How to Allocate Scarce Health Resources

This paper reviews existing responses to the QALY trap and argues that all are problematic. Authors argue that adopting a moderate form of prioritarianism avoids the QALY trap and disability discrimination. Keywords: QALY, disability, priority setting, health care, prioritarianism.

Paper 2. How to Allocate Scarce Health Resources Without Discriminating Against People With Disabilities by Tyler M. John, Joseph Millum, and David Wasserman


Jack Freer May 5, 2020 at 3:49 PM

JMW focus on the "QALY trap" in their critique of CEAs. I also have a more fundamental problem with QALY estimates in CEA or quantitative decision analysis. Even when estimated by means of Time Trade Off or Standard Gamble (see for a nice explanation of these strategies), they are still fundamentally value judgments. Value judgments are, of course, personal. JMW say, "Suppose that the quality of life score of paraplegia is 0.55 and the quality of life score of ankle instability is 0.80". They then go on to compare Yin and Sam in their QALYs based on this supposition. But why do we assume both Yin and Sam place the same value on paraplegia? or ankle instability? or hemodialysis, blindness or deafness? These calculations may be helpful on a macro/societal scale, but there is a wide range of tolerance for life with a disability or other impairment. Applying the same coefficient to each of their expected survivals doesn't make sense.


Stephen Kershnar May 9, 2020 at 4:21 PM

Jack, I'm not sure I see your objection. Laws are crude instruments. For example, the deterrence effect of an anti-battery law will vary with the wrongdoer. We are trying to seek a result that makes the system as a whole work as well as possible. As long as the valuation gets the average correct, the micro-errors are unjust or unfair, but part of the price of having a general system using coercively obtained resources (for example, taxes).

A private medical system allows people to negotiate for things based on their idiosyncratic valuation of them.

Here, then, is my question: Do you see the problem of valuation as a reason, although perhaps not a sufficient reason, to move away from socialized medicine?

For example, in the U.S. roughly half of health care spending is paid by the government.


Jack Freer May 9, 2020 at 9:14 PM

Steve, I think QALYs have limited applicability to individual treatment decisions (including allocation of scarce resources) because of fuzziness in both parts of the QALY calculation. The first (high variability in value judgments) means any quality discount rate (such as "paraplegia = 0.55 of a normal life") is only a population average and not necessarily applicable to this patient. The second part of the calculation (life expectancy) is also fuzzy. Actuarial tables give the broad survival data (, but this is, again, too crude a measure to apply to an individual with a particular set of medical realities.

Regarding your socialized vs. private medicine medicine question, I'm not sure what you mean by each term. The current US system is not "private" in the sense that individual patients negotiate the price of the care they receive. Virtually everyone either has health insurance of some stripe (MC, MA, or one of many private plans), or they are unable to pay and go into debt. The only choice (or negotiating leverage) an individual has is when they are allowed to choose between private plans. The insurer then has the last word in payment decisions.

"Socialized medicine" means different things to different people. Most would include UK (with a centralized NHS paying for and providing that care), and Canada (essentially, "MC for all"). The common feature seems to be universal coverage.

The issue of "who pays for our health care" is convoluted. When hospitals provide care for uninsured patients, somebody has to pay for that. In many cases, that turns out to be Medicaid (the government) since the patient will now eligible by virtue of owing so much to the system. The two large existing forms of US socialized medicine (Medicare and the VA) actually run pretty well.


Stephen Kershnar May 10, 2020 at 11:55 AM




Thank you for the response. Many good points.

(1) You argue that QALYs are not very useful for individual treatment decisions because they are rough averages. I am not sure I agree with you.

First, I think they are helpful to guide production and distribution decisions. Consider, for example, rationing and how to allocate medical spending in different areas. If you don't like QALYs, what would you use instead?

I don't see the problem here as unique to medical goods and services. Worker's Compensation rely on averages. So do the criminal law punishments, minimum wage laws, earned income tax credits, and age of consent laws for driving, drinking, and joining the military. These try to get the proper incentive, punishment, quality of life, etc. for a population. How is this any different?

Second, imagine a patient is trying to weigh having one surgery versus another or having no surgery. She asks you, her physician, for advice. You do not know her well (even if she is a long-time patient). She does not know how the different health outcomes will affect her long-term health and well-being. Your advice should be keyed into QALY measurements, which are averages, because this is the best information you have available. I'm curious whether you think the physician-patient consultations provide better guides to the expected outcomes for the different options.

(2) By "socialized medicine," I mean "government owns or controls the production and distribution of medical goods and services." Insurance is free market medicine to the extent it is not owned or controlled by the government.

Out of pocket spending allows people to purchase goods or services according to idiosyncratic valuation. Consider houses and cars. The idea is that a free market in medical goods and services would do the same. Insurance makes this less affordable. A single payer system prevents it altogether.

(3) You argue that Medicare runs pretty well. In my view, Medicare is a financial mess. Here is my understanding.

Medicare is currently running in the red (it has accounting credits similar to social security but that money has already been spent). It will use up its accounting credits by 2026.

It has $42 trillion in unfunded obligations ($129,000 per person).

The U.S. debt is $25 T (118% of the GDP).

Why do you think that Medicare is not a financial disaster?

I hope life is treating you well,

Steve K


Bob Kelly May 10, 2020 at 12:43 PM


I wonder if JMW could just say that we need to assume there is some true theory of value, and then we would plug in whatever the true values are for each life. It's easy to enter in numbers for the sake of argument to show how the principle would work with some values plugged in. This seems analogous to economists using monetary values to theorize about choice and preference patterns. But I don't think they need to be saying that all paraplegia is (or would be, or ought to be) judged equally bad by everyone. Perhaps there is some objective value or disvalue to add in, perhaps some subjective experience of goods or bads (which maybe produces more objective value or disvalue), and perhaps any number of other value-relevant factors. I think all they need to say is that, assuming we got the values right, here is what a calculation using this principle might look like. They assume that paraplegia is worse than ankle impairment. That seems reasonable. The idea, I think, is to take intuitive cases to secure the ordinal values (x is worse than y), and then plug in some arbitrary numbers to perform the calculations.

Your point is taken that we don't know what exactly the formula is for calculating the value or quality of a life. But it doesn't seem like a criticism of the QALY principle that it involves value judgments. This just seems to be part of the definition of adopting QALY. The *quality* that needs to be determined is the *value* of the life(-years). If your point is to say that that is hard to do, you are likely right. But it doesn't seem to show there is anything wrong with QALY in principle; only that we aren't certain what the true value theory is, and how exactly to weight up the relevant variables, whatever those turn out to be. The true value theory will dictate the correct outputs of a QALY principle, and we'll have to get as close as we can being limited epistemically as well as with respect to resources. This is not an endorsement of QALY. It is just to say that charging the QALY theorist with invoking values is not going to raise any special problem for them. Though asking for as much of the formula as possible for calculating the value of a life(-year) is the right question.


Jack Freer May 10, 2020 at 7:53 PM


(1) You say QALYs are "helpful to guide production and distribution decisions." I agree. I said I didn't think they were useful for individual treatment decisions.

As you point out, averages are valid measures to guide a host of policy decisions. They are also useful for explaining medical outcome data to patients.

Using your example of a patient considering surgery A, B or no surgery, the first question is the mortality rate of either surgery. That number can be estimated and provided to the patient. If one of the surgeries is associated with a 10% chance of a stroke, that can also be explained to the patient. I don't think it's helpful, however if I tell the patient, "on average people feel a stroke would diminish remaining life to 80% the quality of their prior life." Rather, I would explore that particular patient's attitude toward the possible stroke

(2) A single payer system (in which the hospitals and doctors remain private) would not be socialized medicine by your definition.

I agree that a free market in medical goods and services optimizes idiosyncratic valuation. It also carries with it enormous disparities.

(3) Yes, the Medicare Part A (hospital) trust fund is being depleted. Because Medicare covers the elderly and disabled, costs are higher than for populations with large numbers of healthy young people. All bets are off on what the future holds for the economy and health care during and after the pandemic.



I agree that it is possible (and a useful exercise) to estimate the individual's valuation of a particular outcome. The Standard Gamble approach hypothesizes a choice between a certainty and a gamble. The certainty is the remainder of the patient's life with the impairment in question (such as a stroke). The gamble is a possibility of a normal life vs. death. The odds are them manipulated to find the patient's point of ambivalence between the certainty and the gamble.

