It is often stated that the past decades have seen a flourishing of methodological debates within political theory.1 Well-trodden turf includes the ideal and non-ideal distinction,2 the development of political realism,3 and questions regarding political theory’s relationship to the real world.4 The meaning and role of feasibility for theorizing finds its way into all of these discussions either as a way to delimit the proper boundaries of theorizing, or potentially to help adjudicate between competing theories.5 This paper addresses one element of feasibility analysis: its reliance on a distinction between “hard” and “soft” constraints. Hard constraints include such things as logical and nomological impossibilities and are used to provide a clear boundary on the scope of theorizing.6 Soft constraints include a variety of variables, such as motivational and psychological limitations, which render possible world states more or less achievable. Some theorists argue that only hard constraints should restrict the scope of political theory while others argue for greater sensitivity to soft constraints. However, both types of approaches remain in the conceptual shadow cast by this distinction.

This paper argues that when the hard-soft distinction is applied to empirical claims7 it is a misleading heuristic. In particular, it cannot helpfully be used to distinguish between empirical impossibility and very unlikely possibilities—which is exactly what a number of authors seek to use it for. This is because the distinction rests on a misunderstanding of the probabilistic nature of our empirical knowledge. Fundamentally, all of our empirical knowledge is provisional in that it is subject to revision in light of future evidence. Some may see these very slim possibilities of overturning longstanding well-justified empirical facts as of little consequence. But acknowledging this possibility—however small—transforms a strict impossible versus possible distinction into a spectrum of more and less likely features of the world. Redrawing this impossibility-possibility distinction will have different implications depending on the types of theory different theorists wish to produce. I am a pluralist in this sense—there are a great many ways of doing political theory which will serve different functions for the discipline as a whole—but whichever type of theory is practiced, it should not rely on problematically inaccessible distinctions. Following List and Valentini’s criteria for the appropriate usage of concepts in political theory, the application of hard constraints in a way purporting to be distinguishable from very unlikely outcomes is problematic because it is epistemically inaccessible.8 Wherever theorists set the limits of feasibility in developing their theories is a matter of judgement which cannot be avoided by purporting to rely on a categorical hard constraint distinction. A theory sensitive to facts should account for the fact that claims which appear to be nomologically true have a probability of being false or failing to accurately predict the scope of future possibility/impossibility.

This paper will proceed as follows. In Section II I sketch out briefly why feasibility and possibility matter in the context of the dictum “ought implies can.” In Section III I outline how the distinction has been used by a selection of contemporary theorists who have sought to rely on the impossibility-possibility distinction. I also explain how this reliance leads their theorizing astray. In Section IV I suggest various alternative potential implications for political theorizing which could flow from moving past this distinction. I then conclude with a few remarks on the resulting probabilistic feasibility space shared by different theories and the continuing scope for a diversity of theoretical approaches.

There is an extensive debate within political theory regarding the proper relationship between empirical facts and normative claims.9 In one sense it is trivially true that a responsiveness to empirical evidence is a requirement for any normative theorizing which follows the dictum of “ought implies can.” By this I mean that if something is literally impossible it is generally acknowledged that it cannot be normatively required.10 This empirical impossibility constraint is often described, along with logical and metaphysical impossibilities, as a form of hard constraint on theorizing.11 Even as strict a deontology as Immanuel Kant’s did not allow for binding moral duties which were demonstrably impossible.12 This is often taken not to pose a serious limit to theorizing because theorists do not tend to argue for physical impossibilities; in the words of one author: “…political philosophers do not tend to offer theories violating findings of physics, biology or chemistry.”13

Following from this, the question then arises as to how evidence which falls short of hard constraints relates to normative political theory. Some theorists see probabilistic evidence, for example evidence produced by social sciences, as important for political theorizing.14 Others adopt a position of relative fact-insensitivity, especially to findings from social sciences. For example, McTernan has argued that because the practice of social science—such as social psychology—does not produce the law-like regularity of natural sciences, its conclusions cannot count against, let alone rule out, normative theories.15 Similarly, Valentini argues that normative prescriptions should not be ruled out until they have been “proven impossible” in order to ensure that normative claims are suitably ambitious.16 Both McTernan and Valentini, as well as Estlund,17 are concerned for normative theory to be unconstrained by everything short of hard constraints, to avoid status quo bias.18 A key point of agreement between these theorists is that evidence drawn from the social sciences should not therefore constrain normative theorizing about topics such as the nature of justice. As a shorthand I will in this paper refer to these theorists as adopting a “fact-insensitive” position although strictly speaking it is a position which downplays the relevance of evidence short of impossibility for certain types of normative theorizing usually pertaining to higher level principles of justice.19

