2. Value dimensions and comparability

How might we think about and evaluate the permissibility of ordeals, the use of which generates some benefit but also imposes some cost on candidate recipients of the good? For instance, how might we think about and evaluate the permissibility of an ordeal the introduction of which improves the overall health outcome but imposes a cost in terms of time on recipients of the good? Consider an example: Is it ethically permissible to implement a policy according to which only those who sit in a waiting room for two hours get a chance to see a back-pain specialist? There are reasons to believe that ordeals of this kind would weed out patients with lesser health problems so that the specialist’s time would be better targeted toward patients with greater health needs since, presumably, those with lesser health problems are less willing to wait to see a specialist than those with greater health problems. However, the policy imposes a burden on candidate recipients, and it might also lead to unfair distributions of goods since ability to endure the ordeal (i.e. spend two hours in a waiting room) can be expected to be unevenly distributed across the population of candidate recipients in a way that does not correspond to need, capacity to benefit and severity of illness (see Rose Reference Rose2016, MS).

Some conceptual clarifications facilitate the evaluation of ordeals. At a general level, one might say that ordeals actualize two kinds of value dimensions that are relevant to all-things-considered comparisons. First, there is a benefit dimension, the value dimension in which the value of the benefits is expressed. The substantive nature of this value dimension is reflected by the purpose of an intervention, and it is typically the dimension that corresponds to the currency used to measure efficiency. For instance, in the context of healthcare allocation, where the purpose of an intervention is to improve the health of a population, the benefit dimension is often health impact. An intervention with greater health impact is considered more efficient. A more effective targeting of specialist time will render a greater health impact and is more efficient in that it mitigates more ill-health. Second, there are cost dimensions, value dimensions in which the (dis-)value of the costs is expressed. The substantive nature of what I will call the immediate cost dimension is reflected by the design of the ordeal, the deadweight cost that candidate recipients must take on in order to attain the good that is distributed.Footnote ¹ For instance, in the context where an ordeal consists of demanding of candidate recipients of goods to wait, the immediate cost dimension is the waiting time that candidate recipients must endure to receive the good.

As should be evident, one can furthermore draw some distinctions between different ways of thinking about cost dimensions and different kinds of cost dimensions. First, one can – as described above – focus on what one might call the immediate cost dimension, the value dimension that corresponds to the direct currency of the costs that ordeals make use of, e.g. time taken from people, the effort required to fill out forms.

Second, one can focus on the direct impact of introducing these immediate costs on candidate recipients. What are the consequences of asking candidate recipients to spend time as ‘deadweight costs’? The immediate cost dimension and the cost dimension that focuses on direct impact can be expected to stand in a quite complex relation to each other. For instance, filling in forms does not require the same amount of effort from all people and demanding that candidate recipients do it will impact different candidate recipients in different ways. Some will be used to administrative burdens and find no difficulties understanding the questions, some will find it difficult to understand how to fill in the forms, some might be practically unable to fill in forms (due to e.g. illiteracy), and so on. Similarly with the immediate cost dimension ‘time’: the direct impact on candidate recipients of asking them to give up some time differs depending on what they would have alternatively done with the time, and the opportunity cost for a person of spending time in a waiting room differs across heterogeneous populations depending on the circumstances of each individual. For some, the alternative cost of spending time in a waiting room will amount to less time to play computer games, while for others it will amount to failing to help their children study, and yet others might have to give up much needed income since they cannot work, etc.

Third, one can focus on what one might call the distributional impact of introducing ordeals. Since different ordeals are differently burdensome – differ in direct impact – for different individuals, it can be expected that ordeals might affect the distribution of goods. This can happen for different reasons. Some might not be at all able to go through the ordeal. For instance, an illiterate person might be practically unable to appropriately fill in a form, which might mean that she completely loses access to the good if it is distributed through such an ordeal. Yet, inability to go through the ordeal is not the only way in which ordeals might influence distributions. Distributions can also be affected due to the fact that for some the direct impact might be so burdensome that they prefer not to take on the ordeal. For example, this could be the case if someone must choose between waiting to see a specialist to treat their back pain or look after their children, while someone else faces the choice of waiting to see a specialist or going home to take a pleasurable but not badly needed nap.Footnote ²

These distinctions are important because they reveal the complexity of evaluating the cost dimension(s), as well as the significance of taking into account not only the immediate currency of the ordeal but also the direct impact on candidate recipients of goods and the distributional effects. From an ethical perspective, it is obvious that we should be interested not only in the immediate cost dimension, but also in direct impact and distributional impact.