If the gamble side carries a risk of death of 1% (and a normal life of 99%), a person might jump at the gamble (rather than the certainty of life with a stroke). Conversely, a gamble with a 99% chance of death (and 1% chance of normal life) might lead that same person to choose the certainty of a stroke. The numbers are then manipulated (90/10, 80/20, 10/90, 20/80 etc) until the person is ambivalent about the choice. That number (let's say, 70/30) is how that patient values a life with a stroke (70% of a normal life). This is obviously a laborious time-consuming process best suited to research, not clinical care.


Bob Kelly May 13, 2020 at 10:34 AM


That is super interesting. I didn't know about that study approach for valuing life vs death vs unhealthy outcomes. It is very reminiscent of standard behavioral economics "games" that attempt to discover the ways that people value their options, what influences that valuation, and so forth. For instance, we know that people value an object much more if it was first given to them as a prize (e.g. they would sell a newly gifted item than they would pay to buy that same item). We also know that everyone discounts delayed rewards (e.g. the same reward is judged less valuable as its receipt is pushed further into the future). We also know that delayed discounting is hyperbolic rather than exponential, and that addicts' discounting curves are steeper than non-addicts.

I wonder if you think that perhaps this kind of research can help us understand what people's general preferences are, and so eventually be helpful in a clinical setting where that might be harder or impossible to figure out? I assume you'd be hesitant still. Perhaps that is the right attitude. This stuff is super hard and I often don't know where I come down.


Stephen Kershnar May 13, 2020 at 12:00 PM


You suggested to Jack that he consider this approach, "I wonder if JMW could just say that we need to assume there is some true theory of value, and then we would plug in whatever the true values are for each life." I think this is an interesting suggestion and, perhaps, might be filled out in terms of the Sen/Nussbaum or Peter Koch theories of capabilites.

My question is whether you think there is objective prudential intrinsic value (either in part or in full) independent of pleasure or desire-fulfillment.

Steve K


Stephen Kershnar May 13, 2020 at 12:09 PM


Thank you for the interesting reply.

You write the following, "Using your example of a patient considering surgery A, B or no surgery, the first question is the mortality rate of either surgery. That number can be estimated and provided to the patient. If one of the surgeries is associated with a 10% chance of a stroke, that can also be explained to the patient. I don't think it's helpful, however if I tell the patient, "on average people feel a stroke would diminish remaining life to 80% the quality of their prior life." Rather, I would explore that particular patient's attitude toward the possible stroke."

I wonder why you think the particular patient's attitude toward a possible strong is informed, especially when compared to statistical studies of the effect it has on people. This strikes me as similar to asking an average car owner how much he values a positive displacement supercharger on his Suburu Crosstrek. I don't see how he would know the answer to this. Nor would he know the effects on his well-being of a stroke other than through imagery that might be wildly inaccurate.

I think we disagree on socialism. You note the following, "A single payer system (in which the hospitals and doctors remain private) would not be socialized medicine by your definition." I don't think this is true if socialism comes about via government ownership or control. Here the government controls the production and distribution of medicine. The private ownership lessens the degree to which it is socialist but does not eliminate it, perhaps not even significantly.

In addition, if they are exclusively paid through the government, they seem no more a private physician or hospital than an independent contractor is independent when his hours, tasks, and pay is exclusively assigned from an employer.

Thanks again for the comments,

Steve K


David H May 14, 2020 at 7:09 PM

Jack, Steve and Bob,

I share Jack’s worries about QALY. Some of my concerns are in principle, some practical, some perhaps both. But I will start with what is the most clearly an instance of the first type. I wonder about value incommensurability of goods that Pat Lee and Chris Tollefsen explored in papers discussed at earlier Romanell working dinners. So, if there are incommensurable goods than there aren’t better or worse errors in QALY estimates and we can’t be content with “well, we have to ration and so let’s do as best we can” or “there is a right answer and let’s try to obtain it, recognizing we might only approximate it”. I don’t want to commit to myself to this view of the incommensurability as I don’t know enough about the topic of value pluralisms and the natural law framework that Pat Lee and Chris Tollefsen draw upon, just put it out there as a concern.

I not only share Jack’s worry the QALY generalizations will not be applicable to those outliers who don’t think condition x is n times as good as a year of health, but I worry a bit that people cannot even make good equivalence judgments about their own future when confronting transformational experiences that Laurie Paul and others discuss. Again, I am not committing myself to the view since I haven’t studied the literature of transformational experiences, just putting it out there.

While rationing is perhaps unavoidable given scarce resources, it is particularly galling to those given a death sentence based upon an evaluation of the well-being of their life that is not shared by the person fatally denied the resource, or an adverse judgment that can’t admit greater or lesser errors given value incommensurability, or based upon judgments that disregard our epistemic limits given transformational experiences; not to mention the belief I defended in an earlier post that it is wrong to sacrifice people by killing or letting them die on the grounds that others will obtain more well-being than they can from extending their lives.

Here is an analogy in support of the last claim: wouldn’t it be galling if you were a minority and denied a lifesaving resource because if saved, your life wouldn’t go as well as others due to racism? But if unfair social losses of welfare shouldn’t affect the decision, why should “unfair” natural losses of welfare be relevant? The difference is not just that racism is immoral and nature’s cruelties aren’t. We don’t think the NAACP have the right to demand charity from the racist even though we think it correct to judge his character harshly. But we do think the minority patient has the same right to receive the organ as anyone else, it doesn’t just reveal bad character to deny him the organ on racist grounds as it is to deny the NAACP a donation. So, there is a claim of the minority patient to be considered equally for the organ despite the life with lower well-being, and that type of claim belongs also to the patient who, if saved, will have lesser well-being due to natural disability.


Jack Freer May 15, 2020 at 7:38 AM

Bob, Yes, this kind of approach to complex decision-making in the face of uncertainty originated in the field of economics and is applicable to a broad range of decisions: First medical applications of quantitative decision analyses came out of Tufts (Pauker & Kassier: Being a quantitative methodology, it works best with a quantifiable value (such as QALY).

Steve, your point about medical decision makers (patient or surrogate) making informed judgments about options is well taken, but it really is a communications problem for the clinician. This is actually relevant to a lot of advance directives that focus on hardware and technology (tubes, ventilators, CPR) which are akin to your supercharged automobile example. Of course, most consumers are not familiar with such details. On the other hand, if car dealers had a fiduciary responsibility (which they do not), AND were good communicators, the discussion would focus on the consumer's values, expectations, preferences, risk tolerance etc. The dealer would then make a recommendation for or against the supercharged Crosstrek based on that information. Physicians do have a fiduciary responsibility to patients and should be discussing the patient's values, expectations, preferences, risk tolerance etc. Based on that, the doc should explain how a particular treatment modality does or does not further those values. This is my biggest concern about advance directives (including MOLST) which allows (encourages?) patients to opt in or out of specific treatments (intubation, ventilation), I would argue, without sufficient background or context.

I will pass on commenting on your socialized medicine questions since it is getting pretty far afield from the focus article.


Stephen Kershnar May 15, 2020 at 9:38 AM


I think your point is well taken. Still, I don't see why the physician wouldn't introduce QALY-based studies. As consumers, we can use Consumer Reports, looking at bluebook pricing, and so on. Patients don't have these options and don't know what sort of well-being-based outcome they can expect. I think introducing these studies is a good basis for their decision. They may then opt out if they feel they have idiosyncratic evaluation.



I am not sure why incommensurability of value - whether prudential or intrinsic - is plausible. Is it plausible that for a person, Option A is neither better, worse, nor equally good as Option B? This seems odd. Do you think there is incommensurability of the right as well as the good? If not, I would be interested in hearing the basis for the distinction.

Also, consider the following.

"While rationing is perhaps unavoidable given scarce resources, it is particularly galling to those given a death sentence based upon an evaluation of the well-being of their life that is not shared by the person fatally denied the resource, or an adverse judgment that can’t admit greater or lesser errors given value incommensurability, or based upon judgments that disregard our epistemic limits given transformational experiences; not to mention the belief I defended in an earlier post that it is wrong to sacrifice people by killing or letting them die on the grounds that others will obtain more well-being than they can from extending their lives."

The language of a death sentence and sacrificed is odd in that the state is refraining from helping people rather than harming them.