One way to understand this distinction in approaches is as a pluralistic division of labor.20 Academic disciplines contain multitudes, including diversity in assumptions and methodological approaches. There is not necessarily anything problematic about this, and I do not wish to argue that there is “one method to rule them all.”21 That said, I do believe that the working assumption that hard empirical constraints can, or should, occupy a special place in determining the scope of political theory should be challenged. This hard-soft constraint distinction is common to much methodological discussion of feasibility but is particularly relied on by fact-insensitive theorists to distinguish the firm boundary to their theorizing. Moreover, their attempts to draw a firm categorical boundary to distinguish hard constraints from probabilistic possibilities best illustrate the problems with the hard-soft constraint distinction. These fact-insensitive theorists will therefore be the focus of my analysis in the next section, although the points raised apply to the distinction as used more generally when applying feasibility analysis to theorizing.

To be clear on the scope of my argument: it is an epistemic one concerned with the limits of human knowledge as applied to the development of political theory. It is therefore only indirectly concerned with what Estlund—a prominent fact-insensitive theorist—considers to be the core claim of his book:22

I do not have anything to say regarding what standards are or are not relevant for the truth of any specific political theory. It may well be the case that Estlund and fellow fact-insensitive theorists are correct that hard constraints are the only relevant feasibility limitation on true normative theories. My argument is that we cannot know what these hard constraints are when it comes to empirical claims except in a way which is itself probabilistic. Therefore for the purposes of theory development from our epistemically limited human perspective we can only discuss empirical constraints in a way which is probabilistic—just like soft constraints.

Hard constraints find their way into Estlund’s work in a somewhat oblique manner. Rather than defending a single substantive account of justice, he is primarily concerned with defending his methodological and meta-normative position of “non-concessive” theorizing from what he terms anti-idealism and anti-moralism, as well as a range of other critiques.24 In doing so, he repeatedly acknowledges the force of the “ought implies can” principle, albeit “for the sake of argument”25 rather than in a substantively defended manner. Nevertheless, the importance of hard constraints and the impossibility/possibility distinction figures prominently in his argument as he consistently argues for theorizing to remain as ambitious as possible without crossing into impossibility, demonstrating his commitment to ought implying can.26 His framing is one of ability versus impossibility, because he is concerned to argue against theorists who use soft constraints such as motivational disposition to conclude that individuals are unlikely to behave in accordance with evaluative requirements:27

Simply put, for Estlund if something is possible then it is within the realm of normative theorizing, while if it is impossible then it cannot be a feature or requirement of a true normative theory, even an evaluative rather than a prescriptive one. This means that—short of literal impossibility—the likelihood of some or all agents behaving one way or another is irrelevant from the perspective of normative theory carried out at this level.31 What matters for Estlund is that a theory of justice can specify the normative requirements of a just basic structure and thereby its attendant requirements on individuals and/or groups of individuals. To illustrate a strong separation between questions of possibility and probability Estlund repeatedly uses the example of him dancing like a chicken at a future lecture which he describes as very unlikely but not nearly impossible.32 This leads him to argue that, “It follows that an action’s having even zero probability does not entail that it is beyond the agent’s ability (or even that it is difficult).”33

There are two issues with Estlund’s claims here: the first is more trivial but indicative of the type of error common in this type of analysis when trying to distinguish hard constraints from unlikely events, the second is more substantive and calls into question the entire distinction. The first is that Estlund misunderstands what zero probability means. It is incompatible for there to be a truly zero probability of an event occurring while the ability remains for someone to execute it.34 Assessments of probabilities are always evidence-relative,35 and nobody can predict the future with ironclad certainty—even more so when it comes to human behavior which we acknowledge remains within a given person’s ability. Estlund’s example of himself dancing like a chicken is designed to elicit the intuition that because he controls his own actions, he can ensure it is a zero-probability event. He even jokes that he would be unlikely to accept a bribe and therefore tries to rule out exogenous factors causing said dancing.36 But this lacks imagination; threats to loved ones or other forms of coercion could readily procure said dancing (especially given how widely Estlund has publicized this challenge!). The possibility space of the future is wide, and Estlund does not fully appreciate what zero probability really entails. An extreme example of this is the infinitesimally small but non-zero probabilities which result from quantum physics: “Say you want to walk through a wall; quantum theory says it’s possible.”37 We cannot readily draw a strong clear boundary between “very, very unlikely” and “impossible” for any kind of practical purpose, including theorizing. Estlund’s example of a case apparently under his own will should not confuse the matter.

