Researcher at the Center on Long-Term Risk. All opinions my own.
Anthony DiGiovanni
I’m not sure about this, though. As I wrote in a previous comment:
The reasons to do various parochial things, or respect deontological constraints, aren’t like this. They aren’t grounded in something like “this thing out there in the world is horrible, and should be prevented wherever/whenever it is [or whoever causes it]”.
The concern I’ve tried to convey in our discussion so far is: Insofar as our moral reasons for action are grounded in “this thing out there in the world is horrible, and should be prevented wherever/whenever it is [or whoever causes it]”, then shining the spotlight of our active altruism on beings who happen to be salient/near to us is arbitrary. To me, active “altruism” per se[1] is pretty inextricable from anti-arbitrariness.
And I’m saying, suppose for a moment we’re no longer trying to be actively altruistic, and instead consider normative reasons that aren’t grounded in the above. Then, prioritizing those whom you actually have special relationships with isn’t arbitrary in the relevant sense. Because those relationships give you a reason to prioritize them. (Of course, if we started from an impartial altruistic perspective, this reason would be dwarfed by our duty to reduce large-scale suffering overall, insofar as that’s tractable! But the worry is that it’s not.)
Is your position something like, “We also have special relationships with strangers who are near to us”? I might be sympathetic to that, but it seems like it needs more unpacking.
Like I said, I do share the LAR intuition in some limited contexts, and it would be pretty sad if there’s no non-arbitrary way to make sense of active altruism at all. I find this situation unsettling. But I currently feel confused as to how much I honestly endorse LAR.
- ^
As opposed to deontological(-ish) prohibitions against harming strangers.
- ^
(I unfortunately don’t have time to engage with the rest of this comment, just want to clarify the following:)
Indeed, bracketing off “infinite ethics shenanigans” could be seen as an implicit acknowledgment of such a de-facto breakdown or boundary in the practical scope of impartiality.
Sorry this wasn’t clear — I in fact don’t think we’re justified in ignoring infinite ethics. In the footnote you’re quoting, I was simply erring on the side of being generous to the non-clueless view, to make things easier to follow. So my core objection doesn’t reduce to “problems with infinities”, rather I object to ignoring considerations that dominate our impact for no particular reason other than practical expedience. :) (ETA: Which isn’t to say we need to solve infinite ethics to be justified in anything.)
I’ve replied to this in a separate Quick Take. :) (Not sure if you’d disagree with any of what I write, but I found it helpful to clarify my position. Thanks for prompting this!)
Musings on non-consequentialist altruism under deep unawareness
(This is a reply to a comment by Magnus Vinding, which ended up seeming like it was worth a standalone Quick Take.)
From Magnus:
For example, if we walk past a complete stranger who is enduring torment and is in need of urgent help, we would rightly take action to help this person, even if we cannot say whether this action reduces total suffering or otherwise improves the world overall. I think that’s a reasonable practical stance, and I think the spirit of this stance applies to many ways in which we can and do benefit strangers, not just to rare emergencies.
The intuition here seems to be, “trying to actively do good in some restricted domain is morally right (e.g., virtuous), even when we’re not justified in thinking this will have net-positive consequences[1] according to impartial altruism”. Let’s call this intuition Local Altruism is Right (LAR). I’m definitely sympathetic to this. I just think we should be cautious about extending LAR beyond fairly mundane “common sense” cases, especially to longtermist work.
For one, the reason most of us bothered with EA interventions was to do good “on net” in some sense. We weren’t explicitly weighing up all the consequences, of course, but we didn’t think we were literally ignoring some consequences — we took ourselves to be accounting for them with some combination of coarse-grained EV reasoning, heuristics, “symmetry” principles, discounting speculative stuff, etc. So it’s suspiciously convenient if, once we realize that that reason was confused, we still come to the same practical conclusions.
Second, for me the LAR intuition goes away upon reflection unless at least the following hold (caveat in footnote):[2]
The “restricted domain” isn’t too contrived in some sense, rather it’s some natural-seeming category of moral patients or welfare-relevant outcome.
(How we delineate “contrived” vs. “not contrived” is of course rather subjective, which is exactly why I’m suspicious of LAR as an impartial altruistic principle. I’m just taking the intuition on its own terms.)
I’m at least justified in (i) expecting my intervention to do good overall in that domain, and (ii) expecting not to have large off-target effects of indeterminate net sign in domains of similar “speculativeness” (see “implementation robustness”).
