W.W.J.R.D. (What Would John Rawls Do?)
I’m not often inclined to retreat behind the veil of ignorance to decide policy questions, but every now and again it makes sense to take out one’s less-favored philosophical tools and see how they handle a difficult problem. Today, we’re going to take global warming behind the veil of ignorance, and see where it gets us.
Why do I want to retreat behind the veil of ignorance on this problem in particular? Mostly because I think the two big moral accounting methodologies being brought to bear by the two sides in the debate – the Precautionary Principle on the one side and Cost-Benefit Analysis on the other side – each have serious deficiencies. So I thought it might be interesting to come at the problem from a different direction.
The Precautionary Principle (PP) has some very fundamental problems. The most fundamental, I think, is its inability to define what constitutes an action. Stated most simply, PP dictates that there should be a strong presumption against taking actions that have extremely severe negative consequences out in the tails of the distribution of possible outcomes. But you can only apply this rule if you know what constitutes taking an action as opposed to refusing to take action.
Sometimes this is an easy question to sort out. If, for example, we decide not to build a superconducting supercollider for fear of destroying the universe, we might be making a stupid decision – this particular risk is really far out on the tails – but we’re pretty clearly applying PP correctly. But sometimes it’s not so clear what constitutes an action, as opposed to a refusal to take an action.
Here’s an example to illustrate what I mean. We are involved in a war in Iraq. Somewhere out on the tails of the distribution of events that could follow a pullout from Iraq are some pretty bad consequences. It’s also true, though, that out on the tails of teh distribution of events that could follow continuing the war in Iraq are some pretty bad consequences. This kind of thing – dueling nuclear war scenarios at the ends of long chains of causation – should be familiar to high school debaters. Assuming that both pulling out and not pulling out are actions then PP would dictate that you must both pull out and not pull out. The only way to resolve this is either to weigh the tails against each other (which, as should also be familiar to high school debaters, is extremely difficult to do; at the end of a long chain of causation, the massive uncertainty about both probability and impact swamp any attempt at comparison), or to identify something as the status quo that gets presumption.
But this second option is also not easy to do. Which is the action: leaving Iraq or staying? Leaving means an end to a host of actions that we have to take by staying; it is a decision that, at a certain level, puts an end to the need to make decisions. As such, the presumptive posture should be: leave, unless there is a clear benefit to staying. But as we are currently engaged in a war in Iraq, the decision to stay is a decision to maintain the status quo; the presumptive posture should be: stay, unless there is a clear benefit to leaving.
Of course, we could decide the Iraq question without reference to PP, and just look at the expected costs of staying or going (as best we can calculate them) and make the call. But we can do that for global warming as well. PP is invoked because cost-benefit analysis doesn’t weight more remote scenarios very heavily, and under many cost-benefit analyses drastic action to stop global warming doesn’t appear to be worth the cost. Jim Manzi has written a great deal about this kind of analysis; see here and here and here among other places. (I’m not endorsing his analysis here, just citing it.) So how does PP work out applied to global warming?
I think it has the same kinds of problems as my Iraq example above. Advocates of the PP approach argue that we must not continue burning fossil fuels because we shouldn’t radically alter the earth’s environment when it’s so unpredictable what the consequences could be; out at the tails of the distribution, human civilization itself is imperiled, and not quite so far out there is massive death and destruction. Now, this might be a plausible line of argument if we didn’t currently have a fossil fuel-based economy. But we do currently have a fossil-fuel based economy. And drastically curtailing economic growth also has very severe consequences way out at the tails of the distribution. Where does presumption lie? Which is a positive action – burning fossil fuels, or taking legislative action to end the burning of fossil fuels?
Again, this is not the only problem with PP. PP would probably dictate not building more nuclear power plants. That, however, might mean greater dependence on fossil fuels for a longer period of time. But this would be a perverse consequence of correctly applying PP, not a fundamental problem with applying PP itself. I think there are fundamental problems with applying PP itself when you talk about global warming, because global warming is believed to be the result of pervasive and ongoing activities of our economy, and drastic action to change the nature of our economy is, I would argue, a significant action which itself would need to pass PP to be justified. And it probably can’t be.
So much for PP. What’s the problem with cost-benefit analysis (CBA)?
