How Would We Know if We Made a Climate Difference? (Guest Post by Kian Mintz-Woo for Applied Ethics April)

32 Replies to “How Would We Know if We Made a Climate Difference? (Guest Post by Kian Mintz-Woo for Applied Ethics April)”

1. Hi, Kian–Interesting post. But unless I’ve missed something, there’s a false inference here. Suppose I make the following two plausible claims:

   No Difference: Swallowing a single aspirin does not harm a normal healthy adult.
   Group Difference: Swallowing a large bottle of aspirin kills a normal healthy adult.

   Now, imagine the following dialogue:

   MATTHEW: I feel a headache coming on. I’m going to take an aspirin.

   KAIN: Don’t do that! Swallowing a bottle of aspirin would kill you!

   MATTHEW: But I’m not going to swallow the whole bottle. There’s an essential difference between No Difference and Group Difference here.

   KAIN: But there’s no reason to think that there’s anything privileged about *one* aspirin. People often take two. Given your body weight, I even suspect you’d survive taking three. There is no evidence or reason that nature would carve up the results of taking aspirins such that the Difference-making threshold happens to be between the two levels you note.

   Now of course KAIN’s last claim is quite right. But it is also beside the point. Though the boundary between harmless use of aspirin and poisoning is clearly higher than a single aspirin, it remains true that a single aspirin won’t hurt me. It seems to me Kian’s fallen into a simiilar trap.

2. KAIN: Look–Given that you and I have often taken one aspirin before (and doctors often prescribed single aspirins), we have overwhelming evidence that one aspirin is not dangerous. But suppose we had *no* scientific evidence about aspirins. In that case, you might be harmed by just having half an aspirin, or even a quarter.

   MATTHEW: But look. The aspirin comes in convenient single cylinders. Surely one is fine. Otherwise, they wouldn’t be made like this.

   KAIN: Okay, I’ll grant you that in the case of an aspirin. We know someone made it, and it’s unlikely they would have packaged them in harmful quantities.

   But climate change is different. There’s no prepackaged number of emissions that are stamped and packaged by a friendly pharmaceutical company.

   It could be that very small quantities can drive important climate effects and generate harmful climate events.

   MATTHEW: Really? But the volume of carbon in the atmosphere is so gigantic.

   KAIN: That’s part of what I wanted to draw attention to. Personal incredulity seems like a weak appeal for a physical process.

   But also, on a simple linear model, some quite small volume of emissions could make a difference. So, again, that quarter of a person’s emissions, or even smaller quantities, could be sufficient.

   MATTHEW: That seems implausible. But for the sake of argument, we can grant it for the moment.

   KAIN: Well remember I’ve distinguished the No Difference claim from *Some Difference*. My interlocutors say that even when many many people are contributing small bits of carbon, *none* of them can cause dangerous weather events. But even in your original aspirin case, *some* aspirin was sufficient to generate the level of toxicity required to kill you when you took a hundred (and less might have been dangerous but not enough to poison you).

   MATTHEW: Okay, but that’s a familiar threshold point we’ve heard from Kagan and others.

   KAIN: Sure. That wasn’t meant to be the new contribution of this post–the new contribution was supposed to be (a) that incredulity is not a good appeal and (b) there is nothing scientifically privileged about a single individual’s emissions–fractions of those ex ante could have been enough to cause dangerous events. One person’s emissions might be a morally relevant category (because one person is a special kind of moral category, say), but there’s no reason so far to think that it’s a scientifically/physically relevant category. Or at least none without more evidence, preferably scientific evidence from my No Difference interlocutors.

3. KAIN: The new contribution of this post was supposed to be (a) that incredulity is not a good appeal and (b) there is nothing scientifically privileged about a single individual’s emissions–fractions of those ex ante could have been enough to cause dangerous events.

   MATTHEW: I agree that there’s nothing scientifically privileged about a single individual’s emissions. It would be odd if there were–particularly because some individuals emit five, or ten, or even more times as much carbon as do others. But I don’t think that’s what Sinnott-Armstrong et al. are claiming. Their claim is like mine about aspirins: one aspirin just isn’t enough to kill you. And if so, half an aspirin won’t be enough to kill you either.

   KAIN: But this rests on a naked appeal to intuition. They haven’t actually *shown* it.

   MATTHEW: Granted. However, it does seem very hard to believe that a single person’s emissions could raise the temperature enough to make a perceptible difference. ‘Using Nolt’s scenario’, Aaron Maltais observes, ‘we can very crudely calculate that the average high-emitter’s lifetime emissions will contribute on the order of a billionth of a degree to a total 3°C increase in
   global mean temperature’ (‘Radically non-ideal climate politics and the obligation to at least vote green’, pp. 592-93). How could anybody ever notice such a change?

   KAIN: Even if such a rise in temperature is too small to make anyone feel perceptibly worse, it seems possible that it could make a meaningful difference in other ways. And remember Broome’s argument that some individual emissions could have nonlinear effects like causing a storm.

   MATTHEW: These are more persuasive arguments. I’ll grant you that (a) simple incredulity is not a good reason for dismissing them. But observation (b) still seems neither here nor there.

4. Thanks so much Matthew for both these comments; I think this thread is really helpful in clarifying exactly what the issues are. I hope to expand this post into a short article submission so I’ll try to address a response of your type when I do so.

5. Thanks for this thoughtful post, Kian!

   I’m sympathetic with your suggestion that the issue of individual’s causation of climate effects is at least in part a scientific issue, and climate scientists have not spent much time on it (nor, perhaps, should they, with many other pressing jobs to do). So many of us are left with, as you suggest in your exchange with Matthew above, a sense of incredulity that individual emissions are enough to cause harm on their own, perhaps fueled by our inability, as lay people, to grasp the mechanism by which these changes might take place at such low-level changes of levels of emissions. And these seem insufficient, you say, to let individuals off the causal hook.

   I also want to highlight that there are a wide range of normative judgments we can make about high individual emissions even if we conclude they never cause harm: they could be callous, ugly, unhelpful. By contrast, a low emission life-style might be beautiful, ideal, and we might have significant moral reason to pursue it, even if it makes no difference to climate-related harms. I believe many of these judgements actually hold.