(“Speculativeness”, too, is subjective. And while I definitely find it intuitive that our impacts on more robustly foreseeable moral patients are privileged,[3] I haven’t found a satisfying way of making sense of this intuition. But if we want to respect this intuition, condition (ii) seems necessary. If the set of moral patients A seems no less robustly foreseeable than set B, why would I morally privilege B? Cf. my discussion here.)
Some examples:
Trying to reduce farmed animal suffering: Depends on the intervention.
I have the LAR intuition for things like donating to humane invertebrate slaughter research, which doesn’t seem to have large backfire risks on other animals in the near term. (Even if it does plausibly have large backfire risks for future digital minds, say.)
For me LAR is considerably weaker for things like vegan advocacy, which has a lot of ambiguous effects on wild animal suffering (which don’t seem more “speculative”, given the fairly robust-seeming arguments Tomasik has written about).
Trying to prevent digital suffering (even pre-space colonization): All the interventions I’m aware of are so apparently non-implementation-robust that I don’t have the Locally Virtuous intuition here. Example here.
If we instead said something like “this intervention is implementation-robust w.r.t. helping some specific subset of digital minds,” that subset feels contrived, so I still wouldn’t have the intuition in favor of the intervention.
Trying to prevent extinction: (Assuming a non-suffering-focused view for the sake of argument, because lots of EAs seem to think trying to prevent extinction is common-sensically good/right.) As I’ve argued here and here, the interventions EAs have proposed to reduce extinction risk don’t seem to satisfy condition (i) of implementation robustness above. Even if they did, off-target effects on animal suffering arguably undermine condition (ii) (see e.g. this post).
None of which is to say I have a fleshed-out theory! I’m keen to think more about what non-consequentialist altruism under unawareness might look like.
- ^
I mean to include Clifton’s Option 3 as a possible operationalization of “net-positive consequences according to impartial altruism”.
- ^
In the definition of LAR, “trying to actively do good” is the key phrase. I find it pretty intuitive that we don’t need conditions nearly as strong as (1)+(2) below when we’re asking, “Should you refrain from doing [intuitively evil thing]?”
- ^
Maybe the most promising angle is to show that it’s normatively relevant that our beliefs about the more distant moral patients are (qualitatively?) less grounded in good reasons (see Clifton).
- Jun 14, 2025, 3:25 PM; 2 points) 's comment on 4. Why existing approaches to cause prioritization are not robust to unawareness by (
For example, the following passages seem to use these terms as though they must imply consequentialism
I don’t understand why you think this, sorry. “Accounting for all our most significant impacts on all moral patients” doesn’t imply consequentialism. Indeed I’ve deliberately avoiding saying unawareness is a problem for “consequentialists”, precisely because non-consequentialists can still take net consequences across the cosmos to be the reason for their preferred intervention. My target audience practically never appeals to distributive impartiality, or impartial application of deontological principles, when justifying EA interventions (and I would be surprised if many people would use the word “altruism” for either of those things). I suppose I could have said “impartial beneficence”, but that’s not as standard.
Those claims seem to assume that all the alternatives are wholly implausible (including consequentialist views that involve weaker or time-adjusted forms of impartiality). But that would be a very strong claim.
Can you say more why you think it’s very strong? It’s standard within EA to dismiss (e.g.) pure time discounting as deeply morally implausible/arbitrary, and I concur with that near-consensus.[1] (Even if we do allow for views like this, we face the problem that different discount rates will often give opposite verdicts, and it’s arbitrary how much meta-normative weight we put on each discount rate.) And I don’t expect a sizable fraction of my target audience to appeal to the views you mention as the reasons why they work on EA causes. If you think otherwise, I’m curious for pointers to evidence of this.
- ^
Some EAs are sympathetic to discounting in ways that are meant to avoid infinite ethics problems. But I explained in footnote 4 that such views are also vulnerable to cluelessness.
- ^
Some people think we’re entirely clueless, so that we haven’t the faintest clue about which actions will benefit the far future. I disagree with this position for reasons Richard Y Chappell has explained very persuasively. It would be awfully convenient if after learning that the far future has nearly all the expected value in the world, it turned out that this had no significant normative implications.
What do you think about the argument for cluelessness from rejecting precise expected values in the first place (which I partly argue for here)?
Thanks for this Magnus, I have complicated thoughts on this point, hence my late reply! To some extent I’ll punt this to a forthcoming Substack post, but FWIW:
As you know, relieving suffering is profoundly important to me. I’d very much like a way to make sense of this moral impulse in our situation (and I intend to reflect on how to do so).
