Between Subjectivity & Objectivity: The Desiderata of Any Future Ethics

A sketch of my own thoughts on metaethics. This essay tries to do a host of things, perhaps none of them particularly well, but I do say a little bit of what I want to say. Hopefully informative at least in marking out some conceptual space.

I. There are broadly three ways you might carve up positions on the metaphysical foundations of an ethical theory. You could be a moral realist, a moral constructivist, or a moral antirealist. While these distinctions might not be exhaustive, or discriminate existing theories in a fine-grained enough way to capture and appreciate their subtle differences, broadly speaking, they cover most of the really substantive metaphysical differences. Here I attempt to sketch out their general characteristics, some of their greatest difficulties, and gesture towards potential solutions to them.

A moral realist says there is and always has been a fact of the matter about what is good and what is bad. The same moral judgements have, and always will have been, true or false. They are true or false because they refer to some property actually out there in the world: a moral fact. Moral facts are normative properties of the world that determine what we ought to do (and/or what is permissible to do) in each situation. Some moral realists contend they are merely facts, and that there need not be properties over and above simply the facts. For example, the judgement "it is wrong to inflict gratuitous suffering" seems true, and it seems always to have been true. A moral realist says it is true because there is a moral fact "out there" that relates to the descriptive facts about people inflicting gratuitous suffering, such that there could not be someone performing such an action and it could be right. Thus, the moral realist says that morality is absolute, unchanging, and is in no way contingent upon human attitudes and opinions but weaved into the structure of reality itself, out there to be discovered by us.

A moral constructivist says, in contrast, that the truth of moral claims comes from our attitudes, culture, or another 'mind-dependent' truth-bearer. What does 'mind-dependent' mean in this context? It is useful term to use due to existing philosophical terminology, but we need not think of it as dependent specifically on minds, as this could mean a number of things for different people. All it means in this context to say that the truth of moral claims depends on something "mind-dependent", is that they depend on some aspect of human life, the things we feel, the desires we have, the attitudes we hold, the practices we enact, and institutions we erect, to name a few candidates of the kind of thing you should have in mind. A constructivist may put forth one, many, or even all of these things as the kind of thing that they think determine the truth of moral claims. It will be a matter of intra-constructivist debate where the proper emphasis should lie.

 

As an example, take laws. It seems like the sentence "murder is illegal in New Zealand" is true. However, piracy being illegal in New Zealand is not something that is written into the fabric of the universe in the sense that a moral realist thinks moral facts are, prior to human intervention. Nor did we discover this fact after years of searching for it . Rather, its truth seems to depend on facts about legal conventions and agreements that we humans have created collectively, plus the practices we enact that make them legitimate and which make some people legitimate subjects of them. This is all that is meant by mind-dependent. Thus, the moral constructivist will say that judgements like "it is wrong to inflict gratuitous suffering" are true because there is a moral community that agrees or holds attitudes that it is wrong, or partake in practices that forbid it. There are no moral facts "out there" to be discovered, there are just different groups of people at different times and places, who come to agree about what is right and wrong through the enactment of moral practice, just like the law, and this is all that morality is. Though, of course, for the constructivist, it will tend always be a more fundamental a phenomenon in our lives than something like laws tend to be, due primarily to the fact that it more overtly governs our interpersonal relations.

Note that this does not mean that what is right and wrong is for some individual is determined by what that individual thinks is right and wrong (subjectivism, as typically understood). It is that right and wrong are determined by institutionalised conventions or collective commitment to specific norms by some group (relativism, as typically understood). Once the conventions or collective commitments 'establish' some norm, there is a truth of the matter about what is right and wrong (as there are with laws), which means there are moral facts, it will just be relative to a time, place, practice, and community.

However, there is another kind of constructivism that is relative to humanity, though not necessarily to a time, place or community. Call this ideal-constructivism and call what I just discussed, contextual-constructivism. Ideal-constructivism argues, with its contextual counterpart, that morality is relative to the human being. However, it says that there are certain conditions universal to all humans (or rational beings), under which we can derive a morality that is universal for all humans (or rational beings). There are both Aristotelian (where the conditions are supplied by a substantive vision of what the good life is for humans) and Kantian forms (where the conditions are supplied by categorical principles of practical rationality) of ideal-constructivism. But there are also other theories whose ideal conditions are something like "things all rational people would agree to (in this situation)." Once again, (some of) these views think there are moral facts, but that they depend on (are relative to) features of human beings.