   To get to the crux: It really seems we are running up against a burden of proof issue between “harmists” who assume that individual emissions cause morally relevant harm and “no-harmists” who deny this. The harmists believe that *no-harmists* must show that individuals are *off* the hook, whereas the no-harmists believe that *harmists* must show that individuals are *on* the hook.

   I’m still working out exactly what I want to say about this burden-of-proof standoff, which I think we didn’t address in our paper. Here’s one idea. Who seems to have the burden of proof here might depend a great deal on framing. In the framing I think your piece pushes us towards, individual drives (say) are just a “subset” of the larger group of actions (the aggregate group of acts of driving, aluminum smelting, air conditioning, cement production etc) that create say, the US or China’s emissions and thus cause global warming. Then, I agree, it might seem “natural” as you put it, to place the burden of proof on the no-harmist (why stop at a country’s set of drives? why not a city’s? an individual’s?)

   But what if someone asserts, quite plausibly, (a) that the harms of climate change are fundamentally caused not by all the countries’ sets of emitting acts per se, but by the lack of action of governments: their inability to respond appropriately (by mitigation and adaptation) to credible suspicions of climate change back in the 1980s, and/or to proceed more cautiously in developing industries with large-scale impacts. Even more radically, someone might hold that (b) it is not even countries’ actions that are the true cause of climate-harms but features of humanity as a species that fundamentally cause climate change, such our talent for innovation, our greed and desire for comfort, coupled with our lack of foresight, and our susceptibility to collective action problems. Under either of these framings (a) and (b), it is not obvious that the no-harmist is drawing an arbitrary distinction between sets of different sizes. Rather, it appears they are drawing a very “natural” distinction between very different kinds of actions between those undertaken by individuals and a global community of states in (a), and by individuals and a species in (b). So, I suggest, what might be driving our intuitions about who has the burden of proof might depend on how we are framing the root cause of climate change.

   Relatedly, I’m interested in your final sentence, which seems to be doing quite a lot of work here. “When considering potential harm to others, we should adopt a higher bar of justification [than mere metaphysical possibility that our actions cause no harm]”. I thoroughly agree, and I’d be interested to hear (perhaps in the article version) where you think that bar should be.

   Thanks again for a charitable and insightful exploration of this issue.

6. Dear Ewan,

   Thanks so much for your comment and of course your published work that helped occasion these thoughts (and I’m very relieved that you take my reading of your work as charitable!)

   Honestly, I agree with much of what you write in this comment. We can certainly agree that individual excessive emissions might well be callous and unhelpful (I just think in addition they can also be harmful–at least on occasion). As for the burden-of-proof standoff, I’m happy to keep chatting with you as we work through this discussion. We have a good mutual understanding of our disagreement, and that’s an important start.

   So you offered two more challenging points: (a) perhaps the relevant entity is governmental (social/national) action or (b) perhaps even the species. Of course, these are very different kinds of entities than individuals. Perhaps, you suggest, if we take the relevant long-term, large-scale decision-making to be at these scales, that shows an important quantitative distinction between individuals and governments/species. I accept these are real qualitative distinctions. One suggestion that was made to me (in person) today in response to this post might be similar to your picture. The institutions and social structures we are part of have major effects on what we are nudged towards or even what we take to be live options. I do not dispute this; nor do I dispute the point that governments have moved agonizingly slowly (and, in the cases of US and Australia, backwards). However, I don’t think there’s any tension in thinking that scale-effects are valuable (and likely super-additive–I wonder if my interlocutors are going *beyond* this agreed super-additivity to a function with a major step change from 0 to a positive value?) and should be encouraged while also believing some individual effects can be meaningfully harmful in climate effects terms (not in the usual case, plausibly, but at least sometimes). I do not take myself to be committed to the claim that reducing your own emissions is more effective than other kinds of social action; my own view is that it’s difficult to know, but there’s something valuable in doing them in conjunction so you have a consistent sense of self.

   It’s worth saying a sentence about the motivation behind this post (and the short piece in development): it seemed that there was some talking past each other in this discussion. Johns’ (both Broome and Nolt) point that you can divide up the large harms to find individual marginal harms seemed to be missing the target of your work. Trying to consider your position on its own terms was the purpose of these thoughts.

   PS The final sentence is methodologically important to me, but I don’t have a sophisticated theoretical idea of how to spell it out yet, though. Happy to keep talking about that too.

7. Two ideas here, one in support and one against (which I actually hope is wrong, but I’m not sure!) First, it seems obvious that we need to talk here, not about causing harm, but increasing the probability of harm–very slightly, but for large numbers of people across time. So Sinnott-Armstrong’s claim that a joyride does not cause massive harm is true, but utterly irrelevant; it increases the /probability/ of this harm. And it probably doesn’t even increase the probability that a million people will suffer in a storm or drought as opposed to none of them doing so. But it slightly increases the probability that the storm (and many other similar ones) will be slightly worse, and harm one extra person than it otherwise would (or erode away a coastal building slightly quicker, etc.) And that is bad enough, and easy to at least roughly calculate as Broome has done.

   However, it also seems to me there is one possible argument for Difference which you (and its supporters) may have overlooked. I made this argument in an unpublished conference paper several years ago, and confess I don’t know if it is sound, and even hope it is not, but I haven’t developed it further either way. It is that there is such a huge economic incentive–cheap power–to emit greenhouse gases, that we can be confident that the human race will not collectively stop doing so that until the damage reaches a certain point at which time the harm is so palpable that enough actors have an incentive to put in place a collectively enforceable solution to the problem (carbon tax, etc.) Prior to that time, any one person’s refusal to release such gases simply puts off the time before we reach that critical point, and the total amount of gases released over time, and the damage this causes, remains the same. In effect your individual sacrifice just gives others more time to act on their incentives to release such gases. Cf. the standard case of the tragedy of the commons, in which–assuming some amount of human greed and actors who cannot trust each other–the commons /will/ eventually be exhausted regardless of what any given individual does, even though if all of them could impose a collectively enforceable solution the commons could be saved, so you might as well use what you can until the commons runs out (or enough people see the harm that they are willing to bargain towards a conservation agreement). If this is true, then, perhaps, your individual choice to release such gases or not may indeed make no difference, not even probabilistically. Even if this is true, of course, there remains a powerful argument for imposing a collective solution, perhaps again based on Broome-like calculations, if that means that we could step up the time at which such a solution gets imposed. But until that is imposed, individuals may have no duty to release fewer gases, and may not even meaningfully be said to be increasing the probability of harm caused by such gases, until such a solution is imposed. This is of course not only consistent with, but makes all the more urgent their a very strong duty to promote such a collective solution. The tricky thing here, I think, is that there may be no fixed point of actual or imminent damage at which the collective solution gets imposed. But it does perhaps suggest that political or other work to /change/ that point from what it otherwise would have been is far more important than reducing your own emissions when we, manifestly, have not yet reached whatever that point is. Indeed, it might be the *only* important thing, the only duty we have w/r/t gas emissions, prior to its imposition.

