Let’s return to those second-order pairings – power :: liability and immunity :: disability (or if you prefer, to those second-order opposites: power/disability and liability/immunity).
1. So then I took my turn
Consider the criminal law: I have a duty to obey the law; we can suppose that this is correlative to a privilege held by the state, or an individual who holds an office enabling him or her to represent the state, e.g. my friendly local policeman, who I’ll call PC Yellow (for reasons which will become clear later). Now, what can Yellow and I do with this duty/privilege pairing – or rather, what can’t we do with it? The important thing that Yellow can’t do, I think, is waive it. Enforcement of the law can be selective and discretionary – you could say that the texture of the law is open enough for enforcement always to be discretionary to some extent. But it’s not open to Yellow to state that, as far as he has anything to say about it, I personally am free of any duty to obey the law; at least, if he does say that, it’s likely to cause legal trouble for him.
So I have a duty to obey the law, correlated to a privilege in the performance of that duty held by PC Yellow, and Yellow is unable to waive that privilege; in other words, Yellow has a disability of waiver, correlating to an immunity to waiver on my part. ‘Immunity’ may seem like an odd term in this context, but what we’re really talking about is one person altering another person’s legal standing: as a citizen subject to the rule of law I’m ‘immune’ to Yellow placing me above the law, but by the same token I’m immune to being placed below the law, subjected to arbitrary impositions and controls.
This is an example of the ‘second-order’ quality of powers (and disabilities) – the fact that they have effect on other jural relations. Confining myself for the time being to the power of waiver, any holder of a privilege (correlated with a duty) may have the power of waiver over the duty; more to the point, if the holder of the privilege doesn’t have a power of waiver, he or she must necessarily have a disability of waiver. A power of waiver is correlated with a liability to waiver on the part of the duty-holder; a disability of waiver is correlated with an immunity to waiver. Liberties as well as duties may be waived: the holder of a no-right (a lack of entitlement to constrain another’s actions in a particular area) may also have the power of waiver over the correlated liberty – and if the holder of the no-right does not have a power of waiver, he or she will necessarily have a disability of waiver.
At this point it gets (more) complicated. Duties and liberties both represent ways in which one person’s actions are subject to another’s control – or delimited lack of control; but the same can be said of liability and immunity, given that liability by definition involves the potential imposition of a duty. It follows that liabilities and immunities can also be waived – which is to say that they are logically associated, in any given case, with either power of waiver or a disability of waiver.
We can see where this logic leads if we return to our criminal law example. So far we have one duty (to obey the law) and one privilege, plus one disability (Yellow’s incapacity to place me above the law) and one immunity. Now, can I waive my immunity? In general terms, somebody who is immune to prosecution (for instance) may well have the power of waiving that immunity. Can I, in this case, waive the immunity to being placed above (or beneath) the law? We’ll assume that it’s an unwaivable immunity – I can’t opt to be above the law even if I’d like to be. In this case, I hold a disability of waiver of immunity to waiver, which correlates with an immunity held by Yellow. Yellow in turn is unable to waive his immunity, giving him a disability which correlates with an immunity on my part – this third immunity being an immunity to waiver of immunity to waiver of immunity to waiver of duty. I am unable to waive this immunity, which means… but enough already; you get the gist.
Hillel Steiner, in his contribution to A Debate over Rights considers the criminal law in Hohfeldian terms and rapidly heads off in a different direction:
Like ordinary citizens, subordinate state officials are standardly disabled from waiving compliance with criminal law duties. Thus Yellow, a subordinate state official, holds a disability to waive a person’s duty not to rob. Yellow’s superior, let’s call her Black, therefore holds an immunity against Yellow’s doing so. Can Black waive her own immunity? What would be implied in denying her the power to do so? For Black’s immunity to be an unwaivable one she, in turn, would have to be encumbered with a disability: namely, the disability to waive Yellow’s disability. But if Black does hold such a disability then some still more superior official, call him Green, must hold an immunity correlative to Black’s disability.
We could, I suppose, continue indefinitely adding such epicycles to this line of reasoning by imagining that Green’s immunity too is unwaivable and identifying yet another even more superior official, Orange, who in turn holds the immunity correlative to Green’s thereby entailed disability. And so on. Let’s not do that. For the sufficiently unmistakable point here is that wherever we decide to stop this otherwise infinite regress, it can be stopped only by an immunity which is waivable. Unwaivable immunities (eventually!) entail waivable ones. So, yes, there can be unwaivable immunities. But what there can’t be are unwaivable immunities without there also being a waivable one. And the waiving of that one renders waivable whatever (otherwise unwaivable) immunity entails it.
A similar passage in Steiner’s 1994 book An Essay on Rights is discussed in Nigel Simmonds’s 1995 paper “The Analytical Foundations of Justice”; the argument reappears more or less unchanged in A Debate over Rights, albeit with a long footnote in response to Simmonds.
