far as I can tell, the first common longtermist objection to working on x-risk reduction is that it’s actually bad, because future human civilization is of negative expected value. The second is that, even if it is good to reduce x-risk, the resources spent doing that could better be used to effect a trajectory change. Perhaps the resources needed to reduce x-risk by (say) 0.001% could instead improve the future by (say) 0.002% conditional on survival.
About the decision theory thing: You might think (a) that the act of saving the world will in expectation cause more harm than good, in some context, but also (b) that, upon observing yourself engaged in the x-risk-reduction act, you would learn something about the world which correlates positively with your subjective expectation of the value of the future conditional on survival. In such cases, EDT would recommend the act, but CDT would not. If you’re familiar with this decision theory stuff, this is just a generic application of it; there’s nothing too profound going on here.
About the main thing: It sounds like you’re pointing out that stocking bunkers full of canned beans, say, would “save the world” only after most of it has already been bombed to pieces, and in that event the subsequent future couldn’t be expected to go so well anyway. This is definitely an example of the point I’m trying to make—it’s an extreme case of “the expected value of the future not equaling the expected value of the future conditional on the fact that we marginally averted a given x-risk”—but I don’t think it’s the most general illustration. What I’m saying is that an attempt to save the world even by preventing it from being bombed to pieces doesn’t do as much good as you might think, because your prevention effort only saves the world if it turns that there would have been the nuclear disaster but for your efforts. If it turns out (even assuming that we will never find out) that your effort is what saved us all from nuclear annihilation, that means we probably live in a world that is more prone to nuclear annihilation than we otherwise would have thought. And that, in turn, doesn’t bode well for the future.
Does any of that make things clearer?
As long as any of NTI’s effort is directed against intentional catastrophes, they’re still saving violent-psychology worlds disproportionately, so in principle this could swing the balance. That said, good point: much of their work should reduce the risk of accidental catastrophes as well, so maybe there’s not actually much difference between NTI and asteroid deflection.
(I won’t take a stand here about what counts as evidence for what, for fear that this will turn into a big decision theory debate :) )
I was just saying that, thankfully, I don’t think our decision problem is wrecked by the negative infinity cases, or the cases in which there are infinite amounts of positive and negative value. If it were, though, then okay—I’m not sure what the right response would be, but your approach of excluding everything from analysis but the “positive infinity only” cases (and not letting multiple infinities count for more) seems as reasonable as any, I suppose.
Within that framework, sure, having a few thousand believers in each religion would be better than having none. (It’s also better than having everyone believe in whichever religion seems most likely, of course.) I was just taking issue with “it might be best to encourage as many people as possible to adopt some form of religious belief to maximise our chances”.
Still it might be best to encourage as many people as possible to adopt some form of religious belief to maximise our chances.
I’m very sympathetic to the idea that all we ought to be doing is to maximize the probability we achieve an infinite amount of value. And I’m also sympathetic to religion as a possible action plan there; the argument does not warrant the “incredulous stares” it typically gets in EA. But I don’t think it’s as simple as the above quote, for at least two reasons.
First, religious belief broadly specified could more often create infinite amounts of disvalue than infinite amounts of value, from a religious perspective. Consider for example the scenario in which non-believers get nothing, believers in the true god get plus infinity, and believers in false gods get minus infinity. Introducing negative infinities does wreck the analysis if we insist on maximizing expected utility, as Hajek points out, but not if we switch from EU to a decision theory based on stochastic dominance.
Second, and I think more importantly, religiosity might lower the probability of achieving infinite amounts of value in other ways. Belief in an imminent Second Coming, for instance, might lower the probability that we manage to create a civilization that lasts forever (and manages to permanently abolish suffering after a finite period).
Thanks! And cool, I hadn’t thought of that connection, but it makes sense—we want our x-risk reduction “investments” to pay off more in the worlds where they’ll be more valuable.