   Now one still might wonder, how could this argument support Difference? How could it be that global emissions cause (increased probability of) harms but each individual’s emissions do not do so, given that the former consists of the collection of the latter? The answer, I think, is that the former doesn’t actually consist of the latter, and there’s a subtle deception which makes it look like it does. Anyone (say, a world government or treaty-making body) in a position to limit the first is in a very different epistemic situation from any individual outside of such a regime: the former knows that there are *no* others who will retain their incentives to emit gases if they enact some policy, while each individual knows that there *are* many others with such incentives. So for each individual, the expected probability that their emission of gases will cause increased harm could be literally (not just close to) zero, and the total sum of all such probabilities remains zero, at least when each choice is treated as an individual choice in a tragedy of the commons scenario. For we have good reasons to think that it is extraordinarily unlikely that everyone will just stop at the same time, especially when they don’t trust that others will (ok, there’s some very small chance that this could happen, but it is what Dennett calls a “vastly” small number, with a denominator so large it makes astronomical numbers look small). This is consistent with the probability of reducing harms by *enforceably changing global emissions* by someone in a position to impose relevant costs on all emitters being non-zero.

8. Dear Kian/KAIN,

   Many thanks for the insigtful read. I had no idea that there existed such a thing as “individual denialism”. Like Ewan above, I’m a bit puzzled as to whom bears the burden of proof. But if it were to rest on “no-harmists”, I would say that they must offer at least some intuition of what constitutes a “large group” and why. According to your recounting of their argument, countries seem to be large enough groups. What about states/provinces? Cities? Is a group defined by a set of policies, which hence renders them accountable for their actions? Or is sheer population numbers enough (in which case appeal to the fallacy of division is inappropriate)?

   I like your framing in terms of incredulity (which is more pronounced in the response to Matthew’s comments than in the actual blog entry). You seem to be on to something. Also, and even though you focus on causation, it seems that our incredulity is exacerbated by the fact that cliamte change can only really be *experienced* as a group.

   Indeed, any evidence of climate change is statistical in essence: more and stronger hurricanes, more forest fires, more droughts, a gradual expansion of malaria-bearing mosquitos to more northern latitudes. All of these phenomena happen on a large scale, both in space and time. By contrast, when an *individual* loses their home to a hurricane or a forest fire, or their crops turn out poorly because of a drought, or they becomoe infected by malaria in a reion where infecting mosquitos aren’t supposed to be… it is not global warming, it is bad luck.

   My point is that one must be very careful in distinguishing the causation from the effects, as you do. However, the incredulity (about causation) that you contest seems to comingle with another type of incredulity, one about the effects. That said, I realize this comment is only tangential to the question at hand, about causation.

9. I don’t see this as being much different from any other “paradox” of group action. Say one’s government passes a law limiting your pollution emissions or requiring you to pay taxes, or in my jurisdiction requiring you to vote – the same “no individual difference” argument is not accepted. Causation wise, “no individual cigarette” sounds just as specious, even though the lung cancer does arise from a single mutational event presumably arising after a single dose.

10. Dear Scott Forschler,

    Thank you for reading my post and for your comments. I think our views are actually very closely aligned: once you’ve opened up the possibility that some individual’s emissions at some times cause morally relevant climate harms, then the question is how often and how serious those harms are. At that point, it’s very plausible that what is relevant (Broome as cited above, among many others, argues this explicitly) is the expected value over the probability of different scenarios that a given tonne of CO2 will cause harms (which presumably lie amongst higher probability scenarios where that particular marginal tonne does *not* trigger any harms) to determine an expected (dis)value of the action. My interlocutors want to deny that first step–some of them would agree with the rest of this line of reasoning if it were the case that some individual emissions sometimes cause harm could be established. So I think what you are saying is where we want to go, but the individual denialist does not accept that first move.

    ‘So Sinnott-Armstrong’s claim that a joyride does not cause massive harm is true, but utterly irrelevant; it increases the /probability/ of this harm.’

    This I disagree with, at least if we understand probability is subjective terms, which are the terms I think relevant. In order to make that expected value point true, you first need to establish that some of the joyrides *do* cause harm (e.g. given a background level of x tonnes of CO2 in the system, it would not cause harm but, perhaps, given background level of x+1 tonnes, this marginal tonne would cause a morally relevant harm) and then figure out something like the fraction of cases where there is climate harm caused out of the total.

    Another way of saying the same thing is that, from *our* point of view, it is rational to take the harms as expectational/prospective/subject to credences/subjective probabilities even if, from the point of view of the physical world, the actual system is deterministic.

    However, if you mean the system is objectively probabilistic and the increase in emissions adds to objective probability, then I don’t think the statement makes sense. Sometimes the joyride causes harm, unless you think that as a matter of fact all of the emissions increase the probabilities of harm but none of those harms materialize. Or perhaps you think that causation has to be deterministic, but this seems straightforwardly false (cf. my smoking caused my cancer, even though that just means that the objective probability of my developing cancer increased).