Well, the biggest fundamental problem with CBA is how you account for benefits to people who do not currently exist. To take a simple example: I am debating whether to have one or two children. If I have one child, that one child will inherit my entire estate. If I have two, they will split it. Assuming I am not sure whether I would be happier with one child or two, and that I am not sure whether my child will be happier with a sibling or without, and that I am not sure whether the benefit to the world from my brilliant offspring would outweigh the cost to the world of their resource consumption, I can reduce the decision to this one question. Clearly, CBA dictates I should have only one child, so that child gets the undivided benefit of my estate. (Note: this does not pose a problem for PP because PP is structured as a binary decision principle. You don’t need to know how to weigh impacts on future generations against anything; you just need to determine that severe negative consequences are possible.) There are a variety of responses to this problem, none of which I find entirely persuasive, but even granting a solution to this problem I think it is generally agreed that CBA imposes a presentist bias on analysis, since we’re generally debating whether to incur costs in the present to secure benefits in the future, and anything that happens in the future, when discounted back to the present, gets smaller.
Beyond this, there are a variety of other problems attributed to CBA, among the most common cited by folks on the left side of the spectrum is that CBA ignores distribution effects. This can be observed very clearly with respect to global warming. Jim Manzi cites an estimate that global warming could costs as much as 3% of global GDP in the next century. On the one hand, that’s a big number; global GDP now is $66 trillion, and in a century it’ll be much bigger than that. On the other hand, given that the costs of abandoning the fossil fuel economy are enormous, and that those costs are born up-front rather than a century from now, it’s pretty easy to do a calculation that shows that the cost of stopping global warming vastly exceeds the benefit when discounted back to the present. Far better to allow faster economic growth, and divert some of our resources to adaptation strategies and “big bang” solutions to transitioning away from a fossil-fuels based economy (as, indeed, Manzi favors).
But the distribution effects masked by this analysis are potentially huge. Bangladesh, a poor, overpopulated country near the equator and substantially at sea level is a country that is massively vulnerable to the impact of global warming. The Bangladeshi economy is about 0.3% of global GDP, on a purchasing power parity basis. Assuming a static relative position in the world economy over the next century, a forecast of a 3% loss of global GDP is entirely consistent with the complete obliteration of the nation of Bangladesh, ten-times over. We can recall, in this regard, that the economic consequences of Hurrican Katrina were basically negligible, but the failure to prevent the destruction of New Orleans is still viewed rightly as a massive indictment of the relevant public authorities.
For these reasons among others, CBA is rightly viewed as a somewhat suspect tool. It appears most applicable when purely economic questions are being debated; the further afield we go from these sorts of questions, the more suspect CBA appears.
So where am I going with this? Behind the veil of ignorance!
How would a Rawlsian look at the moral question at the heart of the global warming debate? He would say: behind the veil of ignorance, you don’t know who you are. So you should make your decision on a risk-averse basis, as if you are the person most seriously affected by the decision in question.
That person is probably somebody living in Bangladesh. Bangladesh is, as noted above, a country that is exceptionally vulnerable to global warming. But it’s also a country that is very poor (45% below the poverty line, $1,400 GDP per capita on a PPP basis), and hence stands the most to benefit in absolute terms from the continued advance of global economic growth. Conducting a CBA from the perspective of a Bangladeshi answers a lot of the criticisms of CBA. Because Bangladesh is especially vulnerable, the tails of the distribution will be closer to the center for them than they will be for the world at large, so you can avoid a problematic binary rule like PP without feeling like you’ve ignored the severe outlier scenarios. As well, you’ve accounted in a certain fashion for distribution effect questions; by taking the perspective of someone most likely to suffer from downside consequences, you avoid making a decision that will benefit the world on the backs of severe harm to the most vulnerable. Finally, you’re getting the primary benefit of CBA – a serious attempt to account for both sides of the ledger on policy questions – because Bangladesh is especially vulnerable to both downside scenarios, a global economic slowdown and global climate change.
And, to boot, this means we can delegate the entire global warming question to Reihan and his family.
I want to stress one thing in closing: this is not a post about the practical difficulties in attacking global warming. Manzi’s points about the likelihood of regulatory capture, among other things, are very well-taken. All I’m talking about is the moral framework within which we should be making these decisions. Anybody else feel (as I do) like this kind of quasi-Rawlsian framework makes more sense for global warming than it does for, say, setting the level of welfare benefits?