One can now say that when, all things considered, the value in the benefit dimension is at least as great as the disvalue in the relevant cost-dimension(s), it is at least as good to introduce the ordeal as to refrain from doing so. When it is at least as good to introduce an ordeal as to refrain from doing so, it is – on the comparativist account – permissible to introduce the ordeal.

The benefit as well as the cost dimensions of course differ depending on what kind of context and ordeal one is interested in evaluating. However, characteristic for all ordeals is that the cost dimension is non-monetary. The immediate cost dimension is typically some form of minor inconvenience or time cost (e.g. waiting, filling out a form, returning a coupon), while the direct impact might relate to more significant inconveniences or harms and the distributional impact often actualizes fairness consideration. All-things-considered costs might, in light of the distinction between different ways of thinking about the cost dimension, be seen as multi-dimensional. Time-consuming ordeals impose a cost in terms of time on candidate recipients of a good which has different impact on different individuals, but they might also make the good inaccessible to certain groups, and in so far as free time is something individuals have a valid claim on, they can also constitute an injustice by frustrating some individuals’ valid claims (see Rose Reference Rose2016). The benefit dimension, meanwhile, might be monetary but it can also be non-monetary. Typically, the benefit dimension is one-dimensional when ordeals are considered since the reason invoked in favour of ordeals refer to efficiency which is usually treated in a one-dimensional manner. How do the cost and the benefit dimensions relate to each other?

To evaluate how the values in the benefit and cost dimensions relate to each other it is helpful to look at the value-theoretical literature that has considered comparisons across different kinds of value dimensions. How might gains in a benefit dimension and losses in a cost dimension relate to each other? Based on the contemporary literature on cross-dimensional value comparisons, here are four initial pertinent conceptual possibilities:

1. 1. Unlimited Cost-Benefit Trade-Offs: For all amounts of losses in the cost dimension, there is an amount of gains in the benefit dimension that is large enough so that it is at least as good to introduce the ordeal than not to (assuming that the ordeal would indeed create the required amount of the benefit).

If it is the case that the cost and benefit dimension stand in a relation that implies that Unlimited Cost-Benefit Trade-Offs holds, evaluating the permissibility of ordeals would be a matter of determining whether the benefits are great enough to outweigh and justify the costs regardless of how high the costs are. Sometimes, this is referred to as ‘The Archimedean Property of Value’: ‘For any amount of A there is some amount of B which is better’ (Arrhenius Reference Arrhenius2005: 97). It implies that no amount of costs is so significant so that it cannot be outweighed by a great enough benefit. That Unlimited Cost-Benefit Trade-Offs are possible and that the Archimedean Property of Value holds seem to be assumed in the quotation from Nichols and Zeckhauser above. They express both benefits and costs in terms of ‘utiles’ and suggest that ordeals are permissible when the benefits in terms of utiles outweigh the costs in terms of utiles. In their argument benefits and costs are expressed with a single currency, but that is not required for Unlimited Cost-Benefit Trade-Offs. For instance, it is conceivable that health impact and time costs relate to each other so that Unlimited Cost-Benefit Trade-Offs are possible. If that were so, it could for instance be the case that for each bundle of 1,000 hours lost, at least one Quality-Adjusted Life Year (QALY) must be gained for it to be at least as good to accept the introduction of ordeal as not to.