The state does this all the time. Consider the exceptions to the draft. Consider, also, the massive taxes used to support various welfare and old age benefit programs or the sizable tax breaks. What is distinctive here other than the magnitude of loss?

I don't see why Laurie Paul's transformational experience point is relevant. We are bad at valuing lots of outcomes (for example, how bad would it be someone to become deaf?). This is similar to the issue of how much to value a positive displacement supercharger for a Suburu Crosstrek. Here we need studies. The QALY framework provides a way this can be done.

Steve K

Stephen Kershnar May 9, 2020 at 4:25 PM

Tylor John, Joseph Millum, and David Wasserman face the following jointly incompatible propositions. I restated these principles in a more concise manner.

(1) Quantity Matters. Other things being equal, it is better to save the life of someone who will gain more time than someone who will gain less time.

(2) Quality Matters. Other things being equal, it is better to save the life of someone who will gain more well-being than someone who will gain less in well-being.

(3) Transitivity. All things considered, if a is more valuable than b and b is more valuable than c, then a is more valuable than c.

(4) Disability. Other things being equal, if we are saving lives, disabled people’s lives should not receive less priority even when their disability results in (on average) their having shorter lives or lives of lesser quality.

Tyler John et al. adopt the following principle to adjudicate between these propositions.

(5) Prioritarianism. Other things being equal, it is better to benefit someone whose life goes worse than someone whose life goes better.

(6) Better Off. How well someone’s life goes depends on, and only on, his lifetime well-being.

Given these last two principles, the quantity, quality, and priority of a person’s life determines the value of a medical treatment.

Prioritarianism is not a satisfactory theory of the good.

Usually, the good is viewed in welfarist or non-welfarist terms. Welfarism asserts that the good is a function of, and only of, well-being. Non-welfarism asserts that the good is not a function of well-being at all or it is not a function of well-being alone.

This is not a welfarist theory because priority is not a welfarist function because it is trading off more well-being for less well-being.

Nor is it a plausible non-welfarist theory. Usually non-welfarist theories involve virtue and justice modifying the value of well-being or adding to it. Prioritarianism takes neither virtue nor justice into account. It says nothing about virtue. It ignores what a person has a right to or what he deserves. It says nothing about justice unless justice is understood in egalitarian terms.


Bob Kelly May 10, 2020 at 2:19 PM


Can you say more about why it is not, as you suggested, a non-welfarist view that takes into account justice in the form of egalitarianism? Alternatively, why isn't it a non-welfarist account that takes into account current life-value instead of virtue or justice (if it isn't either of those)? In other words, I don't really see what the objection is. If non-welfarism is the view that welfarism is false, then either MP is welfarist or it is not. They don't only take welfare into account (you claim), and so it is by definition a non-welfarist view. It's not clear why (1) this means it has to involve virtue or justice, or (2) why it doesn't involve the latter (egalitarian justice).

John May 13, 2020 at 8:33 AM

This is partly just a test comment (last time my comments didn't appear), but I am interested if the egalitarian justice point was addressed, perhaps below.

Stephen Kershnar May 13, 2020 at 6:26 PM


Good point. Here is my argument.

A non-welfarist theory does not consider well-being exclusively.

(1) If JMW's theory is a plausible non-welfarist theory, then it considers virtue or justice.

(2) JMW's theory does not consider virtue.

(3) JMW's theory does not consider justice (because it does not take into account desert or moral rights).

It does not take into account desert because it does not take into account that which someone is morally responsible for (or virtue).

Perhaps JMW might respond that people have a moral right to equality. I find this implausible because

(A) equality is too vague to be the content of a right (equality of what? Well-being, health, or opportunities),

(B) even if (A) were not true, equality is not a legitimate feature of the justice, rather it is proportionality to what people deserve or something along those lines, and

(C) even if (A) and (B), were false, as a theory of justice equality seems better situated in the context of the right rather than the good (for example, is it unjust as a type of intrinsic badness if Al has a very happy marriage and Bob has only a happy marriage?)

Thank you for pushing me on this.


Steve K

Stephen Kershnar May 9, 2020 at 4:37 PM


Prioritarianism is not a plausible consequentialist theory of the right because it does not set out a plausible theory of the good.

It is not a plausible non-consequentialist theory because it ignores (actual) contracts, desert, and rights. If, as a matter of justice, it were desirable to give priority to the worst off, this would rest on one or more of the following: what someone has a (pre-institutional) right to, what he deserves, or what he actually agreed to. None of these support prioritarianism.

The authors reject a Rawlsian argument for pioritarianism. Here are a few of the reasons to reject Rawlsianism in general. A hypothetical contract ignores people’s moral rights. More generally, a hypothetical contract does not by itself ground a duty. At best, it merely indicates what justifies a duty. In any case, it is likely that a hypothetical contract between free and equal individuals behind a veil of ignorance would justify a maximization principle. Even if it would not justify such a principle, it would not justify a principle that blocks consented-to Pareto superior moves. This is irrational.

If John et al. want to give instrumentalist arguments for favoring the disabled, they should do so in a straightforward manner just as people do for the types of race and sex discrimination that constitute affirmative action.

Stephen Kershnar May 9, 2020 at 4:50 PM


The problem is that the priority function is either context-based (it depends on the individuals being compared) or it is not. If it is context-based, then the Independence of Irrelevant Alternatives is violated. If it is not context-based, then it is mysterious how it functions. Here is a version of their principle.

(1) Medical Treatment Value = Quantity x Quality x Priority

Independence of Irrelevant Alternatives says that an irrelevant third choice should not affect the preferability of two other choices. Consider choice set [A, B] and set [A, B, C] where C is unrelated to A and B. If, for choice set [A, B], A is preferable to B, then, for choice set [A, B, C], A is preferable to B.

Consider these individuals in a two-person world and the priority function seen in (1).

(2) Al has 100,000 utils per life and a priority of 1. Hence, the value of his life is 100,000 priority-utils (100,000 utils/life x 1 priority unit).

(3) Bob has 50,000 utils per life and a priority of 2 (because he is the worst-off). Hence, the value of his life is 100,000 priority-utils (50,000 utils/life x 2 priority units).

Here Al’s and Bob’s lives get equal priority because they are equally valuable.

Now consider this three-person world.

(4) Al has 100,000 utils per life and a priority of 2. Hence, the value of his life is 200,000 priority-utils.

(5) Bob has 50,000 utils per life and a priority of 3 (because he is the worst-off). Hence, the value of his life is 150,000 priority-utils.

(6) Carl has 150,000 utils per life and a priority of 1 (because he is the best-off). Hence, the value of his life is 150,000 priority-utils.

In this world, Al’s life gets priority over Bob’s life. Earlier their lives warranted equal priority. This violates Independence of Irrelevant Alternatives.

Imagine instead that the priority function does not depend on context.

This would be odd because priority depends on how well other people are doing in comparison to one another.

Consider how much value Bob’s life has in a one-person world. If, in a context-independent valuation system, his life has 150,000 priority-utils (same as the three-person world), then it is mysterious what justifies his priority ranking of 3. A priority ranking is necessary even in a one-person world because there still might need to be a tradeoff between goods connected to welfare (for example, Bob’s well-being) and goods not connected to it, such as beautiful things.

Stephen Kershnar May 9, 2020 at 4:55 PM


Another problem is that the authors’ measure of well-being is mistaken. Consider these lives.

(1) Ecstatic Century. (100 years) x (1,000 utils/year) = 100,000 utils

(2) Repugnant-Conclusion Life. (100 billion years) x (0.01 utils/year) = 1 billion utils

If the authors’ measure of well-being were correct, we should strongly prefer the Repugnant-Conclusion Life. However, we do not and probably should not prefer it.

Perhaps a better measure is well-being per time. The problem then is that a super ecstatic minute of life (100,000 utils per year times one minute) should be preferred to an ecstatic century. This is odd.

Perhaps this is a problem for all theories of lifetime well-being, though, and not a distinct problem for this theory.

David H May 9, 2020 at 5:36 PM


I want to tentatively argue against allowing harm-based quality of life decisions that favor or disfavor the handicapped in life saving scenarios. So, it is a defense of a variant of the Equal Value to Saving Lives, appealing to a lexical ordering. The idea is an analogy to claims about the equal wrongness of killing being insensitive to differences in harm. It is widely considered equally wrong to kill the elderly or young, despite the former being harmed less by the killing as they lose out on fewer QALY. So, if we ignore differences in harm when judging killing to be equally wrong, why not disregard differences in harm and judge not saving equally wrong? A quick response is that the right to life is a right against being killed but not a right to be saved. But maybe there is a right to be saved, or perhaps there is a duty of doctors to save, or possibly just a reason to save or if one is going to save, to equally value saving lives. I will explore these options later in a later post as I just want to pursue here the analogy of wrongness of killing being insensitive to differences in harms to those who are allowed to die.

