A sufficiency threshold is not a harm principle: A better alternative to best interests for overriding parental decisions

Douglas Diekema argues that the so-called ‘best interests’ standard is not the appropriate threshold for overriding parental decisions.1 Instead, he proposes that parental authority over children should be limited by a harm threshold. Others may legitimately intervene if parental decisions would expose the child to significant risk of a serious and preventable harm, but not (as the phrase ‘best interests’ suggests) simply because their decision fails to be what is best for the child.2 Diekema’s paper is widely cited and so influential that others now refer to a growing ‘harm consensus’.3 Thus, Diekema is representative of an increasingly popular view on the limits of parental authority, though one that has also come under attack.

Before introducing my own criticism, I wish to clarify and defend what I take to be correct in this argument: neither the state nor medical professionals may override parental authority simply because those decisions are not in the best interests of their child(ren).
Our actual practice is not based on what is literally best for the child.
One might respond that the term 'best interests' is widely understood in less literal fashion. For instance, Kopelman concedes that 'if taken literally and without qualification, it [the best interests standard] instructs us to evaluate all options and act on that option providing the absolute best outcome for the individual in question, without regard to any one or anything else'. 4 Yet, while this may seem like reason to reject this standard, she maintains that 'it does not require us to act in accord with what is literally best for a child, ignoring all other considerations…. Rather, it requires us to focus on the child and select wisely from among alternatives, while taking into account how our lives are woven together '. 5 This non-literal interpretation is how the (so-called best interests) standard is often applied in practice. 6 Clinicians are generally sympathetic to family interests-besides those of the child-and reluctant to involve courts, even when they disagree with parental estimations of best interests. 7 But this is Diekema's point: the standard actually used for legal intervention 'no longer seems to be a best interest standard but some other threshold '. 8 Referring to this threshold as 'best interests', even if that is intended in some non-literal sense, risks causing confusion.
Other authors have defended using the language of best interests. Coggon suggests that there may be pragmatic reasons to retain the phrase 'best interests' because it emphasizes the high level of concern that we should give to vulnerable children. 9 Any alternative label may result in diminished protection. However, this reasoning is speculative; there is no clear evidence that terminological change will weaken protections for children. Even when decisions are couched in terms of the child's 'best interests', judgements of these best interests are often twisted to support whatever intervention is favoured. Thus, paying lip service to the 'best interests' standard may actually distract us from what is really in the interests of the child in question. 10 Birchley argues that 'best interests' is preferable because it is seen as less pejorative than 'harm'. 11 Birchley, op. cit. note 3. However, this is another empirical claim, for which he gives little evidence. Even if it is true that courts see things this way, it may not be the view of parents. They are likely to resent any interference and may prefer a higher threshold for intervention, whatever language it is couched in. 12 Since the reasons for retaining 'best interests' strike me as uncompelling, I agree with Diekema and others that this label should be replaced by some more accurate term. 13 However, though I agree with criticisms of the best interests standard, I think Diekema is wrong to identify this alternative standard with a 'harm principle'.
(2020). The liberty of the liberty principle. Res Publica forthcoming (published online 21 January 2020). 26(3), pp. 337-355. 15 Birchley, op. cit. note 3, p. 114, rightly notes that simply renaming our test will not deal with substantive problems. However, we need not choose between these two options; we can both use a more accurate term and try to tackle problems with the test. Indeed, using more accurate terminology may help in resolving substantive problems. 16 MacDougall, op. cit. note 2, p. 282 sees the disagreement as terminological: 'there is actually widespread agreement that intervention is justified when harm to the child is projected to cross some threshold. The disagreement is primarily about whether this threshold is best described under the best-interest paradigm or under the harm principle'.
Continued reference to a best interest standard simply confuses physicians and others who must determine when parental refusals of consent should be tolerated and when state intervention should be sought…. In practice, it appears that the standard applied to parental decision-making for children is not truly a best interest standard, but rather something else. 17 Here, it seems no substantive change in practices of intervention is required, but merely a more accurate labelling of the threshold that we already use in practice.
However, other points suggest more than a mere change of terminology. Having argued that our real test for intervention is not properly described as best interests, but harm prevention, Diekema proceeds to 'further define the harm threshold'. 18 These efforts suggest at least a degree of substantive change; if his proposal were merely to relabel the standard that is already applied in practice, there should be little need to elaborate on what this is. In fact, he says only that it is 'consistent with the threshold level suggested by most commentators and applied by most courts '. 19 This suggests that some apply other standards.
Thus, when Diekema says that the best interests standard is 'the wrong standard', 20 it is unclear whether he actually means that the standard is substantively correct but wrongly described, as some of his remarks would suggest, or whether he is actually calling for a substantive change in the standards determining when intervention is legitimate. In fact, it is a bit of both.
The potential confusion is because, as noted above, the 'best interests standard' is actually used in different ways for different purposes. 21 If it is understood literally, as requiring what really is best for the child, then this is the wrong standard for intervening in parental decisions. 22 Parental decisions are not generally overridden simply for falling short of what is best for the child, so it is misleading to call our actual threshold for intervention a best interests standard. Medical professionals will rarely challenge, or seek to override, parental decisions, unless they think there is a significant risk of harm to the child. 23 On the other hand, if 'best interests' is understood less literally, as sometimes suggested, 24 then it may be the substantively correct standard, but confusingly mis-described. If this is indeed how it is used in practice, then Diekema's harm threshold need not significantly alter current practice, except insofar as it may reduce confusion over this standard.