A moral antirealist says that there is no fact of the matter about what is right and wrong. There are, broadly, two ways this could go. On the one hand, an antirealist could think that all moral judgements are false. This includes judgements we would ordinarily think true and ordinarily think false. Thus, on this account, both "inflicting gratuitous suffering is wrong" and "inflicting gratuitous suffering is right" are false. They do not think this because they think 'anything goes' but rather because there are no moral facts (propertied or otherwise) to make any judgements true. However, it is an alleged consequence of this view that anything does, in fact, go. (This view is called error theory.) On the other hand, an antirealist could think that all moral judgements are not the kind of thing that could be true or false because they are not judgments of (moral) fact. Rather, they are expressions of attitudes or norms that an agent is committed to. On this account, "inflicting gratuitous suffering is wrong" does not mean to represent some feature of the world, wrongness, which is co-present with any inflicting of gratuitous suffering. Nor a fact which is related to such act. Rather, it expresses our distaste for inflicting gratuitous suffering or our commitment to the norm that it is not sensible to inflict gratuitous suffering on others. While these two accounts differ in substance, the important part for my purposes is that both views think there is no such thing as moral facts, and thus no way in which it is true that something is right or wrong.

II. Each of these views has serious problems. I begin with moral antirealism. The basic and longstanding issue with any moral antirealism is, no matter how ingenious the formulation, that ultimately, it seems to imply that, in some sense, anything goes. What do I mean by that? What I mean is that if moral antirealism is true, then it is impossible that there could be any standpoint-independent criteria, or grounds, for condemning or praising any action whatsoever. By 'standpoint-independent' I mean independent of some or other perspective, such as your own or that of someone else. Suppose a person in front of you starts torturing another person, but you are an antirealist. You will almost certainly feel anger and condemn this this. You will likely try and stop them. You will feel that it is insensible of them to be doing what they are doing, and sensible of you to stop them. However, you would neither be (standpoint-independently) right, and they (standpoint-independently) wrong, to do so. There is nothing in your account of morality that you can point to that would show that what this person is doing is wrong, except for your own negative feelings, your own standpoint, towards it. 

The problem is not necessarily that your own feelings or commitments are not valid for you, they certainly feel that way and in some sense they certainly are. The problem is that the person's feelings seem to be just as valid if moral antirealism is true (if it even makes sense to speak of validity), as their standpoint is just one among many, including your own. But this means that there is no procedure by which to validate one set of attitudes as the correct set, rather than another. Even if your attitude's condemn this person's actions, and do so regardless of their attitudes, they have no reason to listen to you because your attitudes lack the authority that standpoint-independent moral facts seem to supply. If they asked you: "Why should I stop torturing this person?" and we assume that they are fully rational, they have no authoritative reason dictating that they stop. After all, there is no truth of the matter as the wrongness of the act. 

This is not the case with descriptive facts. If we wished to adjudicate whether it is raining outside or not, all we would have to do is check whether it is, in fact, raining outside. If it is raining and someone continues to insist that it is not, then we can say they are wrong and we can do so with an authority that extends beyond our own feelings, we can do so with the truth at hand. We cannot do the same for moral claims, if moral antirealism is true, because there are no facts about whether some action is right or wrong. Morality would merely be a 'matter of taste', in the most extreme sense. If no other account of morality can be made to work, and this is morality as we have to come to terms with it, then so be it. However, it is arguably quite the price to pay. Call this the problem of evil for any moral system.