    As for your other point from this very interesting sounding conference paper, there are several thoughts I have. The first is to note a helpful economic term here–the elasticity of demand. The question is how the demand shifts in response to changes in price (in percentage terms). In your case, that’s how the shift away from conventional energy sources from some non-trivial number of people (if it’s an individual, it won’t shift be enough demand shift to be measurable in the price, even though the demand curve has in principle shifted) generates a change in price and the question is how the market responds. This is structurally analogous to a question I’ve thought about, with the relevance of divestment as a tool to affect demand for market value for conventional fuel companies. There, the issue is how much a group of divesters shifts the value of the company. In your case, if others increase their demand in response to the temporary change in price exactly the same amount to offset those who shift away, you’ve had no net (market) effect. In the divestment case, if others increase their demand for the company in response to the temporary change in price exactly the same amount, the divesters have had no net (market) effect. In both cases, how others respond is an empirical question and is measured by that elasticity–more elastic demand means they are very sensitive to price shifts, less elastic means not. A fabulously clear report from Oxford’s Smith School (https://www.smithschool.ox.ac.uk/publications/reports/SAP-divestment-report-final.pdf) is helpful here. Propositions 1 and 2 (p. 29f) are that what is relevant to making the temporary price change is the relative shift in demand to the firm’s market cap, and that it’s unlikely to produce permanent changes in demand since there will in general be more demand at a lower price if the fundamentals don’t shift. However–and this is crucial–these are actually empirical questions. We should assume that most of the time if the fundamentals don’t shift the market value will go back to the same, but that’s not a law of nature, it’s an empirical question about the decisions of individuals and individuals are subject to all kinds of heuristics and biases. And the quantity that describes the relevant empirical consideration is the elasticity of demand for this product.

    It might be that you are thinking of something simpler, where we really are thinking of the individual, who avoids emitting and reduces demand. As a mere individual, she has no measurable (but in principle some!) effect on prices. But in that simple case, scarcity should also have an effect on prices–so the cheapness of conventional energy would be highly dependent on how easy it is to produce which depends on how much has been used (we extract the cheapest to extract first and then go to more expensive sources). So, again, I think the simple picture requires some more subtlety.

    A second is that it seems odd to me to think of a *fixed* critical point at which sufficient support kicks in for robust policy response (you also say it’s odd to think of it as fixed!). Human beings are too heterogenous and non-deterministic for there to be some threshold (different people have different thresholds; communities act in non-linear ways depending on values and social coordination, etc. etc.) However, this is a bit complex, because it could be that if I knew sufficiently large amounts about people (if I were a Cartesian demon), they would look very predictable and it would look like there would be a critical point. However, I doubt this; I think the critical point would depend on, inter alia, the values and composition of the extant population.

    Another point in support of my doubt is that many jurisdiction *do* have significant carbon policies (including real carbon prices). As Americans, it can be easy to forget the level of actions in other countries. So, for instance, you can see Fig. 3 here: https://www.carbontax.org/where-carbon-is-taxed/ to see that large carbon taxes are available and have been implemented. (Indeed, in Canada, there is a successful carbon tax scheme–I wrote a short intro to it here: https://climatefootnotes.com/2017/09/18/carbon-prices-bc/ .) Of course, these things are complex–Australia set up a very ambitious and exciting program to tax carbon and then the government was ousted (at least partially in opposition to those taxes, but of course Australian politics are a series of people stabbing each other in the back so there are more drivers than that).

    Those are just some first thoughts, so please feel free to come back and tell me whether any of that is helpful and/or engages with your proposal as you understand it.

11. Dear Justin,

    Thanks so much for reading the post and for contributing! As is usual with us, I think we are in complete agreement. I have to admit that I snuck in the national and global emission cause harm claim because it’s unusual that individual denialists make this explicit (although I think they believe it). As you rightly note, that requires some type of story about why those are large enough (and not two people or seven people or a small village, etc.) I like Ewan’s comment above, because he starts to give some stories about entities that could be relevant. So, for instance, he suggests that bodies that have sufficiently robust governance can act systematically (national groups might then qualify but not regions, perhaps?) While I granted that these governed groups could help coordinate and generate super-additive effects, that’s not enough to establish the radical discontinuity we need for No Difference to obtain.

    I am very sympathetic to your point about the scale of effects and that could indeed be part of what drives the intuitions of some of my interlocutors (if that is or is not the case, please tell me, interlocutors! I’d love to know!) In the Broome paper, there are a few models of how climate harms are driven by emissions: there’s the intuitive linear model, which is certainly untrue; there’s a model which has step changes up and down (at climatic and environmental tipping points), but acts as a very coarse-grained linear analogue; and there’s the model he thinks is correct which is *extremely* noisy and chaotic but trends upwards (more emissions/more climate harms). On the chaotic model, it could be that a given unit of emissions *reduces* harm whereas another *radically increases* it, but the overall subjective states we should adopt are that more emissions in expectation lead to more harms. This notion, that some very small shifts in climate forcings lead to major effects/harms, is unintuitive and I understand that it seems weird. But my point is that this is just an appeal to incredulity (together with a related point that on none of these functions is there any privileging of individual-scale emissions compared to any other scale).

    PS If you’ve contributed or just read, especially this far in the comments, I really appreciate it. I am always open to more thoughts so feel free to email me if you have thoughts, including ones that are too unformed for public discussion.

12. Dear Scott Forschler,

    Thank you so much for your comments and for reading the post. I wrote a long reply and I think it’s being moderated because I included a few links and they are trying to filter for spam, so I guess it’s forthcoming!

13. Dear David Duffy,

    Thanks for reading and for your comment! I certainly agree that the structure of this question arises in many other contexts.

    “Say one’s government passes a law limiting your pollution emissions or requiring you to pay taxes, or in my jurisdiction requiring you to vote – the same “no individual difference” argument is not accepted. Causation wise, “no individual cigarette” sounds just as specious, even though the lung cancer does arise from a single mutational event presumably arising after a single dose.”

    I take this as a friendly comment, in that it sounds like you think the No Difference claim (and its counterparts in other debates) is not plausible on standard views of causation. (But feel free to come back to explain if I’ve misunderstood.) And yes, I think it looks a fair amount like the cigarette case: it would require a huge amount of (highly invasive) information to determine which cigarette generated a mutation even if it is implausible that there was no such event after some particular cigarette. In my terms, No Traceable Difference is highly plausible but that’s not sufficient support for No Difference.

14. I really like your post, especially the insight that there’s nothing scientifically privileged about the *individual*.

    One small issue (and in order to exculpate myself in advance for any obvious error: I haven’t thoroughly thought this through at all):

    A conclusion can be based on some premisses which we know a priori *and* on some premises which we know a posteriori. Thus, we know the conclusion on both a priori and a posteriori grounds.

    The claim which you labelled ‘Difference’ might be such a claim. And thus it is wrong to ask whether we know it a priori or a posteriori — since it could be based on premisses of *both* kinds. One could refute it by questioning either an a priori premiss or an a posteriori premiss.