We don’t say someone’s right not to be killed weakens as the harm of death lessens with age, nor are punishments reduced with the harm of death. Shooting up a hospice center is as wrong as a mass shooting in the post office. Persons have a moral status and a right not to be killed because they are creatures of a certain kind, not because of actualized abilities or sufficient quality of life years remaining. Intuitively the same is true of elderly persons who need to be saved by rare medicine also needed by other creatures like young chimpanzees who have more to lose (decades of well-being) from a death than geriatric persons. So, there is a duty or reason to save even the oldest person over the chimp even though the former has a year of life left and the chimp two decades. Persons, unlike non-humans, are of such value that they can’t be killed to save others or be allowed to die to save chimp or two. It seems an offense to their dignity, not taking their moral status seriously, where it can be outweighed by harms to non-persons. (“Person” is just shorthand for members of our morally elevated kind, not tied to any particular metaphysical characterization like Locke or Boethius. Of course, I owe readers an account about what makes our kind so valuable and since it will be a biological classification why incapacitated or immature members share that elevated status.) What I want to suggest in this post that it likewise wrongs the disabled person if the decision to let her die is for the sake of saving another person is the latter will lose out on more QALY if not prioritized for scarce resources. Just as there’s a deontological threshold immunizing the elderly against killing when death isn’t a great harm, the disabled immunized against cost-effective analysis.

Nevertheless, I think we should save a human being’s arm before another’s finger because of difference in well-being. So why treat death differently? I think there is no devaluing equals when one treats endangered arms differently than fingers. Death differs because saving one life over the other will often make a judgement about the value of the whole person rather than treat differently the different harms due to the damage of parts of equals. There is an insult in destroying or allowing the destruction of the entire person that is absent when one prevents one person’s loss of limb rather than another’ finger. Persons are of such value that they are above the threshold where the wrongness of killing or letting them die is based upon and fluctuates with the extent of well-being lost. The wrongness tracks the kind of entity rather than their susceptibility to harm. Tying


Stephen Kershnar May 9, 2020 at 6:13 PM


Interesting argument.

Is status intrinsically valuable or does it have to multiply by well-being to make a person's life intrinsically valuable?


Consider a person with slightly negative lifetime well-being and desert. Could his life be intrinsically good even though he is better off not existing and deserves his sub-zero well-being?

(1) If no, then status is not by itself a basis for intrinsic value. Rather, we have to multiply (or add) status by well-being (Kagan's solution) But then I don't see that your argument works because we still have to multiply status by well-being and this might support the chimpanzees, children, and good young people over evil old people.

(2) If yes, this is strongly counterintuitive. It tells us that God would prefer to create a world populated only by double-negative people (negative desert and well-being) than a world without people.

This would be true even though the world would not be good for anyone.

I find this conclusion to be so counterintuitive that words fail to express my outrage.

Shelly Kagan in How to Count Animals uses the notion of status to explain why we can prefer people over animals. He would provide the "no" answer to the above dilemma.

I am interested in hearing you answer.


Steve K

Stephen Kershnar May 9, 2020 at 6:21 PM

People use dignity to explain why its wrong to torture criminals and terrorists because it insults their dignity. In response to questioning, they further claim that dignity is different from a person's intrinsic value, moral rights, desert, or what grounds them. I never understood what dignity is and why it should trump or compete against someone's intrinsic value, rights, or desert.

Here are my quick questions.

(1) Is dignity different from intrinsic value, rights, or what grounds them?

(2) If yes, then why think it trumps them?

Thanks for the interesting post.

David H May 14, 2020 at 7:11 PM


I take moral status to mean roughly that an entity or its interests morally matter to some degree for the entity’s own sake. That means the entity is entitled to some moral respect and so others can’t do certain things to such a person. Such creatures of great value may at times be miserable and it could be in their interest to die. But that doesn’t remove their value which is a value in them not for them. So, someone’s well-being may be quite low as they suffer greatly from some disease or torment but their value is quite great. They don’t lose their immunity or cease to deserve respect and consideration because they are miserable. You seem to believe that will commit me to the view that God should prefer a world of miserable human beings with great value and moral status to no such creatures. But why would God have reason to create a world of valuable creatures that are miserable? Is his only concern to have creatures with great value even though they would be miserable and wish they didn’t exist? It would seem like God is ignoring their moral status when he makes them exist in a world in which they are miserable. Being good, He wouldn’t do that and so He wouldn’t be faced with a choice of creating miserable creatures with great value and moral status or no one at all.

I am probably missing your point of hypothetical about God’s choices. Are you trying to get me to admit that a human being’s value is determined by how well his life is actually going and so if it is not going well the human being becomes less valuable and perhaps would be less valuable than non-human non-rational creatures whose life is going well with various pleasure? I would deny the human being became less valuable when his life’s prospects waned and thus doesn’t lose any immunities that the non-humans lack.

I think of a person’s value and well-being as different measures though not unrelated as I believe a creature’s value is dependent upon the type and degree of well-being obtainable. But while a creature may belong to a kind capable of great well-being and thus be of great value, that individual might not obtain that elevated well-being, nonetheless that doesn’t lower his moral status or great value and our concern with his well-being. The person’s interests and well-being matter because the person matters. The person matters even when miserable, unconscious, immature or damaged.

David H May 14, 2020 at 7:12 PM


I understand “dignity” to be the name of our value. We have this value because we possess property x or nature x. Our value consists in our having x. I will illustrate this with Kantian terms without committing myself to anything other than a structural analogue to Kantianism There are different types of value – one is named price, the other dignity. You have dignity because of your possession of humanity (rationality or moral rationality).

Stephen Kershnar May 15, 2020 at 10:11 AM



Here are your responses to the case of the person with slightly negative lifetime well-being and desert.

You assert that (1) a person's life be intrinsically good even though he is better off not existing and deserves his sub-zero well-being.

Similarly, you assert that (2) God would prefer to create a world populated only by double-negative people (negative desert and well-being) than a world without people.

I find these results counterintuitive.


You're also committed to God (3) preferring to create a world with many many miserable and evil people (double-negative people) rather than a world with a much smaller number of very happy good people (double-positive people).

In saying we should prefer that one person lose a finger than another lose an arm (other things being equal), you are introducing well-being-based value that adds or subtracts from status-based value. If I remember correctly, and I might not, this is Elizabeth Harman's solution. Perhaps you and Harman could use this to block results (1) and (2).

Now consider this claim, "We don’t say someone’s right not to be killed weakens as the harm of death lessens with age, nor are punishments reduced with the harm of death. Shooting up a hospice center is as wrong as a mass shooting in the post office."

As a theory of badness, this is inconsistent with the status and well-being based theory of value.

As a theory of the right, this depends on the stringency of a right being independent of the harm or loss of autonomy it prevents. This is odd. We normally think that one of these setbacks explains the stringency of a moral right. To suggest this is done at a population level requires a theory of why this is the relevant class of individuals. I don't know any argument for such a broad class.


Steve K

Stephen Kershnar May 15, 2020 at 10:23 AM



Here is what you say, "It seems an offense to their dignity, not taking their moral status seriously, where it can be outweighed by harms to non-persons."

I take it you think dignity is intrinsic value (you say "value," but I doubt you mean to include extrinsic value).

Does “status” refer to intrinsic value or the basis of intrinsic value?

If yes, then we have the claim, “It seems an offense to one’s intrinsic value, not taking their intrinsic value seriously (or, alternatively, not taking the basis of their intrinsic value seriously).”

But it does take their intrinsic value seriously, specifically, their well-being-based value.

The price to be paid for preferring to save one person’s arm over another person’s finger is to introduce a second type of value, specifically, well-being-based value.

Perhaps you want to say that the preference is a matter of the right and not the good.

But then the right floats unconnected to value – whether intrinsic or total – and is mysterious. Worse, the obvious explanation for standard triage decisions becomes mysterious.