This is very helpful, Noah, though I’m still sorting through the implications — especially the implication that when you go behind the Rawlsian veil of ignorance you find . . . Reihan. Weird.
— Alan Jacobs · Apr 11, 02:11 PM · #
Applying Rawls to environmental issues seems problematic for a couple reasons. Do animals get to spend any time behind the veil, or is human welfare the only welfare we care about? Even if we decide to include individual animals in our welfare calculation, it seems like other properties which seem good in and of themselves, like the survival of species and ecosystems, are completely ruled out of Rawlsian thought.
We might conclude that the status quo in Bangladesh is so desperate that a radical five-year plan of global economic transformation and ecological terraforming looks reasonable. As a matter of cosmic justice, if we ask how much Americans should lose if it improves things for Bangladesh and places of similar hardship, the answer is a lot. But if the question is how many rainforests or how many species of cetacean we should be willing to lose, well, I just have no idea how I could answer that.
— Consumatopia · Apr 11, 03:27 PM · #
Consumatopia: I think the problem you’re articulating is much bigger than Rawls. I don’t think any liberal/utilitarian framework of rights and interests can really accommodate non-human, or at least non-sentient, participants. If, behind the veil of ignorance, you don’t know whether or not you’re a lump of granite, then I don’t know how you make decisions about anything at all. (On the other hand if the panpsychists are right, then all matter participates in consciousness to some degree, and so plausibly all matter has interests. But then, the law of conservation of matter and energy means that, in aggregate, those interests can’t be harmed. So nothing matters!)
You say that species and ecosystems are “good in and of themselves” but I don’t think this is the case. Obviously, I think they are good – but they are good because of what is good for humanity, not because there is some transcendant good about them. It may be good for humanity – spiritually, not just materially – to live with a small footprint, to recognize our dependence on a larger ecology, to seek a harmonious and stable existence rather than a dominant and expanding one. Or it may not. But that’s a debate about “the good” from humanity’s perspective, not about the rights or interests of trees. The smallpox virus does not have a right to life, and neither does the territory known as Bangladesh have a right to remain above water. Even the Jains, who abjure harm to any living things, don’t have a program for making tigers vegetarian, so their philosophy is implicitly human-centered in that it is founded on the assumption that we, being able to control our nature in a way that other beings cannot, have special obligations.
As for how many rainforests or cetaceans we should allow to die to promote human interests: that is a question you have to answer one way or another, isn’t it? The question doesn’t go away because it’s hard to formulate an answer.
— Noah Millman · Apr 11, 03:59 PM · #
The problem is present in all utility frameworks, but seems particularly acute with the veil of ignorance. If I’m maximizing aggregate utility, I could apply different weights to different “persons” depending on what species they are a member of, and then we could determine how many chimpanzees or dogs it’s okay to kill to save a single person.
But if you must apply a veil of ignorance, the only possible answers are “zero” and “infinity”—either the entity is a person and cannot be sacrificed for another person or it is no different than granite and must be sacrificed for another person wherever possible without hesitation. Neither of those answers seems plausible to me.
I worded my objection about rainforests and cetaceans poorly, but what I mean is that it seems like the sort of question that would require deep thought about what precisely our moral values are, whereas the veil of ignorance just sweeps the issue away as if it’s nothing.
Precisely what I find admirable about the veil of ignorance when it applies to humans—that the least counts most—becomes a deal breaker when it stretches beyond humans. It forces me to select an arbitrary black or white dividing line between persons and non-persons. Otherwise I’m stuck applying maximin to the stones of my driveway.
— Consumatopia · Apr 11, 06:19 PM · #
This is a pretty heavy thread.
Consumatopia, you say that:
Accepting that I don’t really know what “cosmic justice” is, and we’re not being precise about “a lot”, I think I get your drift. But, why? I mean, virtually no society that I’ve ever seen is organized on these lines, and only exceptional individuals live this way. Pretty much any time some group has tried to organize a society beyond the very, very small along the lines of “to each according to his needs”, it sure seems like it ends up being a disaster. At some point, if a stated desire runs directly contrary to human nature, doen’t this become sort of like wishing away gravity?
I mean these as actual questions, by the way.
Noah:
Pretty awesome point about extending the “I speak for the trees, for they have no voice” argument all the way to quarks.