| PRI OR CRITI QUE S OF D IEK EMA' S HARM PRIN CIPLE
I believe Diekema is right that the proper threshold for intervention is significantly lower than anything short of the best. However, I will later argue that our standard for intervention need not be identified with harm. This is also misleading. Before going into this criticism though, I wish to show why some existing critiques of Diekema's proposal are unconvincing.
First, critics point out that Mill claims to derive his harm principle from his utilitarianism. 25 This may seem problematic for those who reject utilitarianism. It may be possible to provide alternative justification for the harm principle, but Bester argues that 'if it is possible to reframe the harm principle and ground it in a more widely accepted or authoritative different ethical framework, it is no longer Mill's harm principle, but something else at work'. 26 As it happens, I agree that Diekema's principle is not Mill's; this is why I prefer to call Diekema's proposal a 'harm threshold' rather than the 'harm principle'. 27 However, Bester's argument seems to confuse the harm principle with its grounding.
While Mill believed the harm principle to be grounded in his utilitarianism, some interpreters have thought them in conflict. 28  harm principle does not intuitively seem like a foundational moral principle. Therefore, there must be some more ultimate moral theory that justifies it. 32 However, this criticism seems to confuse the logical and epistemic grounding of moral claims. 33 It may well be that some more ultimate theory justifies the harm principle, but it does not follow that we need to identify this ultimate theory before we can have confidence in any putative moral principle. 34 We often have more confidence in the truth of particular moral principles than in the moral general theories constructed to explain them; indeed, this is often how we test the acceptability of moral theories. We usually start from particular judgements or principles and construct theories that coherently explain (most of) them. We do not generally require a complete moral theory first.
There are other criticisms of Diekema, but I have chosen these in particular, first, to illustrate why I find some existing criticisms unconvincing and, second, because these points help to clarify Diekema's project, at least as I understand it. Those who advocate a harm threshold, as an alternative to the best interests standard, need not be concerned about the grounding of their principle. The harm threshold can be justified in the same way as other principles, including the best interests standard itself, by appeal to our considered intuitions and by incorporation into any acceptable moral theory.

| D IEK EMA' S HARM PRIN CIPLE IS NOT MILL' S HARM PRIN CIPLE
Though I have argued that one could endorse Mill's harm principle for non-utilitarian reasons, there is some truth to the criticism that Diekema's 'harm principle' is not Mill's. To be fair, Diekema does not explicitly say that it is, although the fact that he cites Mill before introducing what he simply calls the harm principle certainly gives that impression. In fact, Diekema's principle is closer to Feinberg's, which he also cites. Perhaps confusion arises because Diekema does not emphasize the differences between Mill and Feinberg. 35 Thus, while he might not be confused on these points himself, he is somewhat culpable for any misunderstanding.
While both Diekema and Mill are addressing when it is permissible to limit someone's freedom, their concerns are rather different.
Mill's harm principle does not say that it is wrong to cause others harm, or that we are permitted to interfere with people's actions whenever they do cause harm to others. Rather, it says that preventing harm to others is the only good justification for coercive interference. 36 Mill's purpose is to preclude other supposed justifications for interference, in particular the agent's own good. As he puts it, 'the only purpose for which power can rightfully be exercised over any member of a civilized community, against his will, is to prevent harm to others. His own good, either physical or moral, is not a sufficient warrant '. 37 If an action does not harm others, then it should be immune from interference. However, Mill adds that 'it must by no means be supposed, because damage, or probability of damage, to the interests of others, can alone justify the interference of society, that therefore it always does justify such interference'. 38 Thus, Mill does not argue that we ought to interfere with action that harms others. Again, his point is that we ought not to interfere with what does not harm others. Hence, it is not really apt to characterize Mill's harm principle as a liberty limiting principle.
While it does concern the limits of individual liberty, it does so in the sense of identifying the minimum freedom that should be protected for all competent adults, rather than where our freedom ends. The only liberty it limits is the liberty to restrict the liberty of others. In short, Diekema's references to Mill are misleading, for the 'harm principle' that Diekema proposes is not that defended by Mill.
Since the phrase 'harm principle' is so closely associated with Mill, I prefer to describe Diekema's proposal (as he sometimes does) as a harm threshold or standard for justified intervention, rather than a harm principle.