Steve K

David H May 15, 2020 at 1:30 PM


I didn't say God should prefer a world of more miserable people with high moral status. I just said their moral status did not drop with their well being going south. You then translated that to mean God would prefer more miserable people of high value than no people. I said someone's moral status and how well their life were going were different issues. There is a difference between a value in them and for them. They can have a great value and not care about it or they could have a great value and rationally desire to die sooner than later because they miserable and devoid of most goods. Perhaps it is better for people to die than suffer greatly. That is compatible with their having great moral status and others having to do things for them and not do certain things to them. I also said God would be disregarding their moral status if He made them them miserable and He wouldn't do that. I don't think it helpful to try to explicate my view with how many people in what condition would God prefer. My view of moral status is tied to well-being in that only creatures of a kind capable of great well being have great moral status. So it doesn't float free of well-being as you charge, but it doesn't vary with variations in well-being However, their great moral status is there before (mindless fetuses) and they realize that well-being and after they no longer do (illness). The intuition pump for that if we had a scarce serum that would enable the terminally ill fetus or comatose adult to be healthy, then they get it before conscious animals who lack the healthy potential and interest in realizing the same potential for self-conscious rational life, though they would enjoy it once they possessed it. And we don't save many happy animals by killing an unhappy person. The alternative is very inegalitarian where the wrongness of killing depends upon the degree of deprivation and killing the middle ages is not as great as killing teenagers and perhaps the latter can be killed to save the former. My view of moral status and its connection to intrinsic value is tricky because the intrinsic/extrinsic connection is difficult. Is it the value someone would have if he was the only creature in the world and didn't have the history or future that he does? My views of the moral status of the mindless are tied to what is healthy development but that involves history and contingencies and so is not straightforwardly intrinsic. But it isn't extrinsic in that their value doesn't depend upon others' concerns for them or their affect on others. See Rose and my Res Philosophica paper Volume 95, Issue 4, October 2018 David Hershenov, Rose Hershenov Pages 693-718 which is also on my webpage I think triage in life saving is different from triage in limb saving. I want lives just favored because of surgical success, just as I want the right not to be killed to be independent of well-being lost as one can't practice target shooting in a hospice or with a suicide mid leap. Where it is a matter of limb and not life, then I am more open to differences in well being

David H May 9, 2020 at 5:38 PM


This egalitarian view about our value doesn’t mean that doctors always flip a coin when they can only save one. If it unlikely that one person will be saved but far more likely the other will be saved, then that is form of rationing that respects the equality of persons. It is akin to when the brakes of your car have failed and if continuing to to go straight will kill both people who are standing a few feet apart but you can turn one way and very likely kill the guy on the left or turn the other way and probably injure rather than kill the person on the right. So, rationing is made on medical grounds of who is more likely to survive, not who has more to gain from survival or who has had less across their life. This will mean discriminating against the elderly not because they have had more of good life than someone else or have less good life left than another, but just on medical grounds it is less likely they will survive live. It means triage discrimination against the disabled only if their disability makes it less likely they survive the procedure. It isn’t based upon levels of well-being the disabled will enjoy so doesn’t disvalue the newly disabled as does the authors’ prioritarianism. Nor does it favor disregarding the able-body person’s life in favor of the disabled who have had a worse life any more than it endorses you turning your brakeless car into the person who has had the better life.

David H May 9, 2020 at 5:55 PM


The egalitarian view that I have tentatively proposed seems akin to the lexical ordering position that Wasserman et al ascribe to Harris so I want to consider whether the author’s objections have merit and undermine my attempt to give equal value to saving lives. 1) The authors’ first objection is that just saving someone for a few days can’t have priority. But when we speak of saving what we mean is roughly akin to either curing or stabilizing. We won’t think a surgery that would keep someone alive until the end of the operation would be a successful saving of the life because the patient would have died at the moment the surgery could have begun but didn’t. So, if someone dies a few days later they weren’t really saved in the sense of being cured or stabilized or something similar to that. 2) The Aggregation Problem just seems like a sorites that would affect the Moderate Prioritarianism of the authors as well. The Aggregation Problem is that if we could save one person or prevent a horrible chronic condition of two or more, we would cure the latter. And then we increase the number but slightly lower the severity of the chronic disease and we end up saving many with little hardships. My response is that Moderate Prioritarianism is in the same slowly sinking ship. The authors allow that if the gains to the better off are great that we should bring them about if it prevents a slight gain to the worse off. Well, just gradually increase the numbers of the better off and slightly lower their benefits and we get a parallel problem for Moderate Prioritarianism and end up denying the worse off to obtain insignificant gains to the better off. Tu Quoque. 3) Lexical ordering is alleged to have a problem of individuating life-savings. Harris supposedly is committed favoring ten times saving someone’s life with procedures that each add two years over one life-saving procedure that adds twenty years to someone else. But isn’t this a pseudo problem? The view in question favors prioritizing keeping someone alive, not prioritizing the number of successful attempts at keeping someone alive.

Bob Kelly May 10, 2020 at 3:40 PM


JMW attempt to avoid the QALY trap. The QALY trap can be expressed as a kind of dilemma, where 3 intuitive principles (1-3) together entail a 4th counterintuitive principle (4) as follows:

(1) Equal Value to Saving Lives: it is equally valuable to save the life of someone with low quality of life as someone with high quality

(2) Value to Reducing Morbidity: all else equal, it is better to save someone’s life and cure her disability than save her life but leave her disabled

(3) Transitivity of Betterness Relation: if (A is better than B and B is better than C), then A is better than C

(4) Unequal Value for Equal Outcomes: it is better to save the life of someone with a chronic health condition who will then be healthy, than to save the life of someone already healthy who will then be healthy

JMW criticize extant attempts to reject (1), (2), or (3), and an attempt to keep (4). They solve the dilemma with moderate lifetime prioritization (MP), arguing that it grounds a "mixed view" that allows a balancing act to be completed between (1) and (4). Sometimes (1) can be accepted and sometimes (4) can be accepted (implying that sometimes either (1) or (4) can be rejected). They then respond to three objections to MP.

I want to raise a few concerns, some more general and some directed at specific pieces of the paper. Since the post only allows 4100 or so characters, I'll post them separately.

A. JMW's response to the "mere difference" rejection of (2), p. 8-9:

Their third objection to this move is to say that a mere difference view would not undermine ALL differences in QAL (e.g. life in prison vs freedom). Hence, some form of principle (2) would still apply. However. Why doesn’t this miss the point? If the disability version of (2) is what got us to the trap about QALY being discriminatory against the disabled (hence, ‘morbidity’), then why would a modified version about, say, the value of reducing disability-independent bads associated with prison life get us to the discrimination of disability? If a mere difference proponent rejected the disability version of (2), then don’t they avoid the QALY trap as it relates to disability? They could also still get to keep a plausible version of (2), which would say that it is better to save a life and avoid disability-independent harms (including disruption associated with the appearance of the mere difference) than to save the life and not avoid those harms. That said, they may be right that there would still be the other problems with adopting a mere difference view.



Phil Reed May 15, 2020 at 4:42 PM

Hi Bob,

Regarding your objection A, doesn't the QALY trap generalize to well-being or value in the same way the authors think that (2) generalizes? The mere-difference view could avoid the QALY trap as it relates to disability (supposing the other objections to the view could be met) but they would still face a related QALY trap:

(1) Equal Value to Saving Lives: it is equally valuable to save the life of someone with low quality of life as someone with high quality

(2) Value to Reducing Time Stuck in Prison: all else equal, it is better to save someone’s life and remove her from prison than save her life but leave her in prison

(4) Unequal Value for Equal Outcomes: it is better to save the life of someone in prison who will then be out of prison, than to save the life of someone already out of prison who will then be out of prison. (or something like that)

I take it that the QALY trap is not merely about disability. It's about tradeoffs between saving lives and improving quality of life. From p. 5: "We can generalize the problem further by not limiting ourselves to scarce health care resources, to disabilities as the source of reduced quality of life, or to health-related quality of life. In a world of scarce resources, policymakers must make choices about saving the lives of people with great disparities in expected quality of life, whatever their source. And they must make tradeoffs between saving lives and improving quality of life, whether through health care or the provision of other resources."