It is worth adding that sophisticated non-cognitivists (the view that moral claims are not the kind of thing that are true or false) and subjectivists (who think that there are only standpoint-dependent moral facts) would be outraged by my depiction of their view. They would say that complaining about the fact that there are no (moral) facts of the matter to adjudicate moral disputes is precisely to miss the point of their accounts, which attempt to show that looking for such a criterion is to misconceive the nature of morality, or to ask of it more than we should feel is required. If we are to dispute them, they will say, it should be on these grounds. Such accounts are usually buttressed with complicated accounts or analyses of moral reason and behaviour. However, no matter how much I try to understand these views, they always come out sounding, to some extent, implausible and gerrymandered. Morality is the kind of thing where we do wish to appeal to something independent of our own mere feelings or attitudes towards something, and whether or not that thing turns out to be moral properties or facts, there seems to be something more going on here. There is truth and falsity, right and wrong, or at least some kind of validity and invalidity, in matters of morality. I don't pretend to have refuted antirealism here, just to show what is, in my view, its primary difficulty. Onto realism.


III. It does seem like a virtue of the moral realists' account that the facts which make moral judgements true are independent of humanity. This is because it completely evades the problem of evil, given that objective moral truths are something we can appeal to in any given situation. However, I will argue, it is actually a vice. Moral realists could think that the moral properties are non-natural, or natural.

If the moral realist has it that the moral facts are non-natural, a serious issue arises. Given that moral facts do not figure in the spatiotemporal nexus of the universe (because they are non-natural) they could be changed entirely without us noticing. For example, take two nearly identical worlds. In the first world, there is a finite set of non-natural moral rules R1 governing all actions. Suppose these rules roughly correspond to the values that appear to manifest themselves in our actual world. In the second world, there is also a finite set of non-natural moral rules, R2, governing all actions, except each of the rules in R2 is the negation of the corresponding rule in R1. The worlds are identical in all but these moral rules. Suppose now that each world is identical to the actual world, with all of our systems of morality, passions, desires, viewpoints, cultures, etc. One problem is that if realism is true, we can never know if we are in a world like the first, where the normative properties line up with our actual judgements, or if we are in a world like the second, where the normative properties are orthogonal to our actual judgements. However, even more fatal for realism, even if we did know all of the realist moral facts, they do not actually seem to be the thing that would determine what is right and wrong in each of these worlds.

To see why this is a problem, suppose you somehow knew you were in the first world. Now suppose you were transported to the second, and you somehow know that you were. In this world, you know for a fact that inflicting gratuitous suffering is good, and helping others is bad. Should you, therefore, torture others and be selfish? The answer surely seems to be no, given the world is descriptively identical to your own. If this is right, then we should say that the non-natural moral rules of world 2 would not determine what we ought to do even though the moral facts would make actions adhering to those rules, right. This produces a contradiction: how could it be that we ought, and ought not, perform some action? Yet that is what realism entails in such a world, if it were true. Even worse for realism is that we don't seem to have the ability to tell the difference between these worlds, even though we have the same moral clarity around the same issues in both of them. Indeed, it is not implausible that if there were such a thing as an eternal moral realist morality, the actual facts of the matter could look completely alien to us. But if this is the case, would we really call that morality? Would we really say that that is what we ought to do?

The lack seems to lie in the fact that the possible moral facts are not sufficiently human, not sufficiently anthropomorphic. By this, I mean that it is possible that moral properties could be configured in such a way that what they dictate could not actually deserve to be called morality, nor actually bind us to do what it dictates. This suggests that something else other than non-natural moral properties determines what we should do as humans, something grounded in the way we actually are and actually relate to one another (which is the kind of thing that contextual and ideal constructivists latch on to). Call this the anthropomorphic requirement of any system of morality. Note that I do not mean that our ethics must be 'human' in a sense that would exclude 'non-humans' (animals, environments, artificial intelligence, aliens, etc.), as this would surely be a grave mistake. I just mean that our metaethics must coincide non-accidentally with the actual practice of morality in order to capture what morality is. Theistic ethics faces a similar problem to the moral realist. The idea is to think: if god commanded that we should inflict gratuitous suffering, should we do so? Would it even count as morality if he were to command it? No. In such a case we really ought to devise our own morality, even if there is some objective fact of the matter otherwise, dictated by a supreme being.

Now, I am not saying that moral realism necessarily fails to meet this requirement, or is refuted by it. The problem is that if it did fail (which seems possible), we would neither be in a position to know that it did nor to follow these 'true' rules were to to know. This is a serious problem for the view not only because it is impossible to ever verify whether we are in such a world, but also because even if it were verifiable they would not even ultimately be the last word on what one ought to do. But if they are not even the last word on what one ought to do, then why should we consider these facts the foundation of morality? Which is to say, it seems at least possible that we could be in a situation where we ought not do what we ought to do, and that possibility is unacceptable. What this shows, I think, is that something other than completely mind-independent moral facts is at play in our theorising, and that morality must have something to do with how we actually live, not some realm of facts entirely independent of us.