I agree that it’s totally plausible that, once all the considerations are properly analyzed, we’ll wind up vindicating the existential risk view as a simplification of “maximize utility”. But in the meantime, unless one is very confident or thinks doom is very near, “properly analyze the considerations” strikes me as a better simplification of “maximize utility”.
And if you have any particular ways you think this post still overstates its case, please don’t hesitate to point them out.
My current best guess happens to be that there aren’t great funding opportunities in the “priorities research” space—for a point of reference, GPI is still sitting on cash while it decides which economist(s) to recruit—but that there will be better funding opportunities over the next few years, as the infrastructure gets better set up and as the pipeline of young EA economists starts flowing. For example I’d actually be kind of surprised if there weren’t a “Parfit Institute” (or whatever it might be called), writing policy papers in DC next door to Cato and Heritage and all the rest, in a decade or two. So at the moment I’m just holding out for opportunities like that. But if you have ideas for funding-constrained research right now, let me know!
And sure, I’d love to discuss/comment on that write-up!
Yes, I agree with this wholeheartedly—there are ways for money to be put to use now accelerating the research process, and those might well beat waiting. In fact (as I should have been far clearer about throughout this post!) this whole argument is really just directed at people who are planning to “spend money at some time t to increase welfare as efficiently as possible at time t”.
I’m hoping to write down a few thoughts soon about one might think about discounting if you’ll be spending the money on something else, like research or x-risk reduction. For now I’ll edit the post to make caveats like yours explicit. Thanks.
I think risk-aversion and pure time preference are most likely both at play—I say a few more words about this in my response to Michael above—but yeah, fair enough.
With regard to your second point: I thought I was addressing this objection with,
If this is true, then indeed, we would do less good giving next year than giving this year.
But this one-year relationship must be temporary. Over the course of a long future, the rate of increase in the cost of producing a unit of welfare as efficiently as possible cannot, on average, exceed R. Otherwise, the most efficient way to good would eventually be more costly than one particular way to good—just giving money to ordinary investors for their own consumption. And since the long-run average rate of increase in the cost of welfare is bounded above by R (“5%”), investing at R + RPTP (“7%”) must eventually result in an endowment able to buy more welfare than the endowment we started with.
My point here is that, sure, maybe for farm animals, people in extreme poverty, and so on, the cost of helping them is currently growing more expensive at some rate greater than R (so, >5% per year, if R = 5%). But since the cost of helping a typical stock market investor is only growing more expensive at R (“5% per year”), eventually the curves have to cross. So over the long run, the cheapest way of “buying a unit of welfare” seems to be growing at a rate bounded above by R.
Does that make sense, or am I misunderstanding you?
This is a good point. In general I think the hypothesis that people don’t actually have positive RPTP (in contradiction to the received wisdom from most of the economics literature on this) is the most likely way that my argument fails. In particular, I’m aware of some papers (e.g. Gabaix and Laibson 2017) that argue that what looks like discounting might usually be better explained by the fact that future payoffs just come with more uncertainty.
I currently think the balance of evidence is that people do do “pure discounting”. Defending that would be a long discussion, but at least some evidence (e.g. Clark et al. 2016) suggests that pure impatience is a thing, and explains more of the variation in, for example, retirement saving behavior than risk tolerance does.
In response to your particular argument that if RPTP is a thing it’s weird that financial advisers don’t usually ask about it: I agree, that’s interesting evidence in the other direction. One alternative explanation that comes to mind on that front, though, is that, while advisers don’t ask for the RPTP number explicitly, they do ask questions like “how much do you want to make sure you have by age 65?” whose answers will implicitly incorporate pure time preference.
Thanks! I’ve edited the post to include a link to that article.
That’s right: I agree that there are many other considerations one must weigh in deciding when to give. In this post, I only meant to discuss the RPTP consideration, which I hadn’t seen spelled out explicitly elsewhere. But thanks for pointing out that this was unclear. I’ve weakened the last sentence to emphasize the limited scope of this post.