Bob Kelly May 10, 2020 at 3:41 PM


B. JMW's response to the "life-saving is different" justification for rejecting (2), p. 12:

In response to rejecting (2) on the grounds that life-saving is different than mere morbidity reduction, JMW say that this leads to irrational outcomes, like being "money pumped" (I think this is the right analogy, but might be misusing this term) into paying more for a treatment than another that has equal outcomes for the same patient. However, I think they raise the right objection to themselves here. The assumption JMW seem to make here is that we ought to look at individual options, assign a value to that option, and then compare options (some of which may be combined after we’ve assigned values). This seems elicit since the target account precisely says that decisions involving life-saving are categorically different than those about morbidity-reduction, and this fact matters to how we value morbidity-reduction. In other words, the contrast class makes a difference to how we value individual options. So, for their case, when we combine B and C and compare them to A, it is false that C gets to “bring over” its “extra” value it had in assessing it on its own. There is no extra value, because we have to re-evaluate the options in light of the new contrast classes. JMW’s objection rests on an assumption that begs the question against the proposed principle they are attacking. Sure, maybe this means we need better principles of individuation still. But that is a different objection, and they seem to maybe run them together in this section.

C. JMW's claim that MP is consistent with (3), the transitivity of betterness, p. 19:

One benefit of MP, they say, is that it is consistent with (3). However, it seems like their MP principle will suffer from the same sorites problem that they raised earlier in the paper regarding the vast number of small headaches outweighing the saving of a life. They claim that if a patient has not had their condition for very long, then the small and recent reduction in QAL doesn’t count towards prioritization. If we extrapolate this principle, then if we have to choose between saving Al, who has a health condition that reduces his QAL by some amount and Bob, with a similar health condition that they have had for slightly more time (and so with slightly more QAL reduction), then Al and Bob should be treated equally (if all else is equal). And the same goes for a decision between Bob and Carl, if Carl has had his condition for just slightly more time. And then the same for Carl and Dave, and so on. But then by the transitivity of betterness, we should treat Al and Zach (who appears way down the line of sorites cases) equally. But their view entails that Zach should be prioritized because he’s had his condition and reduced well-being for much longer than Al.


Stephen Kershnar May 13, 2020 at 5:28 PM


I think we agree on this, but the money pump objection is a way of pointing out the problem with an intransitive solution. If a buyer has the following preferences: A > B, B > C, and C > A, then a seller could continue to sell the buyer B, C, and A options.

The point about the reference class relates to the notion that an option has its prudential value to someone regardless of the reference class in which it occurs. I suspect we agree with this principle. Unless the reference classes changes the value of the option itself to the person, in which case it is a different option, it is hard to see how this could not be true.

As a result if any one set of options is well-formed (rankings are complete as well as reflexive, symmetrical, and transitive), then the reference class is irrelevant.

I gather your point is that the change in reference class changes the option. However, I'm not sure why this would be true with regard to the sort of examples they are discussing.


Steve K


Bob Kelly May 10, 2020 at 3:42 PM


D. JMW's objection to Kamm's rejection of (3), p. 13-14:

They first claim this is unmotivated. Why should we allow case I a chance at treatment and not case III (when they are comparatively equal)? And why doesn't this discriminate against disability? Moreover, JMW argue, this violates the independence of irrelevant alternatives principle which says: If A = B out of [A, B], then A = B out of [A, B, C] when C is "irrelevant" to A and B. If you prefer pancakes over eggs, you should not change your preference to eggs when the server comes back and tells you they also have bratwurst (which you have no interest in whatsoever).

In response to the first worry about motivation, one reason is that contrast classes matter. When comparing I and III, they ought to get an equal chance. But we are not only comparing I and III. If we were, then I = III and we should flip a coin. We are comparing I, II, and III. Another option is introduced that is better than III but equal to I, which may shift the overall judgment. If we flipped a 3-sided coin, an outcome (III) that is worse than another alternative (II) is given equal chance at occurring. Maybe it would be better to assign equal probability to I and II and a lower probability to III. But it seems like in order to ensure the best outcome, you would want to try to avoid the potential for your decision to end up resulting in a worse outcome than was possible. Given the weights, having II trump III is the only way to do that. It seems question-begging to say that Kamm's method fails because contrast classes cannot matter (or, alternatively, to assume that the one's she invokes are irrelevant).

In response to the second worry about the independence of irrelevant alternatives, this again seems question-begging. Their case gets at an intuition that it seems weird how it happens that a new third alternative changes the original valuation of the first two alternatives without that option. But it is also intuitive that something changes. Needy is worse off than Unhealthy, and would be preferable if it were just those two. Yet Needy seems equal to Healthy. Hence, something has intuitively messed with our original valuation of Healthy and Unhealthy. All their case can do is point out the problem. It cannot support the adoption of the Independence of Irrelevant Alternatives on its own since it doesn’t give us any independent reason to think that contrast classes don’t matter to the value of an option.


Stephen Kershnar May 13, 2020 at 5:33 PM


I am not sure I see how contrast classes could change the prudential value of an option unless they change what that option is (which is a different question) or unless value is somehow comparative (versus non-comparative). A comparative value is, for example, how attractive a man is to a set of potential daters depends on the other men against whom he competes. A non-comparative value for example is a net 10 util gain.

How does assuming that medical cases involve options that have a fixed reference and are non-comparative beg the question? Perhaps I am missing something.

I hope things are going well. Buffalo is finally getting through the cold spell. Best,

Steve K

Bob Kelly May 14, 2020 at 6:49 PM

Well the principle of the independence of irrelevant alternatives is what JMW claim is violated by Kamm's rejection of the transitivity of betterness. My claim is that JMW's objection to Kamm is simply to assume that this principle cannot be violated--or, perhaps more subtly, that the extra alternative introduced in their case, (III) Needy, is irrelevant to the valuation of (I) Healthy and (II) Unhealthy). Either way, they don't do anything else other than show how denying the transitivity of betterness would make it so that introducing a new alternative can change the comparison between two other (seemingly unrelated) alternatives. But this means that they implicitly reject the possibility that an option's value can change merely in virtue of having a new contrast class. Kamm has to deny this in denying the transitivity of betterness. My point was just that the only objection they have to this move is to say that it looks weird when we have to change the comparative values of two options when another one is introduced, or when we change the value of a single option when it is introduced into a new set of options. For instance, they simply insist that it is wrong to treat Healthy and Unhealthy equally before Needy comes in, and then differently once Needy is in the mix. That is basically just a restatement of denying the transitivity of betterness. You could almost use that case to explain to someone what it means to deny the transitivity of betterness. Hence, they don't object to it really. They explain her view, and then basically just say "that's crazy." Their footnote says "Elsewhere, Kamm (2013: 470) recognizes that her opponents argue that her view violates the Independence of Irrelevant alternatives, but claims that 'such cases help show that the principle is either incorrect or its correct interpretation does not conflict with such a result.’" Then they move on. That's fine that they think it can look somewhat weird in certain cases. But unless you are already committed to one or the other view, it's hard to see how this objection does any work. It's not to say Kamm is right. Perhaps her arguments are similarly dismissive. But this objection on its own should not move Kamm or like-minded people at all.

David H May 14, 2020 at 7:13 PM


I agree with your first point that the prison sentence lowering well-being is irrelevant if the authors’ target is the mere difference account of disability objection to offsetting the Value of Reducing Morbidity principle. Prison sentences aren’t disabilities. You show some sympathy to the authors’ first claim that since even the mere difference theorists admit that there can be transition costs when one acquires a disability then that would bring a QALY trap. The transition to disability would lower the quality of life for a time and so could lead to discrimination of the disabled if we are to choose which lives to save on the basis of comparisons of the resulting welfare. But I think the authors can’t actually help themselves to transitions costs of the newly disabled since they are trying to sustain the Value of Reducing MORBITY principle. Once they accept the mere difference account which means disabilities are polymorphisms and not pathological dysfunctions, then can’t help themselves to transition costs since they are not welfare losses due to morbidity, taking the latter literally as disease. I supposed the most charitable read would be they mean by “morbidity” something along the line of welfare losses due to bodily differences and then transition costs are relevant and return the QALY Trap even for advocates of the mere difference view of disability. But then the trap is also there because of ableism as that will bring about lower well-being due to non-morbid bodily differences. If we should ignore ableist caused welfare losses in distributing life-saving scarce resources, why not ignore the transition costs?