One simple and obvious reply to this line of objection is to say that moral facts systematically supervene upon descriptive facts in such a way that the two worlds described above would be impossible. A property supervenes upon another property if it can't change without the other changing. On this view, you could say that moral facts supervene upon descriptive facts as a matter of necessity, that one world cannot have different moral facts unless it has different descriptive facts. Thus, without changing something about the structure of these worlds, there cannot be a difference in moral facts. There is some plausibility to this, as it does seem that, given the structure of the human nervous system and of human behaviour (the descriptive facts), inflicting gratuitous suffering on another will be wrong in any possible world in which these facts hold and therefore that it is nonsensical to postulate the possibility of worlds in which it would not (and it seems to be precisely this point I was originally exploiting against the realist). 

However, going this route has some issues. For one thing, the necessity involved here does not seem to be logical or conceptual as it neither contradictory nor incoherent to imagine that gratuitous suffering could be good, construed realistically, in a world. This means that some explanation has to be provided for its necessary truth, for why the moral necessarily covaries with the descriptive. But there is some difficulty in doing this given there is no clear logical or conceptual inconsistency with supposing otherwise. And given this is so, their relation must be a  brute, synthetic necessity, thereby introducing severe epistemic and metaphysical difficulties. For another, the problem with accepting this strong supervenience between the moral and the descriptive facts is that the moral facts end up relying on the descriptive too much, such that they become explanatorily superfluous when compared to views that fully explain morality without reference to moral facts, realistically construed. Such accounts, once they connect themselves with our moral life in such a way as to satisfy the anthropomorphic requirement, seem inevitably to be preferable to the views under consideration.

The other alternative is that the moral realist might has it that moral facts are natural facts. In this case, you could either think that goodness and badness are analysable into an unproblematic descriptive property, like pleasure, or that goodness and badness are properties in their own right that we discover by empirical investigation (just as the economy or the family can be investigated in such a way). If we accept the former account, that goodness and badness are analysable into other properties, the moral realist faces a serious problem.

The problem is that someone can always intelligibly question whether any proposed analysis exhausts the meaning of goodness (or badness). Suppose that the natural moral realist thinks that the meaning of "good" is pleasure. When we say something is "good" we mean that it is pleasurable, where pleasure is a purely descriptive state of affairs. Someone could easily think that other features of the world like art, freedom, or friendship are good things too, independent of whether they produce pleasure. It is sufficient that this point of view, that other things are good, is even intelligible for it to refute the naturalist moral realist on this point. This is because the burden is on them is to show that 'goodness' just means pleasure, and nothing else. For, if this view is intelligible, then 'goodness' does not mean pleasure after all, as it could mean a host of other things too. 

Even if you disagree with that person, they still used the word in a way that made sense, meaning 'good' cannot mean pleasure. Any such attempt to analyse goodness in a way seems to deny something essential about goodness: that it is dynamic and heterogeneous, that it could be applied to a number of different things, and that it is essentially open to what we consider good and what we consider bad. We can see this because all static definitions of the good would fail similar tests that pleasure does. Call this the openness requirement of any system of morality. Morality cannot be reduced to a simple formula, meaning, or identity with a descriptive property, lest we deny the essential openness of what could intelligibly be considered good. There is a potential infinity (in the Aristotelian sense) of counterexamples to any definition of good and a moral theory that denies this denies this fact of moral life.

If we accept the latter account of natural moral realism, that goodness and badness are their own distinctive kind of property, this satisfies the openness requirement because we could discover it to be present in all kinds of different situations, with different moral ingredients. However, the issue now is, that it is still a puzzle how this property makes it such that we ought or ought not to act in a particular situation. It is a puzzle because no other descriptive facts, such as the colour of grass, the smell of the sea, the chemical composition of iron, or the state of the economy, at least by themselves, tell us how we should act. This is what Mackie calls the 'queerness' of moral facts (and it applies to both natural and non-natural moral facts). Their queerness consists in the fact that moral facts purport to make demands on us (like that you ought not kill or steal) that descriptive facts do not. However, if we take the moral realist route under consideration, we are arguing that moral facts just are more descriptive facts. But if they just are more descriptive facts, what is it about them that tells us what we actually ought to do? By themselves, it seems difficult that such facts could tell us anything at all.