    Why do I mention this? The reason is that I was confused. On the one hand, I noticed that I agree with you — and was happy that you highlighted this — that we need to know empirical facts (e.g. that CO2 exists and that it makes a difference to the climate system) in order to counter the denialists. On the other hand, I often had the impression that my disagreement with the denialists rested on a priori grounds (something along the lines — and I know: this needs to be worked out much better and it will remain controversial even if stated more carefully — of this metaphysical claim: “It is impossible that A+B cause X while neither A causes X nor B causes X”).
    I therefore thought: Yes, we do need to know empirical stuff in order to claim that individuals make a difference. However, we *also* need to know a priori stuff in order to claim that individuals make a difference — and maybe, it’s the a priori premises that are the controversial ones?

15. Dear Dominic,

    Thanks for reading the post and for your very interesting and promising suggestion. So the idea is that my suggestion of a priori or a posteriori grounds is too simplistic and there might be some kind of conjoined answer (you suggest the model of a mixed conclusion which has both a priori and a posteriori premises).

    The first thing to note is that I don’t think I’ve committed myself to there being the exclusive disjunction of these two options. If I show neither of these works, that is sufficient to show that it is not the case that some conjunction of them works (indeed, if I show either of them does not work, that should be sufficient to show the conjunction does not work).

    However, that does not get to the main point of your idea. So it is definitely the case that one way to address those who adopt No Difference is to say that there isn’t something special about aggregations which make things appear out of nowhere (this is Av Hiller’s point, which is linked above in my Works Cited and I won’t relink to because I’m worried that the reply will get flagged for moderation). That is the standard way (and, indeed, I endorse it). My objection for the purposes of this post is to grant that there is something special about aggregation–allowing the metaphysical possibility of something (which seems odd to many of us who do not adopt No Difference) and arguing that it is still not enough to support No Difference. So I’m arguing along a different track, with different presuppositions (that is supposed to be the original contribution of this argument). If my argumentation is sound, we do *not* need the a priori point to object to No Difference.

    The takeaway is, yes, you are indeed right that the a priori premises are controversial (an excellent discussion of the kind of argument you have in mind is in Section 6 Gunnemyr (forthcoming)), but my contribution is to grant the metaphysical a priori point you want to contest and say that those defending No Difference are under-motivated regardless.

    Gunnemyr (forthcoming). Causing Global Warming. Ethical Theory and Moral Practice. Online first and open access at doi:10.1007/s10677-019-09990-w

16. Kian, thank you for your very informative and clarifying response. We are indeed in strong agreement. First, you are correct that the probabilities in the first part of my response are subjective, and do depend on something like the claim that some emissions do (objectively) cause harms, and then the question is just how to cash this out in expectations. Although I would suggest a modification to that middle claim: it’s not like some emissions fully cause harm, and others just happen not to. Rather, they all contribute very slightly to the source of the harm, i.e., they may each add a few newtons of force to the totality of winds around the world, or a few atoms of water to each hurricane, etc., and this marginally increases the amount of damage the hurricane causes. But I think that’s a minor amendment consistent with your point.

    You are correct that my second argument depends on elasticity of demand, which is an empirical issue, which is part of why I am not certain that my argument is correct (nor can I rule it out yet). I did note towards the end that there is probably not a single fixed critical point of damage at which a global cooperative solution (GCS) will be adopted, and there are probably ways to adjust that point (and it is also possible that the amount of damage so far/about to be caused is *not* the only or even dominant factor in adopting in GCS; technology/economics might preempt this, or interact with it in complex ways, which is another basis for skepticism about this argument). But any individual’s voluntary restraint on emissions, in the context of a tragedy of the commons, is by itself likely to have literally no impact on where that point is (or, perhaps some very small, but *far less than*, say, 1/7 billionth, given that others’ emission incentives will remain strong). Hence, again, we should focus on what we can do to change that point at which GCS is adopted, not on individual restraint, if we want our actions to cause reductions in overall harm. Even if the point moves around randomly, at least with respect to individual’s choices about restraint, my argument may work (ceteris paribus my other caveats mentioned above and below).

    Carbon taxes in individual countries probably fall somewhere between individual restraint and a GCS in their effects. Especially when the country is small–which, of course, Australia and Canada are, even despite their high level of per capita industrialization–the results may fall closer to individual restraint, given that (say) China and the USA might simply take this opportunity to emit more. OTOH, these acts could act as significant feasibility demonstrations (“see, /our/ economy hasn’t collapsed, so we could all do this together with moderate overall costs”), which might accelerate adoption of a GCS. To that extent such moves may be morally significant, even if individual restraint does literally nothing to reduce overall harm.

    One point that occurred to me later is that there is a difference between “causing” by being a physical link (in some sense) towards generating some result, and causing in the sense of controlling (able to make happen or prevent). My emissions may cause some harm given what others do; but if others will emit more if I exercise voluntary restraint, then I don’t cause the harm in the sense of being able to prevent it (or make it worse). And this distinction does not depend upon subjective probabilities; it may also be true in a deterministic world. Either way, of course you can’t meaningfully add up all the zeros and say therefore the collective behavior causes no harm, because the no (controlling) harm of each person assumes the context of others stepping in to act on their incentives to emit more. If you assume that everyone spontaneously exercises individual restraint then this is no longer true, but of course this a philosopher’s-case assumption on a par with evil demons mucking things up; in the real world things like this are vastly unlikely to ever occur. In contrast, a group or persons in a position to make policy can control-cause harm or a reduction thereof; but this causal power is *not* the addition of the (non-existent) control-causal powers of the individuals who make up the group. I’ve seen this addition mistake made many times in arguments about individual vs. group responsibility, e.g., Gruber, PStudies 174: 3097–3108, whose objection to expected utility AC rests upon assuming both that if agent A doesn’t shoot a victim, agent B either certainly or almost certainly will, and the same about B w/r/t A, concluding that EUAC lets them both off the hook (assuming they each have some significant incentive to shoot) when obviously collectively they cause the victim’s death and should not do so as a group. The situation is simply another tragedy of the commons and not a reductio to subjective AC: no individual shooter is a control-causer of the harm, which is compatible with the group being responsible and a control-causer, if there is (or we could create) a locus of group control. If there is no such locus, you can’t simply add A’s (non-existent) control-causal power with B’s to a make a group control-power. Nor can you simply say that if neither shot the victim would live and suggest that this is incompatible with the claims that B will shoot whether or not A does and vice versa, for if you “add” the latter two claims you simply assume that both will indeed shoot, which is obviously incompatible with the assumption that they could both fail to shoot! Regan, in Utilitarianism & Cooperation p18, makes the same mistake in adding choices depending on such incompatible contextual assumptions. I wonder, then, if such mistakes rest partly on a confusion between what I am clumsily calling physical-link causation and control-causation, where I think the latter and not the former is relevant to moral responsibility. Or is my distinction here itself confused? It probably is in some way and I would love if someone could explain why.