Bob Kelly May 10, 2020 at 3:42 PM


E. JMW don't really escape the dilemma:

They give us no principle for deciding when principle (1) [equal value of saving lives] or (4) [unequal value for equal outcomes] should be accepted or rejected. Specifically, for (1), they say that the QAL for saving the worse-off can be low enough such that we shouldn’t go with that principle, and should instead use QALY to trump prioritizing. So, when we need to reject (1), we introduce prioritization; when we need to keep it, we introduce QALY trumping prioritization. But when? How much QALY trumps? They do not tell us this, but only provide an intuitive case on each side. For (2), they do the same thing, but instead of introducing QALY trumping prioritization, they introduce duration of being the worse-off. Sometimes this matters, and so we prioritize worse-off over healthy; sometimes it isn’t enough to matter, and so we treat them equally. But when? How much duration matters? Is this just again QALY trumping or not at some point?

The dilemma is that the intuitive (1)-(3) entail the counterintuitive (4), and the various ways of getting out of this all have problems. Their solution is to say that we can sometimes reject (1), but not always, and sometimes we can keep (4), but not always. They do not say when or how these trade-offs work. Describing one case that each captures is not enough to tell us that their view is better than the alternatives since we know rejecting (1) is sometimes bad (per their objections to this method) and keeping (4) is sometimes bad (per the motivation behind the dilemma). If all we know is that to get out of the dilemma we should sometimes reject (1) and sometimes we should keep (4), we need to know when and how that works, and if when and how it works ends up giving us better results than the methods they discuss and reject.

F. JMW's backwards-looking (lifetime) metric produces counterintuitive results:

Suppose we need to choose to save either Anne or Beth, who would equally benefit from the intervention. Beth was in and out of hospitals from age 5-15, and so during those years had a worse life than Anne. However, from 15 and on, Anne and Beth had basically the same quality of life. They are both now 35 and need life saving treatment due to a completely random incident. Lifetime prioritarianism would say we have to prioritize Beth, even if just a little bit. This seems completely counterintuitive, as does the possibility of providing reasoning linking the two (ill-being 20 years ago with a completely random and unrelated health need now).


David H May 14, 2020 at 7:15 PM


I don't see the counter-intuitiveness of ignoring Beth’s earlier troubles in and out of hospitals. Maybe, if we grant that they have had the same quality of life ever since. But how plausible is that? Isn’t it likely that Beth is behind educationally, socially, athletically in that she couldn’t do what Anne did in that decade and a half. It is hard to believe it doesn’t have ramifications in mental and social abilities, not to mention sadness and regrets that she didn’t have the normal childhood and adolescent with its memories and fulfillments. Even if she didn’t lag behind physically and emotionally and educationally and socially, she might still feel cheated out of the pleasant childhood others can recall. I actually find more troubling than favoring those who were unhealthy in the distant past, discounting the newly disabled who haven’t lost out in the past but will have lower QALYs in the future. That really seems really offensive and unfair that they lose out on more life before the life is judged not as good as alternatives. Wasserman et al seem to discriminating terribly against the disabled, devaluing them on the grounds that they won’t benefit from the future as much as others. They might view the first prioritization as pity and the second disqualification as degrading

Phil Reed May 15, 2020 at 6:20 PM


I had the same objection as the one you mention in E, though I think you put it especially clearly. This is really an important point that goes directly to the heart of their paper.

My guess is their response would be: we can't give you a clear criterion about when to favor (1) and when to favor and when to reject it, but that doesn't threaten the account. The cases on the extremes are clear enough (hence their illustrative examples) to adopt prioritarianism. Their solution is an improvement over alternatives to the QALY trap because they save the intuition that both (1) should be defended but that (4) sometimes has to be defended too (and sometimes not).

Maybe that isn't satisfying. One way it might not be at all satisfying is with respect to implementation challenges discussed in 5.3. They write, "anyone who thinks that an allocation scheme should care about more than just maximizing total benefit must have some way of accounting for the other considerations that matter. This is a challenge that we acknowledge, but cannot attempt to address in this paper" (22). It's an especially difficult challenge if they aren't going to give us a criterion for distinguishing when to accept or reject (1).

Jim Delaney May 11, 2020 at 11:54 AM

JWW propose a moderate prioritarianism (MP) as a solution to the QALY trap. A couple of thoughts although they echo some of what has been brought up in earlier posts and comments.

First, like Bob, I worry about the trade off between when we should favor Equal Value to Saving Lives and when we should favor Unequal Value for Equal Outcomes. The appeal to prioritarianism in the first place was supposed to find a principled way to escape the QALY trap, but without a principled way to do the trade off, the account risks being ad hoc.

Second, I refer back to their case of Cecile and Roberto. On MP, we should treat Roberto even though doing so will result in less total benefit, because Roberto is worse off that Cecile. I don’t have this intuition. I’d have to think more about it, but I am almost inclined to say it should be a coin flip. An additional concern is about whether or not desert and responsibility should play a role in decisions if we accept MP. Suppose Roberto’s well-being is lower than Cecile’s because he engaged in immoral or reckless behavior. Does he still get priority just in virtue of the fact that he is worse off? If so, this seems like it runs afoul of the principle that one should not profit from their own wrong. If he gets lower priority because he put himself in a worse off position, we need to get into the business of how much responsibility people bear for making their own lives go worse. This is going to be controversial and difficult.


Bob Kelly May 13, 2020 at 10:55 AM

Jim, I think you raise a good point that they didn't consider. If the person who is worse-off is at fault for being worse-off, should they still get priority? And how do priority and responsibility interact--what is the function? I think if they say it DOES matter, they are in a worse position. Hence, I think they should just say that it DOES NOT matter. Here are some reasons why it shouldn't matter. Some of this is taken from Dien Ho's paper on responsibility in organ donation. Some are mine.

1. No one is responsible

2. The conversation would inevitably be confused by talk of disease, which is irrelevant to control and responsibility

3. It is impossible to have a standard that would consistently apply the appropriate control/responsibility adjustment since determining this would require information we wouldn't have

4. Why would responsibility matter her (for life-saving from illness) and not other times (relief agencies, education)?

5. May jeopardize doctor-patient relationship/trust if they have to inquire about past drinking, etc. and then use it against them

6. We standardly don't punish people for risky behavior that leads to a health care need by withholding health care (e.g. skydiving, speeding, bungee jumping, etc.).

7. There may be paradoxical results - non-smokers who responsibly choose not to smoke might deserve less priority since they are responsible for putting themselves in a position to need more health care for longer (they live longer).

I think (1) trumps, but the other reasons are good enough for me to say that JMW should just reject responsibility as a legitimate factor in determining priority.

Stephen Kershnar May 13, 2020 at 5:41 PM

Jim and Bob:

I like what both of you have to say on this matter. What I think this shows is that equality is unclear in its conception (even if it is a clear concept). Hence, deciding what equality requires depends on whether you think it requires the equalization of one or more of the following.

(1) Well-being (total, average, or average per time)

(2) Health

(3) (1) or (2) with a deficit for which a person is blameworthy being discounted or ignored.

(4) Equal outcomes given what people deserve.

(5) Equal rights against force, fraud, or theft.

Because these are more specific issues and equality piggybacks on them, we should ignore equality and focus on these issues.

In short, death to equality-based theories of equality. This includes prioritarianism.


Steve K

David H May 14, 2020 at 7:21 PM


While I share the intuition that it is problematic to benefit from a wrong, I am not sure that those who are to blame for their medical needs are benefiting from a wrong. I think of benefiting from a wrong to be like keeping the money one earned from investing stolen funds. But getting an organ because one was imprudent doesn’t seem like benefiting from a wrong as the imprudence lowered one’s well-being and the organ, at best, just restores one to an earlier pre-imprudent level of well-being. I suppose acquiring the organ is a benefit relative to the organ going to someone else. But it is also a benefit to marry one’s defense attorney and that is not grounds for annulment.

Admittedly, I find it disturbing if virtuous you need a liver and it goes to someone in prison or someone whose liver failed due to their drinking, or someone who in unwilling to be a liver donor at their death. But on second thought, I don’t think it right that prisoners get lower priority because that unfairly increases their sentence of X years to include risk of life or limb. The person who takes organs but won’t give is not a free rider unfairly benefitting if donation is a gift and comes with no duty to reciprocate. The case of the hard drinker who damaged his liver is plagued by worries about alcoholism beings a responsibility undermining disease or earlier stronger initial dispositions, even if one is not a global skeptic of responsibility as I think are Rob and Steve.