And indeed, if this is the case, what stops the same problem that we had with the moral antirealist from arising? Given that goodness is merely another descriptive fact, i.e., given that it is just like being red, or being made of iron, what makes it the case that people actually ought to act in accordance with the dictates we hope would follow from such facts? It seems that obligations to act this or that way, or within this or that range of behaviours, only follow from the existence of descriptive facts if we also have some goal in mind. For example, if I want to melt lead to make sinkers, then I ought to use an apparatus that can get the lead to its melting point, where the ought in question is wholly dependent on the ends I have in mind. However, if naturalist moral realism only has authority for us in this sense, then what is to stop those who simply don't care about the proliferation of goodness and the destruction badness? If we say they ought to care, they could simply reply: "why? I don't care whether good things are brought about and bad things are destroyed." If the ought is only an ought in this instrumental sense, then the natural moral realist has no leg to stand on in their reply. Call this the bindingness requirement, the requirement that morality is authoritative for us in a way that is not reducible merely to our goals, but in some distinctively moral sense.

The solution to this could perhaps be to revert to a non-natural moral realism with a strict superveinence between the moral and descriptive; the authority is right there in the facts. However, a line indicated by the demand the anthropomorphic requirement sets out could also be pursued. If we can identify descriptive facts about us that actually do bind us in practice, and in the right kind of way, then we may be able to show what descriptive facts really are binding to us in a way that is not simply instrumental, but distinctively moral. Onto constructivism.


IV. Constructivism certainly has some virtues. First, it almost necessarily satisfies the anthropomorphic requirement. The whole idea is that morality is explicitly based on the norms and attitudes of existing humans, existing communities, and their ways of relating to one another. It is impossible to have a constructivist morality that is not ultimately grounded in human activity. This is true for both its contextual and ideal forms. Second, it satisfies the bindingness requirement, at least those versions that countenance the institution of distinctively normative facts (rather than merely descriptive ones). This is obviously tied up with how it relates to human activity. However, there are a few problems with these views. I begin with contextual constructivism.

Contextual-constructivism has two major problems, as I see it. First, it is historically fragile. I mean this in a double sense. On the one hand, it means that if there were societies in history that accepted some obviously immoral practice, the constructivist seems to have to bite the bullet and say that it was good for them to do so, at the time. We do not even have to come up with ridiculous fictional examples to see that this has happened in our actual history. Many past communities engaged in and endorsed the enslavement of others. No doubt there were dissenters in each of these cases, but in many of these cases they were the exception rather than the rule, and it is surely the rule that determines rightness and wrongness. But surely the enslavement of others was not good, even at the time in which people accepted it. On the other hand, this means that if there come to be societies in the future that largely endorse enslavement, or other seemingly immoral actions, being a contextual relativist means we cannot say that they will be wrong, from their perspective, to engage in those actions. But surely they would be just as wrong then, as they would be now. 

We do not even have to traverse time to find this problem, we can cross space too. Look around the world: there are norms that people engage in that are wrong, from our perspective, but not from theirs. We can still think we are right to condemn such norms, but constructivism seems to make that condemnation kind of like condemning someone facing you for thinking that something to your left is to their right. We tend to want one or another of any set of contrary positions to 'ultimately' be the right one, and a rudimentary contextual-constructivism struggles to provide this. Some story must be told about how to legitimately adjudicate disputes between incompatible sets of norms in such a way that the adjudication process could validate, or at least attempt to validate, one course of action rather than another, for both groups.