17. Most discussion of these questions revolves around the recent literature, but these problems go way back. In particular, with respect to whether ‘it is impossible that A+B cause X while neither A causes X nor B causes X’, there was a debate in the 1960s and 1970s that’s still worth reading today.

    David Lyons’ *Forms and Limits of Utilitarianism* (1965, chapter 3) argued that the whole has to add up to the sum of its parts. In 1974 and 1979 papers in Phil Review (‘Simple and General Utilitarianism’) and Noûs (Utilitarianism and Group Coordination’) respectively, Harry S. Silverstein sought to show it didn’t. Silverstein’s 1974 paper prompted a rebuttal from Jan Narveson (Noûs, 1976, ‘Utilitarianism, Group Actions, and Coordination, or, Must the Utilitarian be a Buridan’s Ass?’).

    I believe that Silverstein made a good case, though it’s far from clear to me that he settled the issue. (Mike Otsuka’s 1991 P&PA paper ‘The Paradox of Group Beneficence’ gives some strong reasons for thinking the whole *does* have to be the sum of its parts.) In any case, the Lyons-Silverstein-Narveson exchange deserves to be remembered, lest we risk reinventing the wheel.

18. Dear Scott

    Very glad my points did properly intersect with your considerations and that we are on the same page. And thanks for continuing the discussion. This is very enjoyable.

    [1] It sounds like our pictures are basically the same. Also I think your friendly emendation of ‘contributing very slightly to the source of the harm’ is very plausible and close to my mental model.

    [2] Also glad that you think the elasticity of demand it a (the?) relevant concept. I understand from your response that you are thinking of humanity as a large whole which requires a form of global cooperative solution. We are in agreement that, in markets as they are currently constructed, it’s not a simple linear function of individual inputs–an individual’s (or even a firm’s and, in the absence of relevant trading arrangements, a country’s!) refraining from emitting is less than the simple fraction of emissions because there will be some demand reallocation from other agents due to a temporary drop in price.

    I’m also pleased to note that we also agree that (national or regional) carbon taxes are probably some type of intermediate case between individual restraint and a global system–they generate systematic incentives over a large group of people, but not at a global scale. This is among the arguments that proponents of carbon taxes often point to in trying to advocate a harmonized global social cost of carbon (SCC).

    [3] I am intrigued by your distinction with respect to causing. To me, ‘controlling’ (being able to intervene to determine an outcome) is not a form of causation (political theorists like my colleague Philip Pettit call it ‘domination’ because they are concerned about governments stepping in to manipulate citizens when they are dissatisfied with outcomes, even when in the actual case there is no stepping in). But, if I follow, I do share your concern that philosophers think that the relevant baseline is actual alternatives–and that things like EUAC do not help us with overdetermination cases. Early in my studies, I read Woodword’s manipulationist causal account, and (if I recall correctly!) he says the relevant counterfactual involves a local miracle a la Lewis–we remove the alternative cause (e.g. the other shooter) and ‘manipulate’ our cause away. That seems to give the right causal story at the cost of introducing local miracles into the relevant counterfactual and/or baseline. That seems to me an acceptable cost to avoid the kind of mistake that–if I follow–you and I are concerned with.

19. Dear Matthew,

    Thanks so much for pointing me towards that debate–As you suspect, I have indeed *not* read those papers (although I’ve read and been very interested in Narveson and Lyons’ other work–for instance, I think Lyons’ breakdown of different forms of ethical relativism should be the standard categorization, but I don’t meet other people who are familiar with it!) so that is extremely valuable. I will read those as I prepare for this post to be a piece on its own.

20. Kian, you might be right to distinguish control as not necessarily involving “cause” at all (although presumably it often does involve causing). I’m not wedded to the terminology. But if we separate those, then it still simply seems to me that control over harm is morally relevant and causing harm, by itself, is not (some deontologists, and even some consequentialists, may disagree of course).

    I’m also pleased with Matthew’s citations; turns out I’ve read some of those articles, and the missing one I read quickly in the last hour. 🙂 I’m not sure who is right in the debate or exactly how it intersects with my thoughts because they are often making some very fine distinctions which I confess it would take more study to follow closely. But I noticed that Silverstein seemed to define cause as necessarily involving control, which perhaps is important at some stage in his argument. Again, whether this is correct or not, my worry remains that at least some conflicting intuitions about the morally relevant ways in which individuals or groups “cause” some harm might be due to a confusion between cause and control. I’m not sure how that intersects with the Silverstein-Narveson debates and cases, though it seems to be potentially involved there from my cursory glance.

21. I was always mesmerized by the fact that philosophy and climate change could walk together to a better world.
    Luckily we still have people like you Kian that can create that bound and can show some of the ways that we can build.
    I don’t have much to add or to improve your text, but I wanted to write here some gratitude words towards your thoughts.
    Thanks for your text. You make the difference!

22. Dear Cami,

    Thanks so much for reading this and for your kind words! I miss our times in Vienna~ Hope all’s well with you!

23. I think Silverstein is indeed talking about what Scott’s causation/control distinction; note his remark about Williams’ Jim and the Indians: ‘Jim is not responsible for the death of the Indian he shoots precisely because, since nothing he can do will prevent it, he does not cause it; hence, since “to kill” is at least “to cause the death of,” Jim’s act is not properly described as “Jim’s *killing*” of the Indian’ (‘Utilitarianism and Group Coordination’, p. 355).

    Here Silverstein uses ’cause’ as Scott does ‘control’, adopting a counterfactual view of causation. But I think it’s a matter of the different ways we use ’cause’–both to describe the physical process (in which case Jim causes the death) and the difference it makes (in which case he doesn’t).