But it’s very important that the problem isn’t that we don’t know “the single best thing” to do. It’s that if I don’t ignore my effects on far-future (etc.) suffering, I have no particular reason to think I’m “relieving suffering” overall. Rather, I’m plausibly increasing or decreasing suffering elsewhere, quite drastically, and I can’t say these effects cancel out in expectation. (Maybe you’re thinking of “Option 3” in this section? If so, I’m curious where you disagree with my response.)
The reason suffering matters so deeply to me is the nature of suffering itself, regardless of where or when it happens — presumably you’d agree. From that perspective, and given the above, I’m not sure I understand the motivation for your view in your second paragraph. (The reasons to do various parochial things, or respect deontological constraints, aren’t like this. They aren’t grounded in something like “this thing out there in the world is horrible, and should be prevented wherever/whenever it is [or whoever causes it]”.)
I think such a view might also be immune to the problem, depends on the details. But I don’t see any non-ad hoc motivation for it. Why would sentient beings’ interests matter less intrinsically when those beings are more distant or harder to precisely foresee?
(I’m open to the possibility of wagering on the verdicts of this kind of view due to normative uncertainty. But different discount rates might give opposite verdicts. And seems like a subtle question when this wager becomes too Pascalian. Cf. my thoughts here.)
Which empirical beliefs you hold would have to change for this to be the case?
For starters, we’d either need:
all the factors discussed in this section to be much simpler (or otherwise structured in a way that we could model with the requisite precision); or
sufficiently strong evidence that our intuitions can implicitly weigh up such complex factors with the requisite precision.
(Sorry if this is more high-level than you’re asking for. The concrete empirical factors are elaborated in the linked section.)
Re: your claim that “expected effects of actions decrease over time and space”: To me the various mechanisms for potential lock-in within our lifetimes seem not too implausible. So it seems overconfident to have a vanishingly small credence that your action makes the difference between two futures of astronomically different value. See also Mogensen’s examples of mechanisms by which an AMF donation could affect extinction risk. But please let me know if there’s some nuance in the arguments of the posts you linked that I’m not addressing.
Ignoring everything after 1 min, yeah I’d be very confident listening to music is better. :) You could technically still say “what if you’re in a simulation and the simulators severely punish listening to music,” but this seems to be the sort of contrived hypothesis that Occam’s razor can practically rule out (ETA: not sure I endorse this part, I think the footnote is more on-point).[1]
for how long would the effects after the actions have to be taken into account for your indeterminacy to come back?
In principle, we could answer this by:
trying to model the possible consequences that could play out up to some time point T; estimating the net welfare under each hypothesis that falls out of this model; and seeing how much precision we can justify without making arbitrary choices;
checking at which time point T* the UEVs of the options become incomparable, according to these models.
(Of course, that’s intractable in practice, but there’s a non-arbitrary boundary between “comparable” and “incomparable” that falls out of my framework. Just like there’s a non-arbitrary boundary between precise beliefs that would imply positive vs. negative EV. We don’t need to compute T* in order to see that there’s incomparability when we consider T = ∞. (I might be missing the point of your question, though!))
- ^
Or, if that’s false, presumably these weird hypotheses are just as much of a puzzle for precise Bayesianism (or similar) too, if not worse?
That’s right. I don’t expect you to be convinced of this by the end of this post, to be clear (though I also don’t understand why you think it’s “clearly” false).
(“To increase expected impartial welfare” is the key phrase. Given that other normative views we endorse still prohibit option #1 — including simply “my conscience strongly speaks against this” — it seems fine that impartial altruism is silent. (Indeed, that’s pretty unsurprising given how weird a view it is.))
Hi Vasco — your die roll example is helpful for illustrating my point, thanks!
So there is a sense in which I am “unaware or only coarsely aware” of not only “some possible worlds”, but practically all possible worlds. However, I can still precisely estimate the probability of outcomes of rolling a dice.
When you’re predicting the “outcome” of a die roll in the typical sense, all you care about is the number of pips on the upward-facing side of the die. So all the physical variations between the trajectories of the die rolls don’t matter at this level of abstraction. You don’t have coarse awareness here relative to the class of outcomes you care about.