1. 2. Weak Superiority in Value: For some amount of losses in the cost dimension, there is no amount of gains in the benefit dimension that renders introducing an ordeal at least as good as not doing so.Footnote ³

If it is the case that the cost dimension is weakly superior in value to the benefit dimension, evaluating the permissibility of ordeals is partially about determining whether the cost is significant enough so that no amount of benefits can outweigh it. It is only partially about this because cost-benefit trade-offs might be possible when the costs are not this significant. This possibility reflects a kind of discontinuity in the cost dimension(s) (see Griffin Reference Griffin1986). At some amount of costs, the relation between costs and benefits radically changes so that trade-offs are no longer possible. This might be the case according to certain normative frameworks and with certain kinds of ordeals that actualize a cost dimension where some amount of burden imposed on a candidate recipient constitutes a rights violation. For instance, if – as Julie Rose has argued – all humans have a right to a certain amount of leisure time, it would constitute a rights violation if candidate recipients of some good had to give up so much time in deadweight costs that it would infringe on the leisure time they have a valid claim on (Rose Reference Rose2016). At this specific point in the immediate cost dimension ‘time’, where the ordeal not only imposes a cost in terms of time but also a rights violation, a discontinuity occurs, and someone who takes this right very seriously might argue that there can be no amount of benefits that justifies imposing this cost on candidate recipients of goods.

1. 3. Strong Superiority in Value: Irrespective of the amount of losses in the cost dimension, there is no amount of gains in the benefit dimension that renders introducing an ordeal at least as good as not doing so.Footnote ⁴

If the cost dimension is strongly superior in value to the benefit dimension, evaluating the permissibility of the ordeals that actualize costs in this value dimension simply consists in concluding that it is always impermissible. This possibility reflects a strict ordering of the value dimensions in terms of their importance and relevance for evaluations of alternative courses of action. Sometimes, this is called trumping (see Griffin Reference Griffin1986). Regardless of the amount of costs, the costs always outweigh the benefits when there is Strong Superiority in Value. Whether this is plausible when one evaluates the permissibility of ordeals might seem questionable. However, that this relation obtains between two value dimensions might explain why certain kinds of ordeals are considered unthinkable. For instance, the idea that it is permissible to have patients clean the bathrooms before getting access to a back-pain specialist will to most people seem outlandish (even if this turns out to be a very effective targeting tool). A plausible explanation of why many of us have this reaction is that the cost dimension that reflects the cost of having people with back problems clean bathrooms is strongly superior in value to the benefit dimension.

The three conceptual possibilities introduced above are located on what might be called a spectrum of continuity in possible trade-offs: From the possibility that trade-offs are always possible between two value dimensions to the existence of some discontinuities that rule out some trade-offs to the strict ordering of value dimensions that rule out all trade-offs. There is, however, a fourth conceptual possibility that seems relevant in light of the significant differences between the value dimensions that must be put together when the permissibility of ordeals is evaluated:

1. 4. Nondeterminacy: No conventional comparative relation can be determined to obtain between an alternative that includes an ordeal and an alternative that does not (see Chang Reference Chang2002; Rabinowicz Reference Rabinowicz2008; Herlitz Reference Herlitz2019).

Nondeterminacy – which is sometimes ambiguously called ‘incommensurability’ – has received significant attention lately under different names and within different explanatory frameworks. John Broome and others have argued that it is occasionally impossible to determine that a specific conventional comparative relation (i.e. better than, worse than, equally as good as) holds between two alternatives, and the reason why – he suggests – is that standards/metrics used to evaluate alternatives sometimes are vague (Broome Reference Broome and Chang1997; Elson Reference Elson2017). Ruth Chang and others have argued that some non-conventional comparative relation such as ‘parity’ sometimes hold between two alternatives (Chang Reference Chang2002; Carlson Reference Carlson2010). Wlodek Rabinowicz has shown that within the so-called ‘fitting-attitudes’ framework to value, according to which value relations should be understood in terms of which attitude it is fitting to have toward an alternative with respect to another, it is conceivable that a whole range of different attitudes – and in extension preference orderings – might be permissible with respect to some set of alternatives (Rabinowicz Reference Rabinowicz2008). This is not the appropriate place to address which of these explanatory frameworks is correct. Instead, I will use the term nondeterminacy to stay neutral with respect to the appropriate explanation for why it is impossible to determine that a particular conventional comparative relation holds between two alternatives. I will say that at least at a conceptual level, it is possible that sometimes some alternatives are nondeterminate in their ranking.