24. I think Silverstein is indeed talking about Scott’s causation/control distinction; note his remark about Williams’ Jim and the Indians: ‘Jim is not responsible for the death of the Indian he shoots precisely because, since nothing he can do will prevent it, he does not cause it; hence, since “to kill” is at least “to cause the death of,” Jim’s act is not properly described as “Jim’s *killing*” of the Indian’ (‘Utilitarianism and Group Coordination’, p. 355).

    Here Silverstein uses ’cause’ as Scott does ‘control’, adopting a counterfactual view of causation. But that’s because we can use ’cause’ both ways–both to describe the physical process (in which case Jim causes the death) and the difference it makes (in which case he doesn’t).

25. Thanks, Kian and Matthew. Nothing to add! I now see your strategy more clearly, Kian. And, Matthew, definitely agree about the danger of reinventing the wheel.

26. Dear Matthew and Scott,

    I was away for a bit while I read some of the pieces that Matthew suggested. They are truly excellent and I am very happy Matthew brought them to my attention. I especially like that they are in discussion with each other; I learn a lot more from papers that respond to each other than papers alone.

    One of the relevant classes of cases that Silverstein describes is where there is an ideal number of candidates to pursue an action, simultaneously, say, and more is unnecessary (or, in the climate case, where as sufficient number of candidates (emissions) would generate some bad effect, and more than enough emit). Silverstein says, simplifying greatly, that there is something non-linear (non-additive) going on here and that we can neither say simply ‘you should perform the act where everyone’s performance of the act would be optimal’ nor ‘given that you are part of a set which already is sufficient to generate the optimal outcome, you should refrain’ (which would be suboptimal if everyone followed it) ‘or you should perform it’ (which would be suboptimal if everyone else’s contribution was fixed).

    Silverstein concludes that there isn’t a coherent way of determining what the contributions are or what the individuals should do on a standard utilitarian picture.

    Narveson’s response is, I think, a pretty solid one. Narveson draws attention to the point that Silverstein is trying to make marginal contributions to utility out of average contributions but you can’t do this in moderately complex circumstances. The other point of Narveson bears closely on the discussion from my post: it depends what kinds of assumptions we make about others’ actions. Here are a couple: they will act predictably in some fixed determinate way (then the utilitarian response is straightforward) a;nd they will act according to a utilitarian calculus (then the utilitarian response is calculable, although slightly less straightforward, and might involve some probabilistic assumptions about the others–if symmetric, they each act in such a way that the expected value of the random strategy is maximized). But in either case, one may or may not make the difference (depending on the expectations about the other). I think this picture is most congenial to me, but I’m open to hearing more.

    As an aside, it was cool to see that Narveson distinguishes between ‘participatory’ and ‘stimulatory’ actions where the former are trying to generate a large enough coalition through personally adding to action and the latter are trying to generate a coalition through encouraging or rallying others.

    PS I really enjoyed Narveson’s “E.g., the group elects Trudeau, though no individual may be described as having done so-they can only vote for or against the relevant parliamentary candidates” (in “Utilitarianism, Group Actions, and Coordination” page 183). I had a moment of confusion where I thought this can’t be right because it is an old paper. Then I remembered that Narveson was writing long enough ago he was writing about Trudeau *pere*.

27. Dear Scott

    I’m not why I didn’t think of this earlier, but the toy models that Hilary Greaves (2018) sets up in her recent paper share a lot of similarities with your argument. In particular, (a) focus on absolute cumulative emissions, not rates of emissions, and (b) focus on total possible emissions (in your case, because there is supposed to be an amount that is socially/politically possible but in her case because there is a finite set of retrievable fossil fuels). Hers does not include an economy–so the idea that there will continue being extractions and emissions as long as it is profitable is not part of that model. But perhaps you can play with that a bit and see how modifying it would allow you to discuss your idea. The article was published only a couple months ago and is available at https://academic.oup.com/monist/article/102/1/42/5255706 .

    Works Cited
    Greaves, H., 2018. Climate Change and Optimum Population. The Monist, 102(1), pp. 42–65.

28. I’m coming to this discussion rather late, but I just got a chance to read this post…

    I think one of the best argument that an individual act like a joy ride won’t make a difference to climate change harms comes from an unpublished paper of Mark Budolfson’s. (The point might come up in his published work too; I’m not sure. But I know the argument from his paper “Collective Action, Climate Change and the Ethical Significance of Futility”, available on his webpage).

    First, start with the point that – as I think is true – the best argument that an individual act like a joy ride can make a difference is some version of the ‘threshold’ argument. (Arguments that don’t work include, for instance, claiming that we can just take the total harm, divide it up and ascribe each individual act a share. Parfit showed us that that sort of argument is confused with his discussion of the Share-of-The-Total View in Reasons and Persons.) So, take the threshold picture: there are emissions thresholds, at which higher levels of greenhouse emissions cause harmful climate changes. (These could be chaotic, non-linear, etc. But the point is there are tipping points where a little more emissions will change harmful outcomes.) The common claim is that this means that there is a chance of making a difference, because there is a chance that your drive will push things over a threshold. But is that correct? Here is Budolfson’s argument that it isn’t correct:

    To make a difference, it needs to be that a threshold for climate change harm that was crossed with your act would not have been crossed without your act – or, that it would not have been crossed at approximately the same time. But that, Budolfson argues, is essentially impossible, because there is an “entire planet of other people” constantly engaging in emissions producing activities. So, if a given threshold is crossed at a given time with your joy ride, then it would still have been crossed without it, and at essentially the same time. In other words, even if your joy ride emissions pushed things over a given edge, this edge would still have been crossed at essentially the same time without your joy ride, because there’s always more emissions coming simultaneously/right after yours. (Budolfson’s paper, pg. 36) Note that this point does not require (as Scott Forschler’s might?) any assumption that others won’t reduce their emissions until a certain breaking point; it just requires the clearly true point that many others throughout the world will be emitting at the same time and right after your act. So any given emissions tipping point will be crossed around the same time with or without you.

    I discuss this argument in an in-progress paper, “Climate Change and Inefficacy”, and there I say that either Budolfson’s point shows that your individual joy ride has essentially no chance at making a difference for the worse, *or* at least it can be used to show that the chance is extremely tiny. (My paper argues that the expected utility view faces a problem either way.)

    All that said, I completely agree that individual actions non-trivially causally contribute to climate change harms, and that this gives us an obligation to reduce our emissions. But I don’t think the right way to think of this is in terms of individually making a difference.