— Jim Manzi · Apr 12, 01:38 AM · #
I don’t think you need “to each according to his needs” to argue my point, I think you need “might makes right” to argue against it. When one American is poorer than another, the reason is likely a combination of discipline, wisdom, and luck. But when Americans as a whole are so very, very much richer than inhabitants of some other countries, the reason has more to do with power and force. Power and force will always be with us, but that doesn’t make them just.
I suppose I would be close to a panpsychicist in that I think it plausible that all adapting patterns have consciousness related to the complexity of their adaptations. Conservation of matter and energy does not imply conservation of such patterns.
— Consumatopia · Apr 12, 04:04 AM · #
Consumatopia:
I agree that the way to go after the “I speak for the quarks” argument is to argue that what’s lost is information, pattern content or whatever when we rearrange matter-energy. Of course, this begs the question of why Person A should value some existing set of patterns more than some alternative set of patterns.
You say that:
This is surely true in part, but it also has a lot to do with different (and in material terms, superior) forms of social and political organization in the US (and Europe, Japan, etc.) than in poorer parts of the world. My question was really motiviated not by this, however, but by a kind of practical question: what would somebody propose to remedy this perceived injustice that wouldn’t create worse problems?
— Jim Manzi · Apr 13, 05:29 PM · #
I apologize if this is a double post, but my last attempt came up blank. (Too long re-editing after preview?)
That’s a fair question on practicality, but I intentionally phrased my statement as a conditional to avoid the practical question, since it wasn’t really my main point in this thread. (I’m anti-veil here, not pro). Still, I suspect additional foreign aid would give us more credibility when suggesting social and political reforms. So even if we thought that all poverty was due entirely to social and political organization problems, that still suggests an avenue for sacrifice. (I’m picturing some sort of Gandhi-style protest where libertarians begin a fast abstaining from subsidized farm products as part of a campaign for liberalized trade and other Pareto improvements.)
Well, from my perspective, that’s sort of like valuing some existing set of people more than some alternative set of people. I would rephrase it as asking which of two colliding patterns the evaluator should intervene on the side of. If the apparent course of our industrial system is to destroy our ecological system, when should we intervene? Since the life dependent on the latter is a superset of the life dependent on the former, this isn’t really the hardest case—we would like to preserve both systems as far as we can.
But in the general case it’s going to be a matter of aesthetics, analysis, empathy, humility and prudence. Value is an emergent phenomenon of the world rather than being a neutral place from which the world can be evaluated according to calculable axioms.
The complexity and apparent subjectivity here leaves us a lot of room to cheat and pretend that what is convenient to us (but destructive of other patterns or people) is good. Therein lies the usefulness of more parsimonious moral systems like the veil of ignorance, utilitarianism, or classical liberalism—they are all incomplete, but they all serve as heuristics built on time-tested intuitions keeping us sort of honest. However, I think they break down once we reach the scale of entire ecosystems and species of large, sentient mammals. (Keeping in mind the Thomas Nagel essay “What is it like to be a bat?”, we’re not merely losing consciousnesses, we’re possibly losing entire unique types of consciousness.)
In practice, though the philosophical issues are deep, I think we can just tack on an additional heuristic of respect for biological and cultural diversity and that would work reasonably well enough. The sorts of patterns worth protecting are going to be too complicated to be fully grasped by our limited analysis, so in humility we must adopt such heuristics.
I should say that given how many of the world’s poorest people are particularly dependent on functioning ecosystems for their livelihood, I’m not actually sure I’d disagree with human-centric Rawlsian ecological conclusions.
— Consumatopia · Apr 13, 08:47 PM · #
Consumatopia: I think I can agree with the bulk of your last comment. Value is indeed an emergent phenomenon, and I would indeed dearly like to know what it is like to be a bat. (Right now the closest I can come is reading The Bat Poet by Randall Jarrell – which, as it happens, is very good, but probably not much like what it is actually like to be a bat.)
Though I do think the libertarian mass-protest against ag subsidies would have to involve iPods in some fashion.
— Noah Millman · Apr 14, 02:13 AM · #
I’ll definitely have to that bat book out at the bat library, some bat time soon. It’d be cool to know what it’s like to be a bat, but I guess in the meantime I’m glad bats can know what it’s like to be a bat.
— Consumatopia · Apr 15, 03:23 AM · #