When two value dimensions stand in a relation to each other that admits of nondeterminacy, i.e. when they stand in a relation to each other so that it is sometimes the case that two alternatives that instantiate values from the two dimensions cannot be determinately, conventionally ranked, it will not be possible to always determine whether it is at least as good to introduce an ordeal as to refrain from doing so. Nondeterminacy means that the differences between value dimensions undermine completeness. If the relevant value dimensions stand in a relation to each other that admits of nondeterminacy, evaluating the permissibility of an ordeal is partially about determining whether the alternatives introducing the ordeal and refraining from doing so are nondeterminate in their ranking. Furthermore, a view of what nondeterminacy implies in terms of permissibility must be amended to the cost-benefit analysis – an issue to which I will return in the next sections.

3. Comparability problems in practice

Are the conceptual possibilities outlined in the previous section relevant for evaluations of ordeals? I think they are. Evaluating ordeals from a comparativist perspective requires cost-benefit analysis. The costs are compared with the benefits in order to reach a judgement. The conceptual possibilities presented above provide a way of nuancing cost-benefit analyses and to see the complexities surrounding the introduction of non-monetary costs with the purpose of increasing efficiency. In this section, I will outline in what way the different conceptual possibilities might be relevant to ordeals evaluations in further detail by looking at what conditions entail what possibility.

Strong Superiority in Value at first sight seems highly implausible in the context of ordeals. If, for instance, we can improve the health impact by some very significant amount at a very low cost in terms of time taken from candidate recipients, this must plausibly be permissible. However, as already indicated above, I suspect that an idea similar to the one behind Strong Superiority in Value might lie in the background when we consider what kinds of ordeals we should at all be talking about. Some kinds of ordeals are simply out of the question. There is a reason that we do not, for instance, consider introducing an ordeal that involves eating pork in order to weed out members of a Muslim majority population that might make less use of some good, and there is a reason we do not speak of deeply humiliating deadweight costs when we speak of ordeals. One explanation for this is that we simply think that religious integrity is strongly superior in value to health benefits, and more generally that non-humiliation is strongly superior in value to many benefits that humiliation could generate.

Are discontinuities that support Weak Superiority in Value at all plausible? The value-theoretical research on superiorities in value provide reason to believe that this phenomenon is rare, if at all existent (see Arrhenius Reference Arrhenius2005; Arrhenius and Rabinowicz Reference Arrhenius, Rabinowicz, Hirose and Olson2015). Several questions must be addressed to establish discontinuities. First, at what point(s) in the value dimension do discontinuities occur? Second, what makes these locations within specific value dimensions so special that the locations in the immediate vicinity do not share the property that undermines trade-offs? What could make it so exceptionally bad to impose an ordeal that takes say 57 minutes from a person’s leisure time so that it cannot be outweighed by any amount of benefits, and why would this special property that this ordeal supposedly has not be there if the ordeal only took away 56 minutes and 59 seconds from a person’s leisure time? Weak Superiority in Value implies that for two adjacent locations in a value dimension there is such a radical difference. Questions such as these make value theorists sceptical of discontinuities of this kind as well as of the general relevance of Weak Superiority in Value.

However, I believe that there is an argument to be made for the relevance of Weak Superiority in Value when ordeals are evaluated. This argument invokes the importance of distinguishing between different cost dimensions. In the previous section, I introduced three different cost dimensions: immediate costs, direct impact costs and distributional impact costs. I believe that it is plausible that the immediate cost dimension might be weakly superior in value to the benefit dimension in virtue of how certain direct impact costs and certain distributional impact costs arise at certain levels of immediate impact costs.