29. Dear Julia,

    Thanks so much! [I appreciate comments any time!] This is a very clearly expressed objection.

    I am a big fan of Mark’s work. But I think (and it sounds like we might agree on this) that it’s only a more sophisticated of WSA and EK (and others). Responding to Kagan and others, Mark’s earlier main example is that the agricultural system contains all kinds of redundancies, and multiple orders of decision-makers, so my avoidance of buying meat on any given occasion is unlikely to shift all of the thresholds at different levels at the same time (e.g. the restaurant isn’t very sensitive to one beef order on a night, and even if it is, the butcher isn’t very sensitive to slightly changed orders from the restaurant, and even if she is, the farmer isn’t very sensitive to slightly changed orders from the butcher, etc.). I’ve talked to Mark about this and he agrees that his claims are a way of saying that the marginal effect of avoiding beef on any given occasion is *more often* nothing than is commonly supposed–and definitely more often than Kagan appears to believe–although that is not to say that there are no cases that it does have a positive marginal effect. I’m sympathetic, but that’s not a qualitative/conceptual point, just a quantitative one.

    While I disagree that this is really an objection to Kagan et al.’s threshold argument (multiple thresholds are *more complex* thresholds, not something sui generis), it’s worth saying that this isn’t relevant in the climate case. There are not a bunch of gatekeepers or decision-makers between my choice to emit and potential slow- and fast-onset climate effects (assuming we hold fixed some things like people already live in places where climate effects could harm them).

    So we can put that point aside. However, having put that point aside, Budolfson’s point is not different than Kingston and S-A’s one. In Budolfson’s words: “in the real world you know that reducing your greenhouse gas emissions will have some very small effect on the level of such gasses in the atmosphere. Do these differences make for a difference in what you are required to do?” and, as you phrase it “So, if a given threshold is crossed at a given time with your joy ride, then it would still have been crossed without it, and at essentially the same time.” Putting aside what I think is a mistaken assumption (contra Budolfson, China and the EU and many other major countries *are* acting, and the people within them are also a fortiori changing their behavior so Budolfson’s ‘Non-Cooperation Principle’ isn’t triggered), this is another version of the article that K/S-A offer on page 177: “if a single joyguzzle had not occurred and emitted 14 kg of CO2, another 14 kg of CO2 would have been emitted and would have begun to disperse only a fraction of a second later, and would have dispersed widely before any harm occurred. This suggests that, even if the partial causation theorists are right about their other assumptions, the overall amount of CO2 in the atmosphere would have reached the same levels only a fraction of a second later than it would have with this particular joyguzzle.”

    While you are absolutely right that my post did not address this point, I would take the same line that Broome takes, which is that ‘very small effects’ and ‘essentially the same time’ and ‘a fraction of a second’ are all ways of saying that there is a difference, but–at least in expectation–it is small and negative. That is enough to undermine No Difference. Broome (2019, p. 117-8) makes the same point, saying “Provided concentrations of carbon dioxide in the atmosphere are increasing, it is true that the effect of cancelling a small act of emission is to delay slightly the time when each particular concentration is reached. It will also lower the peak concentration, if a peak is eventually reached. If we ignore the atmosphere’s instability, the effect is to delay slightly the time of each harm that in due course will result from global warming, and if a peak is reached to prevent some harms altogether. All this implies that cancelling the emission makes the total harm suffered at each time less than it would have been. The total benefit of cancelling the emission is the integral over all future times of the reduction in harm at each time. This integral is what is measured by the social cost of carbon.” As an addendum, he adds that “All of this discussion of Kingston and Sinnott-Armstrong’s second argument ignored the atmosphere’s instability. Given that the atmosphere is actually unstable, the effect of cancelling an emission is unpredictable in its nature, size, and timing. All that can be said is—as before—that it diminishes the expectation of harm, and this is what matters.”

    In many contexts, what individuals do has only very little contributions to global problems. It may well be that this contribution is more effectively spent helping generate systematic changes–but reducing greenhouse gas emissions does help *somewhat* to preventing climate harms, at least in expectation. That’s enough to show No Difference untrue.

    I’d be more than happy to hear more (and also read a draft of your piece in case you are interested in sharing)!

30. PS Let me add that I have learned a lot from reading your work on this topic. In your terminology, it might be that climate change is an example of perceptible or imperceptible differences. But the new work this post (and the paper extending it) does is point out that this can’t be determined a priori–and if it is determined a posteriori then that case needs to be made, not just assumed. Saying the system is large, or that there is only some fraction of a second difference, is not enough (why does that privilege personal emissions as opposed to familial or regional or, indeed, subpersonal?).

31. Hi Kian, thanks very much for your reply!

    It’s been a while since I read the Sinnott-Armstrong and Kingston paper, and I forgot that they made a similar argument to Budolfson. Thanks for pointing this out! I’m not sure I agree that Broome’s point that a difference of a fraction of a second is still a difference actually replies to the argument. We already knew that a single joy ride makes a difference to some things – e.g. it makes a difference to exactly how much C02 is in the atmosphere. The question is: does it make a morally relevant difference in harm? I’m not sure that an imperceptibly tiny fraction of a second difference to the timing of a harmful outcome counts as a morally relevant difference in harm. (The stuff about atmosphere instability is another matter though. I think Budolfson has a great answer to this, but I don’t want to quote him about it here because it just came up in an informal conversation.)

    I do completely agree with your point that these questions about whether you individually make a difference cannot be settled a priori. (I tried to argue that in the paper I have on Kagan’s argument – I ague there that the a priori expected utility argument does not work.)

    In the end, my own view is that we don’t need to settle this ‘will I individually make a difference’ question (a complex and largely empirical question) to know that one’s emissions-producing actions have a non-trivial negative causal influence on climate change, and thus to know that there is good reason (and indeed, a moral obligation) to reduce one’s emissions.

    In any case, thanks again! And I’d love to read your paper when it’s ready for viewing.

32. Thanks for that! I will send you the paper when ready (it’s very close to what you’ve read, so shouldn’t take much time at all). And I’d be happy to swap for anything related you’re currently working on.

    As for the takeaway, we can all happily agree that emitting will have non-trivial negative causal influence. That does indeed seem overdetermined. I appreciate you coming by and contributing to this discussion~