The second issue is that Estlund misses the wider problem that our knowledge of what is possible or not in the world is itself only ever provisional and therefore probabilistic. This is related to the point made regarding quantum uncertainty above. The issue is that things we have good evidence to believe, and reasons to rely upon, could still be incorrect. Empirical knowledge at the level of our own beliefs is probabilistic, even if sometimes the likelihood of it being incorrect is so low as to approach zero. The reason for this is the methodology of empirical science—including the natural sciences—proceeding by experimental observations which are used to test and develop hypotheses about how the world works.38 It is exactly these types of hard constraint claims that Estlund and others want to rely on to segregate impossible requirements away from other types of evidence which suggest things as more or less likely.39 However, the way we gather knowledge about our reality—through continuous observations and hypothesis development and testing—means that it is always subject to future revision. The mutability and revisability of scientific hypotheses is a bedrock of the scientific method. This point has been made a number of times by various authors and is generally accepted by professional scientists.40 One of its most famous expressions comes from Einstein himself:

Einstein is in fact probably too optimistic in what he claims experimental science can achieve in terms of disproving theories—nature does not in fact say “no.” This is because the theoretical underpinnings of any experiment designed to challenge a theory can themselves be questioned. A more apt description is as follows:

The simplest way to put this point is that scientific knowledge is only ever provisional—it is subject to future revision and refutation. Now, this is not to descend into some kind of relativistic free-for-all whereby all scientific claims are equally valid. We can have more confidence in theoretical paradigms and associated empirical claims that appear to survive replication, withstand empirical testing, and accurately predict outcomes. Nevertheless, it is important to note that this is a relative form of confidence. Even if we are happy to take such theories as a given and rely on the factual claims they produce for practical purposes, we nevertheless have to acknowledge that there is a possibility that they are incorrect. This uncertainty stresses the type of categorical divisions which fact-insensitive theorists such as Estlund seek to draw, between very very unlikely and impossible. While it may be very very unlikely that our more robust scientific theories are incorrect, it is certainly not impossible. This key point was in fact identified by Gheaus over a decade ago but has not been sufficiently recognized in subsequent research when considering feasibility and impossibility:

Issues arise in other fact-insensitive arguments which similarly try to rely on impossibility as a category distinct from improbability. For example, Valentini argues that “If prescriptions have not been proven impossible, and they strike us as highly morally desirable, then we have good reasons to continue to defend them.”44 Valentini wants our range of theorizing to be as expansive as possible to avoid potential status quo bias. However, the language of “proving” an impossibility simply does not accord with how we develop our empirical scientific understanding. This is further illustrated by the example Valentini herself uses of potential future possibilities. She states that it is impossible for a state to eliminate crime and that therefore this cannot be a normative requirement: “This prescription is not valid, since, given the limits in power and capabilities any imaginable state has, guaranteeing an altogether crime-free society is arguably impossible—even for the best of states.”45 There are two issues here. The first concerns what it would mean to prove some future eventuality as impossible. As discussed above with reference to Einstein and particularly Lakatos, this type of language and thinking does not fit with the actual conduct of science where possibilities are not in the final instance determinable by current knowledge. Proving impossibility may be possible as a matter of mathematics or logic but it is not a relevant category for assessing empirical claims.

Second, Valentini’s example illustrates the confusion between very very unlikely and impossible. Valentini denies that any imaginable state has the power to eliminate crime and so appeals to our intuitions about what power any future state might have rather than any empirical study which would show this conclusively. As stated above, it is not clear how one would even go about proving an empirical impossibility in this way. However, in terms of thinking about possibility, imagine existing authoritarian population control technologies were magnified such that the daily lives of all individuals, both offline and online, were subject to state surveillance.46 Combine this with criminal law reform which perhaps reduced the scope of criminal law and again magnified burgeoning technologies such as invasive biotech microchips which can be used to both monitor and control human behavior.47 With sufficient authoritarian surveillance and human bio-alteration I see no reason in principle why a state could not in fact eliminate crime48 or disobedience to the state. Of course, Valentini might suggest that my imagination has run away from me at this point; but mere appeals to intuition are insufficient to rule future possibilities in or out absolutely. I can acknowledge that my vision of a dystopia is unlikely to come to pass, but this intuition is not sufficient to rule possibilities in or out categorically. Consider whether anyone two hundred years ago would believe it possible that billions of people around the world would have handheld devices which could instantly access more information than all of the world’s physical libraries combined and communicate in real time with people on the other side of the world. Any reasonably informed person would have likely thought it clear that this was impossible, and yet they would have been proven wrong by the advent of smartphones and the internet.49 The point is simply that Valentini’s analysis conflates very very unlikely with impossible, reinforcing the implausibility of assigning this distinction a fundamental role in our normative analysis. Despite the fact that both Estlund and Valentini seek to rely on a strong separation between impossible and possible-but-very-unlikely, their own flawed applications of the distinction highlight just how problematic it is to make in practice.