By contrast (as I note in this section), impartial altruists intrinsically care about an extremely detailed set of information: how much welfare is instantiated across all of space and time. With respect to that class of value systems, we do have (severely) coarse awareness of the relevant outcomes. You might think we can precisely estimate the value of these coarse outcomes “better than chance” in some sense (more on this below), but at the part of the post you’re replying to, I’m just making this more fundamental point: “Since we lack access to possible worlds, our precise guesses don’t directly come from our value function , but from some extra model of the hypotheses we’re aware of (and unaware of).” Do you agree with that claim?
still informative, at least negligibly so, as long as they are infinitesimaly better than chance
I recommend checking out the second post, especially these two sections, for why I don’t think this is valid.
That post also spells out my response to your claim “It is obvious for me the 2nd option is much better.” I don’t think anything is “obvious” when making judgments about overall welfare across the cosmos. (ETA: I do think it’s obvious that you shouldn’t do option 1 all-things-considered, because of various moral norms other than impartial altruism that scream against that option.) Interested to hear if there’s something in your claim that that post fails to address.
A glossary of technical terms used throughout the sequence.
An executive summary, besides the intro + sequence summary at the top of this post. (Note that I think the topics here are to some extent intrinsically subtle, and it’s easy to end up misleading people by simplifying them. But maybe there’s a simpler presentation that still avoids this risk.)
An FAQ about various objections to the arguments in the sequence. (I have attempted to explicitly address what seem to me to be the most common objections, to be clear. The idea would be to have these responses be more self-contained, and also link to other writings besides mine.)
Meta: I’m interested to hear if any readers would find either of the following accompanying resources below (not yet written up) useful. Please agree-vote accordingly, perhaps after an initial skim of the sequence. :)
Something like Holden Karnofsky’s approach here (which Anthony DiGiovanni shared with me on a recent post on insect suffering)
(Context for other readers: To be clear, I don’t endorse Karnofsky’s model, which I think is kind of ad hoc and doesn’t address the root problem of arbitrariness in our credences. The least bad epistemic framework for addressing that problem, IMO, is imprecise probabilities (accounting for unawareness).)
Thanks Toby, I’m glad you find this valuable. :)
On how this differs from complex cluelessness, copying a comment from LW:
I think unawareness is a (major) source of what Greaves called complex cluelessness, which is a situation where:
(CC1) We have some reasons to think that the unforeseeable consequences of A1 would
systematically tend to be substantially better than those of A2;
(CC2) We have some reasons to think that the unforeseeable consequences of A2 would
systematically tend to be substantially better than those of A1;
(CC3) It is unclear how to weigh up these reasons against one another.(It’s a bit unclear how “unforeseeable” is defined. In context / in the usual ways people tend to talk about complex cluelessness, I think it’s meant to encompass cases where the problem isn’t unawareness but rather other obstacles to setting precise credences.)
But unawareness itself means “many possible consequences of our actions haven’t even occurred to us in much detail, if at all” (as unpacked in the introduction section). ETA: I think it’s important to conceptually separate this from complex cluelessness, because you might think unawareness is a challenge that demands a response beyond straightforward Bayesianism, even if you disagree that it implies complex cluelessness.
I think of crucial considerations as one important class of things we may be unaware of. But we can also be unaware/coarsely aware of possible causal pathways unfolding from our actions, even conditional on us having figured out all the CCs per se. These pathways could collectively dominate our impact. (That’s what I was gesturing at with the ending of the block-”quote” in the sequence introduction.)
Some of the examples that stood out were those I included in quotes above
I’m still confused, sorry. E.g., “reliably” doesn’t mean “perfectly”, and my hope was that the surrounding context was enough to make clear what I mean. I’m not sure which alternative phrasings you’d recommend (or why you think there’s a risk of misrepresentation of others’ views when I’ve precisely spelled out, e.g., the six “standard approaches” in question).
Sorry, I’m having a hard time understanding why you think this is defensible. One view you might be gesturing at is:
If a given effect is not too remote, then we can model actions A and B’s causal connections to that effect with relatively high precision — enough to justify the claim that A is more/less likely to result in the effect than B.
If the effect is highly remote, we can’t do this. (Or, alternatively, we should treat A and B as precisely equally likely to result in the effect.)
Therefore, we can only systematically make a difference to effects of type (1). So only those effects are practically relevant.
But this reasoning doesn’t seem to hold up for the same reasons I’ve given in my critiques of Option 3 and Symmetry. So I’m not sure what your actual view is yet. Can you please clarify? (Or, if the above is your view, I can try to unpack why my critiques of Option 3 and Symmetry apply just as well here.)