| PROB LEMS WITH THE HARM THRE S HOLD
Thus far, I have chiefly sought to clarify and defend the critique of the best interests standard. We do not generally take ourselves to be justified in overriding parental decisions whenever they fail to maximally promote the interests of the child. It may be that no one has ever seriously defended such a literal best interest standard, instead using the term 'best interests standard' in a looser manner. 48 However, if the standard that we use does not actually require us to do what is in the child's best interest, then I see no good reason to continue calling it a 'best interest' standard. This merely invites confusion, particularly when such language is used in front of laypersons not trained in medicine or law.
However, while our actual standard for intervention is not best interests, at least if it is taken literally, I believe Diekema is wrong to identify this standard for intervention with harm. After surveying a number of proposals for when intervention is justified, Diekema sums them up by saying: These commentators seem to be suggesting not a best interest standard, but rather a harm-based standard for intervention. The real question is not so much about identifying which medical alternative represents the best interest of the child, but rather about identifying a harm threshold below which parental decisions will not be tolerated. 49 To my mind, the negative part is right: the actual standards used do not refer to what is literally in the best interests of the child. However, it does not follow that our actual threshold for intervention is harmbased either. In fact, only one of the passages that Diekema quotes (that from Deville and Kopelman) explicitly mentions harm. Others simply refer to what standards are minimal, tolerable, or adequate. I see no reason to equate these notions with avoiding harm.
The problem with defining standards for intervention in terms of harm is that the term 'harm' is notoriously slippery. If there is disagreement over whether to intervene in some particular case, such as with parents who refuse to vaccinate their children, then it is likely that there will be consequent disagreement over whether this is harmful. 50 The point here is not simply that the notion of harm is indeterminate, but whether it is the right standard in the first place.
The indeterminacy of harm has already been noted by several critics, 51 but Diekema acknowledges this himself. 52 I do not think this is a particular problem for the harm threshold. Whether our preferred criterion for intervention is the child's literal best interests, the weaker 'best interests' standard commonly used in practice, or harm, we will need an account of what interests children have and further work to apply it to the particular cases at hand. 53 These are difficult interpretive tasks and there is no reason to expect ready agreement. However, best interests are similarly indeterminate. 43 Bester, op. cit. note 25, pp. 13-14, is right that Mill took parent-child relations to be 'other-regarding', but wrong to suggest these obligations go beyond harm avoidance.
Mill, op. cit. note 37, does suggest that a man can be compelled to support his children (p. 295), but he considers the failure to discharge one's duties a harm (pp. 225, 281). Therefore, this is not an exception to his harm principle, but an application of it. See also  In fact, pace Bester-who argues that 'concerns about indeterminacy and value imposition [are] just as applicable to the harm principle as the BIS [best interests standard]' 54 -and Birchleywho claims that 'harm is at least as problematic as best interests in this regard' 55 -harm must be no more indeterminate, and probably less, than best interests. Bester, following Angus Dawson, argues that the harm principle (or threshold) is superfluous, because any determination of a child's best interests requires us to consider harms, as well as benefits. 56 However, if the notion of best interests incorporates consideration of harms, then it must therefore inherit any indeterminacy inherent in the notion of harm.
If determinations of best interests also incorporate further elements-benefits-that are themselves indeterminate, then determinations of best interests are, if anything, even more indeterminate than determinations of harm. It might be responded that there is no additional indeterminacy around benefits, since they are simply the opposite of harms. This is not obviously the case.
There is widespread agreement that certain things-such as pain, frustration, or degradation-are harms and so avoidance of those things is indeed a benefit. But benefits also cover things that are positively good and there is also much disagreement over what constitutes a good life.
Some have suggested that the apparent diversity of goods can be reduced if we adopt a sufficiently thin or abstract notion of good.
For example, Finnis offers a list of seven 'basic goods': life, knowledge, play, aesthetic experience, sociability, practical reasonableness, and religion. 57 He suggests that: … other objectives and forms of good will be found, on analysis, to be ways or combinations of ways of pursuing (not always sensibly) and realizing (not always successfully) one of the seven basic forms of good, or some combination of them. 58 But, even if it is true that all objectives can be reduced to these seven headings, these still leave open many ways that we can be benefited, not only because there is no objective hierarchy amongst these different basic goods, 59 but also because each of these goods is sufficiently indeterminate that it can be pursued in many different ways-as knowledge is served not only by science and philosophy, but also detective stories and gossip. 60 Thus, while harm may be indeterminate, there is also considerable indeterminacy around the notion of positive benefits. Further, note that even if there was no indeterminacy whatsoever about benefits, the best interests standard is still no less indeterminate than the harm threshold since, as noted above, it also considers harms. We have one standard (the harm threshold) that considers only harms and another (the best interests standard) that considers harms and benefits. So, however indeterminate we think harms and benefits are, the latter standard must be at least as indeterminate as the former. 61 Thus, the above remarks suggesting that the harm threshold is at least as problematic as the best interests standard in this respect are puzzling. In fact, the harm threshold cannot be more indeterminate than best interests since, while both are affected by the indeterminacy of harm, the best interests principle (but not the harm threshold) incorporates additional sources of indeterminacy. 62 To be sure, neither standard will be straightforward to apply. To the extent that both are indeterminate, either may be misused. My point here, however, is merely a comparative one. Concerns about indeterminacy cannot give us any reason to favour the best interests standard over the harm threshold. 63 At worst, the two are equally situated (if anything, the harm threshold is less indeterminate). There are cases where parental decisions can reasonably be seen as harmful, because for instance they make the child in question worse off, but do not license interference, because the child is still all 54 61 The indeterminacy of H+B must be at least as indeterminate as H alone. To think otherwise seems to exhibit a conjunction fallacy similar to those in Tversky, A., & Kahneman, D. (1983). Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychological Review,90,(4), Birchley, op. cit. note 50, p. 33, suggests that indeterminacy is less problematic for an aspirational standard than a minimal one. This is probably true, since aspirational standards are not legally enforced, but here we are considering minimal standards. So the (potentially greater) indeterminacy of best interests is a problem when it is used in this context, even if it can be welcome for other uses. 63 Of course, they might give us reason to reject both of these standards and seek some third alternative. Perhaps this is Birchley's point, given the suggestion in Birchley et al.,op. cit. note 7,p. 934, that a new standard may be needed. However, while I do offer a third alternative here, I do not claim that it is any less indeterminate. I am not convinced that any standard can avoid indeterminacy here. things considered very well off. 65 Parents might, for instance, have to relocate for a variety of reasons, such as work or other family commitments. This may cause disruption to children, who have to change schools, leave behind friends, etc. It is plausible that, as a result of such relocation, the child will be worse off than they were before and worse off than they would have been. Thus, on either baseline, the child is harmed. Further, depending on the particulars, this harm might be significant. However, it does not seem that this is reason to interfere, provided that the child on the whole still enjoys a good life.