The upshot of all this is that because scientific theories can only be provisional, our belief or confidence in them must be probabilistic in nature; there is a possibility that they are correct and a possibility that they are not. To treat them otherwise would be to place more certainty on scientific theories than scientists themselves do.50 This is distinct in magnitude but not in kind from social science claims that under certain conditions a human phenomenon is likely, but not guaranteed, to occur. McTernan tries to distinguish the two on the grounds that, in contrast to the natural sciences, social scientific investigation does not produce exceptionless regularities and tend to concern particular contexts.51 This may provide additional reason to challenge the findings of social sciences but does not explain why normative theorists could not challenge natural scientific claims too on the grounds that they are uncertain and revisable, subject to future enquiry. It may be that we simply have not discovered the relevant contexts or exceptions which cause the purported natural scientific laws to vary. The question then for normative theorizing is not whether everything beyond a hard constraint of impossibility should be fair game for philosophers, but rather where we ought to draw the line between likely and unlikely and for what purposes. It is entirely fair to suggest that evidence from social sciences—due to their methods and subjects of analysis—may in some sense be less robust than that from the empirical natural sciences. Yet this is a matter of degree, and not the hard impossibility versus everything else that theorists such as Estlund, Valentini, and McTernan suggest constitutes the relevant dividing lines between what suitably aspirational political theory should countenance and what it should not. Probabilities and social science work in a scalar feasibility space, not a binary one.52 This means that it becomes an open question—and a matter of judgement on the part of the theorist—regarding how to incorporate empirical evidence about the probabilities of various courses of action into our normative theory. We cannot simply rely on categorical distinctions such as impossible versus possible to settle this matter for us.

This paper’s argument so far has advanced one—arguably narrow—epistemological point, regarding the application of the hard and soft constraint distinction in political theory. It is therefore understandable for the reader to now ask “so what?”53 As alluded to in Section I, there has been a significant focus devoted to political theory methodology in recent years, and this includes attention to the multiple ways in which feasibility itself ought to be defined and deployed in theorizing.54 If we think of this paper’s argument as advocating for changing one element of theorizing, it will necessarily have different implications and consequences depending on how theorists are conceptualizing feasibility and also how these different conceptions are being used in different modes of doing political theory.55 The effects downstream of this paper’s argument are therefore not deterministic and will vary according to many other variables. With this pluralist disclaimer in mind, I will now sketch out what I consider to be plausible implications or effects on theorizing of accepting this paper’s argument.

It is worth acknowledging at the outset that there is at least one approach to political theory which remains entirely untouched by the above insight into feasibility. That is political theory in a purely evaluative key. Perhaps most closely associated with Cohen,56 this approach posits that theorizing, especially about justice, can be a purely evaluative exercise without normative action-guiding content.57 As an exercise in theoretical and not practical reason it is possible to theorize in this manner entirely unbothered by feasibility concerns, and so my argument will have no necessary implications for theory conducted in this manner.

For theorists who aspire to action-guiding implications, but also as much latitude as possible for theorizing, there is a way to read my findings as a form of liberation. For them the upshot from the foregoing could be that if impossibility is not readily determinable then it should not pose a constraint on normative theorizing. In other words, if we cannot identify categorically what is empirically impossible then the scope for theorizing is practically unlimited. Without the restraints posed by “ought implies can” theorizing can shift in a utopian direction entirely untethered from anything but logical or conceptual impossibility. For example, say a key block to achieving justice is communicative misunderstanding. A fully untethered theory might therefore require individuals to utilize telepathy when communicating to eliminate mental opacity and to ensure perfect understanding without miscommunication. The fact that such capabilities do not exist, and may never exist, is no impediment to such theorizing. Another untethered theory might premise institutions satisfying distributive justice on humans not requiring food or water to survive. There is, I do not doubt, a place for highly utopian political theorizing. Estlund’s book, after all entitled Utopophobia, is a salutary reminder that we should not necessarily be afraid to think and theorize boldly in the face of highly non-ideal or adverse circumstances.58 It may be that useful, interesting, or even true prescriptive theory can be developed which does not hold constant everything currently believed to be a hard constraint.