I am also worried about “punishing” those who voluntarily engage in reckless or risky jobs like race car drivers and football players as it seems wrong to lower their chances for a transplant or other medical resources when their needs arise from their risky occupations, even if their salaries compensate them for the risk. It doesn’t look like it will be easy to distinguish good risk taking from bad risk taking. And if we could, why not send those responsible for their conditions to worse doctors, poorer performing hospitals, - assuming scarcity in such goods?

Even if I had more confidence about which moral evaluations should determine one’s place in the health care queue, I worry that it is illiberal for doctors to impose them. Would sexual activity be condemned as illicit risk taking even if not frowned upon as promiscuity? Why not judge too risky those non-contracepting “geriatric” pregnancies that lead to complications and possibly require a demand for scarce resources, perhaps a ventilator during a flu epidemic?

Nor are doctors trained to make overall well-being judgments or allot responsibility, and they can’t do so with an unknown, dying John Doe in the ER. If the job is passed off to committees – which might avoid Rob’s concern about corrupting the doctor/patient relationship, though it might corrupt the hospital/patient relationship - I fear that those whose values I don’t share will hijack the committees and determine who is deserving of their place in the queue. So, I would rather just distribute non-judgmental health care except in rare cases where one wrongfully harms another who as a result needs a medical resource that the wrongdoer also needs.

Phil Reed May 15, 2020 at 6:51 PM

I have an objection that hasn't been raised in the previous comments. I'm worried that the conditions of scarcity that motivate the QALY trap are unrealistic and create problems where none exist in the real world.

Take, first, the example of Yin and Sam, which is supposed to be a defining example. This example (p. 3) seems bizarrely unlikely. Under what scenario would we be able to treat someone for paraplegia but not for ankle instability? Perhaps if there is only one surgeon? But would there be a single surgeon who could do both surgery for severe lumbar spinal torsion and ankle surgery?

Similar things can be said about a lot of the examples. On p. 7 we are asked to consider a case (from Beckstead and Ord) about Alice and Beth who have contracted "an unusual disease" where they will be left blind or with a reduced lifespan. The case is unusual indeed! The product of a philosopher's creative imagination! It is simply stipulated that resources are scarce. Why? We have actual conditions of scarce resources organs, pandemics, etc. But how is this one? It's fun to think about, but not necessarily worth thinking about. On p. 11, the problem of individuating life-saving interventions is less severe during a genuine problem of scarcity. Are we supposed to believe that meals, ventilation, and 8 (!) different medications are all simultaneously scarce?

To generalize: My concern about posing the QALY trap in the way that they do is that it is distractingly unrealistic. The cases where we can tradeoff someone's quality of life for saving lives are not (except perhaps for philosophers' thought experiments and very rare actual cases) cases of distributing scarce resources. QALY Comparisons are beside the point. Where we actually do need to distribute scarce resources, disability discrimination is a genuine problem and our goal should be to eliminate it. In other words, how often does scarcity require us to make tradeoffs between saving lives and improving quality of life? Not that often. In the cases that it does, Equal Value to Saving Lives should be used to avoid disability discrimination. I would likely defend Harris's approach to "avoid" the "QALY trap".

Harvey Berman May 17, 2020 at 3:07 AM

Apologies for not reviewing what has gone before, but I hope this offers points not contributed. 

In the late 1940s and early 1950s, at the birth of television, there was a daytime program Queen for a Day, in which women competed for the one whose life was most filled with despair, sorrow, bathos, pathos, misfortune, and overwhelming loss. Not that those were bad things, because the story and pitch were milked for emotional effect.

The women—as contestants in a competition—would describe losing a husband, thereby losing income and straining to support children. If they lived on a farm, their cows died, the collie dog their son Timmy lovingly named Lassie had to be put down, the utilities were shut off for nonpayment and they were living not only without electricity but without … television, their three-year old daughter was diagnosed with a lazy eye, and it went on and on, one tragedy piling on another. There was no end to the suffering the women described. Consumption, illnesses of no known etiology and undiagnosable, there were innumerable losses of brothers and sisters, children, mothers and fathers, grandparents dropping off like flies and, not to be marginalized, yet the contestant felt gratitude for small kindnesses of strangers, and that life was important as God had a plan for each of us.

The TV audience at home loved it, as did the audience live in theater. This was must-see daytime TV. Using an applause-meter, the live audience voted on the woman whose life was the worst, likely projecting their own idea of who suffered the most, who evinced the most self-pity and received the most sympathy. The one contestant whose past life was one of abject misery and whose future was most uncertain, won the competition, receiving a number of gifts, and was crowned Queen for a Day.

This is the model of moderate prioritization proposed by Wasserman to remedy “invidious disability discrimination,” that impedes access to health care and allocation of resources for those with disabilities. And, like the studio audience projecting their own biases about suffering, Wasserman offers a similarly undefinable criteria for determining who gets treated when treatments are scarce, or very expensive, or there are competing individuals who also merit such treatments, and what scale survive to survive a stringent definition. Yet, in the end, Wasserman offers to replace allocation schemes such as cost-effective analysis and QALYs, with another plan that is no less arbitrary and discriminatory (but against the more abled).

Harvey Berman May 17, 2020 at 3:08 AM

Wasserman refers to his solution as a “moderate prioritization” that should solve the problem of the QALY trap, a straw man argument that fills most of the paper. I found this labored and boring, so I looked past it and came to realize that it was not necessary to trek through much of it. But I do note that this paper written before the COVID crisis yet presents an opportunity to discuss allocation of scarce resources during a pandemic.

Three problems jumped out at me:

1. The requirements for moderate prioritarianism to enhance access of the disabled to scarce medical care can easily work against the disabled and in favor of the more abled. Wasserman views the disabled as disadvantaged, little or no different from those who are “poorer … [or] a widower … [or one] with chronic disease of disability.” Since he believes at the outset that the disabled have a lower life expectancy, he recommends using the blanket cover of people he labels as the prioritarians, that the disabled should be given “some [emphasis in original] priority,” by which the word some can mean a lot of priority, a moderate amount of priority, or a little priority. And, who is to make these critical decisions regarding the meaning of what amount of priority is meant by some priority?

Harvey Berman May 17, 2020 at 3:08 AM

2. In turn, Wasserman places far too much power in the hands of hospital administrators to make fine decisions based on nebulous or at least unenumerated criteria. In attempting to eliminate health disparities, Wasserman believes that “all welfare considerations should … be taken into account.” How does one consider all welfare considerations? What exactly is a welfare consideration? Poverty? Lesser education? Lower social status? Homelessness? Receiving social assistance? At one point the authors bring up, but don’t strongly discount, the idea of “lifetime prioritarianism,” a policy that requires that the kind or quality of life the patient lives be counted in the patients favor along with the amount and duration of suffering prior to making his appearance at the hospital. Such a requirement clearly discriminates against those who may have lived joyful, productive and satisfying lives, with or without the disability.

3. Finally, after affirming that no disabled person be denied access to the same care as the abled, implying there must be strict guidelines although they do not provide the, the authors highlight one interesting feature of their moderate prioritization, that it is … flexible and does not give “absolute [emphasis in original] priority to the worst-off.” That is, if treating one disabled individual will result in fewer benefits than treating a different individual with a lesser disability, then by all means place the worse off patient aside and treat the better off patient. This mandate, further empowering the business end of the health care industry, is that physicians, nurses, ethics consults, and hospital administrators, would have to rate the severity of one disability versus another disability.

This is not a trivial matter, as Wasserman’s moderate prioritization asks policy makers to “care” about a “patient’s degree of disadvantage” and the “amount of benefit” they can provide him. The irony is this binary prescription — disability versus benefit — is little different than the definition of QALY-- which incorporates both the life expectancy and the perceived impact of illness and disability on the quality of life—the Wasserman’s straw man.

As one example that troubled me, consider two individual who had been afflicted with polio as children. In the first child, the effects of polio remained throughout his life and, in turn, would offer him points favoring his eligibility for treatment in an emergency situation, as polio represents a potent mark of disability.

On the other hand, if the second individual had overcome polio, and went on to live a full life without any evident disability, his success would count against him in being afforded treatment.

In the end, I view the Wasserman paper as less of a policy prescription and more of plea asking medicine to be more cognizant of intrinsic discrimination of the disabled, discrimination that appears to reflect a perceived lack of worth in the lives of those living with disabilities. And while moderate prioritization is well-intended, it suffers in the same imperfection as the expression "that the operation was a moderate success."