| WHY HARM IS THE WRONG THRE S HOLD
Conversely, intervention may be justified where children suffer poor quality of care, even if we would not consider this to be a harm. This is harder to illustrate, because we may be inclined to think that parents are harming their children whenever they fail to provide them with adequate standards of care. Nonetheless, one might think that parents have duties not only to refrain from harming their children but also to benefit them. 66 On this account, parents who make their children neither better nor worse off, though not actively causing harm, are still falling short of what they ought to do. Again, one might construe this as harm, given that the children are worse off than the normative baseline. 67 However, though the notion of harm is commonly moralized in this way, it need not be. One could coherently think that (i) there is no harm done here, but (ii) there is a parental failing sufficient to warrant intervention. 68 One might respond to these counterexamples by suggesting that we could simply define harm in terms of what might be interfered with. Indeed, we often characterize something as harmful because we think it wrong, rather than vice versa. 69 For example, we may think that parents harm their children by not feeding them adequately, but do not think third parties harm those children, even though they (the third parties) are also not feeding them (the children). Consequently, we might want to describe parental choices that license interference as harmful, and those that do not as non-harmful.
However, this proposal cannot rescue Diekema's threshold. If something's being harmful is a consequence of intervention being justifiable, then it cannot (without circularity) also explain why that intervention is justifiable. If harm is to do the explanatory work that Diekema requires, it must be something that can be identified independently of any judgement about the appropriateness of intervention. And if these two things (i.e., what is harmful and what justifies intervention) are different, there is no reason to suppose that they coincide.
None of this precludes the possibility that, in some particular case, the threshold for intervention will happen to coincide with the threshold of harmfulness. However, if these two thresholds do coincide, that is merely coincidence. The threshold that determines when we may justifiably intervene might be either higher or lower than the threshold that distinguishes harm from non-harm. Further, we may also have pragmatic reasons to avoid referring to harm here, since this may be taken as pejorative. 11 However, rejecting the harm threshold does not commit us to accepting best interests. This would follow only if these were the only two alternatives, but this is a false dichotomy. The next section outlines a third alternative.