The different cost dimensions differ in kind, but they are related to each other in complex ways. When certain amounts of immediate costs are accrued – at certain locations in this value dimension – specific direct and distributional impact costs arise. This provides an explanation of how discontinuities arise, and it can explain why certain amounts of immediate costs cannot be outweighed by any amounts of benefits: some new kind of cost has been generated, and no benefit can justify that. Furthermore, it can explain why two adjacent locations in a value dimension might have radically different properties: one location actualizes a cost that is different in kind, whereas the other does not.

This reasoning that invokes the idea of Weak Superiority in Value is – I believe – a charitable understanding of what Reference RoseRose argues when she distinguishes time costs that infringe on one’s fair share of time from time costs that do not. On Reference RoseRose’s view, something happens in the time dimension – there is some kind of discontinuity in this dimension – when the amount of time one asks for from some candidate recipient crosses into the individual’s fair share of time. However, embracing discontinuities on these grounds renders evaluations difficult: How do we determine where they occur? And: How should we deal with the fact that they might occur at different locations for different individuals? As I mentioned in the previous section, the impact of a fixed amount of immediate costs often differs between individuals depending on their circumstances. If one bases discontinuities on impact costs, and if these differ, then the discontinuities will arise at different locations in the value dimension immediate cost for different individuals. This fact significantly complicates the task of evaluating ordeals.

Nondeterminacy is also relevant when we assess the permissibility of ordeals. Although the research on the conditions under which nondeterminacy arises is underdeveloped, some general ideas presented by Elizabeth Anderson regarding conditions for when ‘incommensurability’ obtains support the view that some alternatives are sometimes nondeterminate in their ranking when we consider introducing ordeals. On Anderson’s view, incommensurability (i.e. nondeterminacy in the language of this paper) arises when:

These conditions seem to plausibly hold in many contexts in which ordeals might improve targeting and increase efficiency. For instance, if we take ‘overall goodness’ to be the general standard with which we evaluate whether it is permissible to introduce an ordeal or not, it is obvious that individuals’ possession of leisure time and individual health benefits meet the standard in very different ways. Furthermore, there do not seem to be gross differences in the degree to which leisure time and health benefits exemplify their own way of contributing to overall goodness, and – save for the cases in which strong superiority in value occurs – it is not categorically superior to preserve individuals’ leisure time (or generate health benefits) than to generate health benefits (preserve individuals’ leisure time).

However, what to do when two alternatives are nondeterminate in their ranking is not obvious, and this has been a topic that has attracted some attention lately (e.g. Sen Reference Sen1997; Rabinowicz Reference Rabinowicz2008; Chang Reference Chang2013; Herlitz Reference Herlitz2019). Two views stand out in the literature. First, it might be held that the appropriate response to nondeterminacy is to revise one’s conception of justified choice – what it is permissible to do – so that one holds it to be justified to do anything that is not determinately worse than any alternative (see Sen Reference Sen1997, Reference Sen2017; Rabinowicz Reference Rabinowicz2008; Broome Reference Broome2009; Herlitz Reference Herlitz2019). That is, insofar as neither of two alternatives that are nondeterminate in their ranking can be determined to be worse than any alternative, one is justified in choosing either (both are permissible). Second, it might be held that nondeterminacy comes with what Ruth Chang has called a ‘resolutional remainder’, i.e. something remains unresolved when two alternatives are nondeterminate in their ranking, and in order to resolve this some further reason must be introduced (Chang Reference Chang2002; Wasserman Reference Wasserman2004). That is, if two alternatives are nondeterminate in their ranking one is not justified in merely picking either but must invoke some other reason in order to be able to form a justified choice. What this reason is, and how one justifies its introduction, arguably depends on context (Fröding and Peterson Reference Fröding and Peterson2012; Chang Reference Chang2013; Herlitz Reference Herlitz2019). In the context of ordeals, the first view amounts to holding both the introduction of an ordeal and refraining from introducing an ordeal to be permissible when the two alternatives are nondeterminate in their ranking, while the second view amounts to demanding that some reason that is external to the initial choice situation must be invoked to settle what is permissible and what is impermissible.