The second problem with this view, at least on a particularly vulgar reading of it, is that it is formally (not politically) conservative. Here's why. If the right action in some situation is determined by what the existing community accepts as norms, then no one is ever justified in acting in any way other than the ways already sanctioned by existing norms. But this means that no one is ever right to attempt to overturn the existing norms and institute new ones that they think are better. This is absurd. People plainly were justified, in many historical situations, and now, to attempt to overturn existing norms. Of course, constructivists will point out that societies do, in fact, change their moral norms over time by coming to accept new ones. However, just because we do change over time, does not mean that the account can justify those changes, at each point in time. Once the new norms are accepted, the norms justify themselves (merely in virtue of their existence), but we were, and never can be, justified in getting to that point. This is a rather strange result, one that must be avoided. Moral improvement must be justifiable.

A more sophisticated moral constructivism thus requires higher-order norms that govern the acceptance of new norms such that we can (1) adjudicate disputes between people or communities that have incompatible sets of norms, and (2) justifiably correct or improve current norms, or generate new ones to face new situations. That is to say, the constructivist must postulate some procedure or test that is internal to any system of norms that determines the conditions upon which a change can be legitimately posed and then accepted or rejected. To see what I mean by this, note that we have precisely such norms in systems of law. 

The existing set of laws does not uniquely determine the lawfulness or unlawfulness of every possible action, for all time. Rather, there are processes built into the existing laws (further laws) that tell us when it is legitimate for a new law to be created, or an old law to be changed. They are higher-order laws. In the case of New Zealand, laws are legitimate when they are passed through the proper legislative process, which requires both the giving of reasons and convincing others of the force of those reasons. When such processes are followed, both the proposal and acceptance or rejection of some law that contradicts existing ones are sanctioned by the higher-order laws governing all of them. This means there is no formal conservation of existing laws because there is a process internal to the system that allows it to justify its evolution. And because such laws are internal to the system, the law can justifiably change and evolve without appealing to standards beyond the existing system, but standards that are actually internal to it. 

Higher-order norms like these will be necessary for the contextual relativist not only to avoid the formal conservation of morality and justify moral change but also because they explicitly deny the realist view that there are standards over and above individual practices of morality. These are unnecessary with higher-order norms, the thought goes, because they are internal to the practice of morality, not external to it. However, some story about what such norms would actually look like in moral practice must be told and postulated to actually exist, for this to work. 

This is problematic for the following reason. If we are to have higher-order norms, they seem to ought to be good higher-order norms. But on this view, what is good is only that which is already sanctioned by such norms. And if the higher-order norms are not permissive enough such that they can be overcome in favour of better ones, then morality gets trapped in a cycle it can not escape. Thus, the higher-order norms need not only to govern the acceptance of any old norms but also contain within them the possibility of their own overcoming, for better or more appropriate norms for the time. Call this the dialectical requirement of any system of ethics.


V. The main problem with any ideal-constructivism that relies on the nature of human beings is that it is objectionably essentialist about the nature or goodness of human (or rational) beings. That is to say, it must accept a substantive view of what is the good life for all human beings when it is not clear that there is a single vision of what the good life is. On an account like this, one might say that what is good for all humans is virtuous or rational activity, or acting in accordance with the categorical imperative. But what does this actually mean? There are two ways this could go. 

First, the ideal-constructivist could give a specific account of what virtuous activity consists of in all humans. However, if one does this, they engage in what we might call 'bad empiricism.' Such an account ends up inevitably being bad empiricism because whatever criteria could possibly be put forward for living the good life, we can always observe, quite plausibly, that someone could come along and live a good life without following the route set out by the view for achieving the good life. And if there are people that can do this, then the kind of activity set doesn't seem to be able to support itself as the metaphysical foundation of morality.

For example, if we say that the good life is seeking pleasure for yourself, we can imagine someone living a good life sacrificing their opportunities for pleasure so that others can have it. If we say that the good life is a life spent contemplating philosophical questions, we can imagine a person living a good life helping others, but engaging in little contemplation about it, or anything else. If we say that the good life is to be a stoic, we can imagine someone living a good life as an Epicurean, and vice versa. If we say that the good life is to be a Buddhist, we can imagine someone living a good life as a Confucian, and vice versa. As the naturalist moral realist did above, this view fails to satisfy the openness requirement. It fails because it denies that there are vast and heterogeneous ways in which one can live a good life, because there are vast and heterogeneous ways in which goodness is actualised. This cannot be denied, at risk of failing to account for the heterogeneity of human existence, at the risk of engaging in bad empiricism that postulates a single concrete end of all human existence.