| A B E T TER ALTERNATIVE TO B E S T INTERE S TS
Diekema rightly notices that our actual threshold for intervention is not determined by the best interests of the child, taken literally. 70 In practice, we do not intervene whenever parental decisions are suboptimal, but only where they are bad enough to fall below some threshold of adequacy or acceptability. However, he is wrong to identify this adequacy threshold with the threshold for harm (or significant harm). The problem here is not the indeterminacy of harm, since any other standard-including 'best interests'-will also need careful interpretive work to apply it to a given case at hand. Rather, the issue is that there is no reason to assume that those cases where intervention is justified will coincide with those involving harm (unless these two expressions are assumed to be equivalent by stipulation, in which case harm ceases to do explanatory work).
After rejecting literal best interests, as demanding too much, the obvious alternative is to focus on a 'good enough' standard. This standard is based on what is sufficient, rather than what is optimal, but that is what we ought to expect if we are seeking a threshold below which interference is justifiable. Of course, we might hopeand even expect-that parents will do more than this. Thus, maximally promoting the child's interests might have some role to play as a guiding ideal for parents (and for others, when overriding parental decisions). 71 It is not, however, necessary in order to avoid interference.
We are justified in interfering only where parental decisions fall so far short of the ideal as to be bad, rather than merely suboptimal. 65 Diekema, op. cit. note 6, p. 131, notes that 'Not all harms should trigger state intervention', suggesting that intervention can only be warranted in cases where there is significant risk of serious harm. Thus, the threshold he endorses is not simply harm, but serious harm. Simply calling it the harm threshold may also be misleading, since it may suggest that any harm justifies intervention.  67 Even without a normative baseline, Feit, op. cit. note 42, argues that some failures to benefit constitute harm, since the could-be beneficiary is worse off than they would have been if they were benefited. 68 Taylor, op. cit. note 25, p. 507, makes an analogous argument regarding non-serious harms. 69 Bradley, op. cit. note 52, p. 410. Brummett, op. cit. note 52, adds that judgements of harm are not only value-laden but also sometimes depend on metaphysical assumptions.
In more Rawlsian language, they are grounded in comprehensive views of the good (MacDougall, op. cit. note 2, p. 292). However, Coggon, op. cit. note 9, p. 225, observes that 'best interests includes not only straightforward welfare issues, but also arguable metaphysical interests' too. 70 Diekema, op. cit. note 1, p. 249. 71 As suggested by Diekema, op. cit. note 6; though, again, it may be reasonable to choose less risky options rather than those that are expected to be better but have high uncertainty (Holm and Edgar, op. cit. note 10, p. 199).
This 'sufficiency threshold' raises many questions that cannot be settled here. For instance, one might ask whether the good enough threshold is an absolute standard, presumably the same for all, or whether what is good enough is relative to the best available in a given case. These are important issues that any account of when parenting is good enough must address. 72 I do not propose to tackle these issues here. My positive contribution to the debate is that the appropriate standards for parental decisions need not coincide with harm avoidance. We might demand more than this of parents, while still not requiring the best, and/or we might regard certain harms as tolerable, provided sufficient standards are maintained overall.
Thus, rejecting the 'best interests' standard should lead us to search for appropriate standards of adequate-or 'good enough'-parenting, without any supposition that these standards coincide with harm avoidance.

ACK N OWLED G M ENTS
I thank two anonymous referees and Clancy Pegg for their helpful comments.

CO N FLI C T O F I NTE R E S T
The author declares no conflict of interest.