I note, however, that theorists defending fact-insensitive positions seem to want to hold fast to this flank of their argument, not entering into what I term truly untethered theorizing. McTernan asserts that normative theorists do not propose theories which conflict with the natural sciences, as she, “Perhaps charitably… assume[s] that political philosophers do not make such obvious mistakes: none, for instance, propose societies where human beings can live on air.”59 Similarly, Estlund notes that “disputes about whether political philosophy properly respects human nature are not normally about issues like our need for oxygen or our tendency to age. What political philosopher puts these aside?”60 Both McTernan and Estlund frame these reflections as if one would be engaging in a mistake to ignore these scientific findings when theorizing. However, the sociological observation that normative theorists do not tend to challenge natural scientific conclusions does not entail any reason in principle, given my argument, that they could not. As discussed above, the theories and findings of both natural and social sciences are revisable subject to future inquiry. What distinguishes them is the higher degree of confidence we—or at least many of us—place in the conclusions arrived at through natural scientific inquiry. If likelihood is irrelevant to questions of feasibility and only impossibility is accepted as a limit, then both natural and social scientific claims are equally irrelevant as hard limitations on theorizing. This ability to simply gesture at limits to theorizing—implying that it would be a mistake to breach certain empirical feasibility constraints—while holding open other possibilities no matter how implausible or unlikely, is undermined by my findings. Given these qualifiers by fact-insensitive theorists, the route I imagine many political theorists will, therefore, want to tread is somewhat different from truly untethered theorizing.

One effect of my argument is to draw attention to the necessity of specifying and justifying assumptions when theorizing. Political theorists already working within the hard constraint feasibility paradigm will have to recognize that what they are doing—as with all theory-making—is making a number of inevitable judgement calls as they develop their theories, including when it comes to feasibility constraints.61 It is important for us all to demarcate the scope and working assumptions underpinning our theorizing.62 Setting these out and scrutinizing them is itself an intellectually critical process. This is well illustrated by the sprawling debates spawned by the ideal and non-ideal distinction requiring theorists to reflect seriously on both the purposes of their theories and the appropriate assumptions to underpin them. Grappling with the proper scope of theorizing, including its limitations and assumptions, is a necessarily humbling scholarly process as we acknowledge the limits of what we know and can therefore credibly claim. It can also help clarify the political theorist’s position in a wider division of intellectual labor.63 Absent the fallback of a hard constraint, theorists who still wish to hold current empirical scientific understanding constant in their theories will have to provide different justifications. Some may acknowledge their general inexpertise (although no doubt some broad-minded political theorists are dual-qualified in the relevant natural sciences) to opine on the probability of current natural scientific laws being revised and so accept them for the sake of argument. More prosaically, it may be the case that it is harder for us to imagine possible worlds where our current scientific paradigms and laws do not hold compared to possible worlds with widespread preference or motivation changes.64 A pragmatic approach may reason that relying on highly improbable features such as certain laws of science being overturned may mean that the resulting theories are unlikely to be persuasive or useful to audiences.65

One somewhat odd potential upshot of this attention to assumptions might be a kind of bifurcation when it comes to using feasibility in the development of theories as opposed to critiquing or analyzing them. An oft-cited usage of feasibility is as a binary—to rule in or rule out theories as feasible.66 However, in a world where theory construction explicitly includes various abstractions or idealizations certain feasibility constraints can be assumed or imagined into or out of existence for the purposes of theory development.67 Indeed, the very same presuppositions could be held constant at one point but varied at another in order to develop an argument. An exemplar of this method is Carens who, in the appendix to his 2013 book, explicitly discusses varying which features of the world his argument assumes and challenges, what he calls a “shifting presuppositions” approach.68 Carens thereby provides a clear example of combining this level of methodological self-awareness with substantive theorizing. The binary conception of feasibility may be employed here but its application will depend on what the theory is trying to do and what assumptions are being made in service of this aim.69 As Carens notes:

Having adopted certain assumptions or presuppositions it may be that certain moves in the development of theory are thereby rendered feasible or infeasible. However, it is crucial to note, as Carens does above, that these feasibility assessments are downstream of the decision as to which assumptions or presuppositions one is incorporating into one’s theorizing and not whether they are necessarily “hard” or “soft.” What may result is the type of iterative theory development illustrated by Carens’s work—varying assumptions to see the impacts on the theory, or even in response to empirical developments, and then continuing to amend the assumptions and/or other elements of the theory.71 What this will produce is a series of binary feasibility assessments—depending on what is assumed and how the theory is constructed, different possibilities will be rendered feasible or infeasible. These categorical feasibility distinctions mirror, and are enabled by, the categorical usage of assumptions.