Second, in order to get around this, you could relativise what virtuous activity is to the individual and say that what the good is life depends either on the individual peculiarities or historical circumstances of that person. (The more common route is the latter.) However, this route has two issues: the former is vacuous and the latter succumbs to a problem of evil. It is vacuous if we say that all virtuous activity is indexed to the individual because it does not actually say what a good life is, so it fails to offer an actual theory with actual guidance. It merely amounts to the tautology: what is good for you is what is good for you, what it is to do well for you, is to do well for you. For it not to be vacuous it must either give some positive guidance or rule some things out (which really amount to the same thing). This is what a historical ideal-constructivism will do. But, it succumbs to a problem of evil because it seems to lack the resources to condemn actions or activities internal to and accepted within the historical paradigm that determines the good of some action or class of actions. If someone is an extremely good serial killer, should they try their best to kill as many as they can? If what is virtuous just is what one could flourish doing, relative to your individual capabilities, I do not see why not. If someone grew up in a tradition of slaveholders, should they try their best to carry on that tradition? No, they should reject its moral trappings.

These last two points show two things. They show (1) that we must accept that there is no single way of living a good life that could be captured in a formula that we can read off of human nature but also (2) that there must be some criteria or mechanism according to which we can rule out certain ways of living as wrong (without also running afoul of [1]), and it must be done somehow from inside the system (the dialectical requirement). Morality must neither constrain itself too much, nor too little. Call this the plurality requirement which constrains the prescriptions of any moral theory.


VI. To sum up the preceding, the requirements outlined above are as follows:

• The Problem of Evil: a successful moral theory cannot be such that there are no situations in which some actions can legitimately be said to be right and wrong, or at least valid and invalid.
• The Anthropomorphic Requirement: a successful moral theory must not be such that it could prescribe something utterly alien to us, it must coincide with the actual practice of morality as it is, or at least possibly be within reach of actually practicing moral agents.
• The Openness Requirement: a successful moral theory cannot exhaustively define what it means for something to be "good" by reference to a single formula or single good, such as pleasure.
• The Bindingness Requirement: a successful moral system must give an account of how whatever morality consists of obligates an individual to act according to its dictates. It cannot be rightly ignored.
• The Dialectical Requirement: a successful moral theory must have norms internal to it that govern the acceptance or non-acceptance of new norms; or, at least somehow allow for the valid interrogation of any moral precept. It must say what this consists of.
• The Plurality Requirement: a successful moral theory must be prescriptively plural. However, the amount of plurality must be right, it must not prescribe too much, or too little, and it must have some procedure for settling disputes about it.

These are the desiderata of any future ethics, desiderata that may or may not be able to be satisfied by the views I have just discussed. Because of course, at no point above did I tackle the strongest versions of them, nor develop them at much more than a sketch. For example, I implicate Aristotle as someone who is not able to strike the right balance between these criteria, but I see no textual evidence in his Ethics that he puts forward an objectionable kind of essentialism, even in the argument from function. Squinting, I think that his notion of practical wisdom, outlined in Book XI (and as expanded upon by John McDowell), is his attempt to bridge the gap between the above two fetters of ideal-constructivism by positing a kind of 'higher principle' that avoids settling in advance what kinds of activities one would do well to enagage in. However, that being said, I do see them as some of the most fundamental difficulties any such theory will face, or at least those difficulties that point us towards the fundamental ones. And indeed, my current view is that no form of realism I am aware of (as I have defined it) can get around the anthropomorphic requirement, and no form of anti-realism I am aware of satisfyingly gets around the problem of evil. Though there are still some days where I wake up and non-cognitivist expressivism makes sense to me.

Thus, the question now becomes: does any existing meta-ethical theory fulfil each and every one of these requirements? Is it possible that a system even could? I think the closest approximation of such a system is Habermas' discourse ethics, which I outline here. The question then becomes: can we do better? I suspect, with Habermas, that a satisfactory answer must begin, if we are to unearth the nature of morality, from the the distinctive phenomenology of moral practices, the way these practices are actually lived, and the way that our practices of justification prop up, or eschew, specific moral judgments or specific ways of life, in specific situations.