4. Identifying how to reason

When cost-benefit analysis is called for, it is often assumed that cost-benefit trade-offs are unlimitedly possible, and in light of the quotation from Nichols and Zeckhauser in the Introduction to this paper we can also note that people often think of situations where unlimited cost-benefit trade-offs are possible when they speak of cost-benefit analysis. There is, however, nothing inherent to cost-benefit analysis that commits proponents of such approaches to these assumptions. Strong Superiority in Value, Weak Superiority in Value and nondeterminacy are all compatible with cost-benefit analysis. Rather than undermining cost-benefit analysis, thinking of the relation between cost and benefit dimensions in terms of the possibilities outlined and discussed in the previous sections allows us to nuance the way we think about cost-benefit analysis and start asking more acute questions: When might we expect to be in which of these situations? When will the cost and benefit dimensions stand in such a relation that Unlimited Cost-Benefit Trade-Offs obtains, when can we expect Weak Superiority in Value, when can we expect Strong Superiority in Value, and when might it be the case that the alternatives of introducing and refraining from introducing ordeals are nondeterminate in their ranking? In this section, I outline some ideas about how to answer these questions.

The discussion of the conceptual possibilities above provides some insight into how to practically evaluate the permissibility of ordeals in a way that is more nuanced than conventional cost-benefit analysis which presumes that trade-offs across the cost and the benefit dimensions are unlimitedly possible. Certain pertinent questions ought to be addressed prior to the evaluation in order to determine what kind of choice situation one is facing. Is it a situation where strong or weak superiority in value characterizes the relation between the value dimensions? Is it a situation where the relation between the value dimensions admits of nondeterminacy?

Unlimited Cost-Benefit Trade-Offs can, I propose, be identified in a negative way. If there is no reason to believe that any relevant value dimension is strongly superior in value to another, if there is no reason to believe that there exist any discontinuities that support weak superiority in value of one value dimension compared with another, and if there is no reason to believe that alternatives that instantiate the relevant values are nondeterminate in their ranking one can hypothesize that the value dimensions admit of Unlimited Cost-Benefit Trade-Offs.

Negative approaches like this might be thought to indicate that the property in question is ubiquitous, and that it is only in very exceptional cases that it is absent. I believe that it is a mistake to hold that view. Although I believe that strong superiority in value is rare, and I also think the case for weak superiority in value often is hard to make, I contend that nondeterminacy will be ubiquitous in contexts in which ordeals might seem attractive. By their very nature, ordeals actualize value dimensions that are significantly different from the motivation behind ordeals. This means that the conditions that Anderson outlined for when nondeterminacy can be expected to occur often will be met in these situations.

Here are three questions that I propose those interested in evaluating the permissibility of introducing some ordeal ought to answer:

1. A. Do the relevant cost dimensions reflect costs that are of such a kind that they are unacceptable regardless of magnitude?

2. B. Are there discontinuities in the relevant cost dimensions such that certain amounts of costs are unacceptable regardless of the benefits one can generate?

3. C. Are the cost and benefit dimensions so dissimilar that one can expect that they admit of nondeterminacy in the sense that it will sometimes be the case that no alternative is determinately at least as good as the other?

How to think about whether an ordeal is permissible or not depends on how one answers these questions. If strong superiority in value obtains, it is obvious that the ordeal is impermissible. If weak superiority in value obtains, one must first establish whether the costs one considers are of such magnitude that no benefit can justify them, and second either deem the ordeal impermissible (if the costs are of this magnitude) or apply more conventional cost-benefit analysis (if the costs are not of this magnitude). If the value dimensions admit of nondeterminacy, one should not expect that it will necessarily be the case that some alternative is optimal (i.e. at least as good as all alternatives), but one can still use cost-benefit thinking in order to determine what alternatives are maximal (i.e. not determinately worse than some alternative) (Sen Reference Sen1997; Herlitz Reference Herlitz2019). If one believes nondeterminacy actualizes resolutional remainders, one must introduce further reasons to determine if a maximal but not optimal ordeal is permissible. If one believes that all maximal alternatives are permissible, one can determine that an ordeal that is maximal is permissible.