When critiquing theory along feasibility lines, however, it may be more common to adopt the scalar conception of feasibility.72 Critical interlocutors may challenge either the conclusions of theories or their assumptions as more or less feasible. Just as evidence exists on a probabilistic scale, so too does feasibility. The extent to which these challenges will concern a theorist will depend in part on whether the challenges are suitably attuned to the theorist’s aims and approach to theorizing. For example, ideal and fact-insensitive theorists are unlikely to be moved by arguments that existing human motivations are unlikely to meet their prescriptions.73 I therefore do not propose to argue for conditions or principles governing what presuppositions or levels of probability have to be accepted by political theorists as feasible or infeasible. Pluralism within the discipline entails a range of plausible approaches to normative theorizing, ranging from the most utopian or abstract theorizing to the highly empirically informed.74 However, it is crucial to note that all empirical assumptions can be considered more or less probabilistically true. As identified above, it is this failure to recognize the continuity of hard and soft constraints in a probabilistic manner which leads theorists astray who try to distinguish between empirically impossible and extremely unlikely. It requires recognizing that drawing such distinctions about which empirical claims to accept or challenge is a matter of judgement and not categorical impossibility distinctions. This places more fact-sensitive and insensitive theorists on essentially the same probabilistic terrain. Interestingly, this is a conclusion Estlund himself reaches:

A consequence of this realm of possibility and recognizing the probabilistic nature of feasibility assessments is that the exact boundary between feasible and infeasible is uncertain, and perhaps unknowable.76 This is a consequence of the probabilistic nature of the knowledge upon which our theories and their empirical assumptions are constructed.

This paper has advanced one main point—that our knowledge of what constitutes an empirical hard constraint is, like all empirical knowledge, probabilistic. From this flow one necessary consequence and a number of other plausible consequences for theorizing. First and foremost, hard constraints cannot be used in a categorical manner to demarcate feasibility boundaries between empirically impossible and extremely unlikely possibilities. At the epistemic level bounded by our limited knowledge, these two categories effectively occupy the same probabilistic space. Grasping this insight would help theorists referencing or discussing possibility avoid the types of errors discussed in Section III of this paper. Readers of this paper may therefore come away with the sense that what is being called for is conservative or more concessive theorizing. It is true that the paper’s critique has focused on theorists who have relied on hard feasibility constraints in order to provide a clear principled way to engage in ambitious theorizing. Reaching for hard empirical constraints has helped buttress the approach adopted by fact-insensitive theorists, and this usage will have to give way. With this option foreclosed, fact-insensitive theorists will need to return to the theoretical drawing boards to reconceptualize how exactly feasibility considerations feature and may be justified in their theorizing. However, a broader consequence of my argument may be to encourage theorists of various methodological approaches to be more explicit in setting out and justifying their assumptions. Some could be inclined to follow the example of Carens and theorize while holding various assumptions constant at different times or in different parts of their theory. To return to a point made by Carens referenced above:

Setting aside hard constraints means theorists will have to engage on a more appropriate shared terrain to work out their relevant presuppositions. When a theorist proposes or assumes that certain options or possibilities are open or foreclosed on grounds of feasibility, it will invite further inquiry as to what evidence they have for this assessment. Moreover, it will also encourage explicit reflection on why a theorist’s feasibility tolerance is set at any given probabilistic level. This may not always be a comfortable reflective experience for a theorist used to assessing the weight of normative argumentation rather than probabilities. This type of exercise may be more familiar to academic colleagues working on problems related to public policy, social choice theory, or evidential thresholds. Unlike those areas, however, the call here is not for a single normative procedure or standard such as expected utility theory or beyond a reasonable doubt. How feasible or infeasible a theory’s proposals are will justifiably vary for reasons which themselves are not simply a matter of counting probabilities, because the goal of a normative theory is not simply to redescribe reality or predict future events with the highest degree of accuracy.78 Still, to the extent that empirical feasibility is a live concern in our theorizing, our sensitivity to facts should incorporate the entire range. This includes the fact that claims which appear to be nomologically true have a nonzero probability that they too may be false or fail to predict future possibility/impossibility.