There are three things to say about Steiner’s argument here. One is that an Austinian command model of law seems to be creeping in here, with antinomies in the law resolved by reference upward. The thinking here seems to be that official A’s unwaivable subjection to the law is a disability held by official B, who in turn is bound by the effects of a disability held by official C, and so on up the chain until we reach Permanent Secretary Z, whose superior is the sovereign; the latter holds a position above the law, which enables him or her to waive Perm Sec Z’s disability, enabling Z in turn to set the underlings free. Some such model can explain how the rule of law is compatible with change in the law. Interestingly, you can turn the whole model upside down without much loss of explanatory power: official A is above the law relative to you and me, but holds a disability making him or her liable to the law relative to official B, who in turn is above the law relative to A but not to C… until we meet Perm Sec Z, directly subject to the sovereign, who in turn is subject to nobody but empowered to make the law. In this case we would have explained how official freedom of action is compatible with the rule of law. But I think we’re dealing in fables either way, and (more importantly) fables based on a very limited model of the law.
The second point to make – and one that’s made very clearly in Simmonds’s paper – is that Steiner is at best departing from Hohfeld. If we follow Hohfeld, there’s no reason to bring Yellow’s superior into the picture. Yellow’s disability vis-à-vis you or me does not correlate with an immunity held by his or her superior; it correlates, precisely, with an immunity held by you or me. The regress is not vertical but spiral: it consists not of Yellow referring his/her immunity up a Kafkaesque chain of superiors, but of me and Yellow running up an infinite pitch while passing the immunity ball back and forth between ourselves. There’s also something odd – and un-Hohfeldian – in Steiner’s apparent belief that the infinite regress could be stopped with a waivable immunity, i.e. by substituting a power for a disability. In Hohfeldian terms (as Simmonds points out) this would make no difference at all: if you did have the power to waive your immunity, this would correlate to a liability held by Yellow, who would in turn either be able or unable to waive that liability, and off we would go again.
On the other hand (and thirdly), I do think Steiner has identified a genuine problem. I’ll discuss it in the next section.
2. Enough! No more.
If we use Hohfeld’s model, the reasonably plain-language term “unwaivable” apparently can’t be defined without presuming an immunity to waiver of immunity to waiver of immunity to waiver, to say nothing of an immunity to waiver of that immunity, an immunity to waiver of that one, and so indefinitely on. Simmonds talks of these higher-order immunities and disabilities being ‘generated’ through inquiry, which I think is a useful way of looking at it; as if to say, the question of the waivability of the immunity to waive (etc) only arises once you ask it, but once asked it has to be answered. One can imagine MacCruiskeen in the Third Policeman being an expert in this field:
“Ah now. You’ll be talking about the immunity to waiver of the immunity of waiver.”
I supposed that I was. The policeman gave me a look of indescribable craftiness.
“That’s the cleverness of it, you see? I’m talking about the immunity to waiver of the immunity of waiver of the immunity of waiver. And I know what you’re wondering. Can that immunity be waived of its own self?”
I said nothing. The policeman’s ingenuity was rapidly ceasing to be a thing of fascination and becoming one of horror. MacCruiskeen caught my eye and – ye Gods! – winked.
“It cannot, and that’s the truth. There exists an immunity of waiver of the immunity of waiver of the immunity of waiver of the immunity of waiver. What do you think of that now?”
I agree with Steiner in finding this line of thinking troubling. Apart from anything else, it makes me wonder what would happen if I somehow acquired a liability to waiver of immunity to waiver of immunity to waiver of immunity to waiver of immunity to waiver of duty, as it were by accident – would that liability ripple down the chain, leaving me liable (under certain conditions) to waiver of duty? And then, how could the acquisition of some such nth-level liability be ruled out? In his footnote Steiner argues that “any form of infinite regress … cannot be part of anything describable as a normative (much less legal) system; there are necessarily insufficient persons and/or time to sustain it” (emphasis in original); the ‘spiral regress’ proposed by Simmonds resembles “a game whose rules include a stipulation that, at the end of any round, either player is entitled to demand a further round” – an instruction set so open to being prolonged that (pace Wittgenstein) it would be difficult to describe it as a game.
The ‘spiral regress’ thus raises two inter-related problems. On one hand, in specifying a second-order relation – a liability or immunity – it seems as if we can never stop. A liability which can be waived is a liability associated with a power of waiver, correlated with a liability to waiver; this second liability in turn may be waivable, in which case it in turn will be correlated with a second power of waiver… and so on. Different possibilities seem to open up at every stage, and the stages can multiply indefinitely. On the other hand, when interpreting second-order relations – working, as it were, from the outside in – the appearance of multiple, ramifying possibilities seems to collapse. To say that I can waive my immunity to waiver of an immunity to waiver of duty is to say that, in some circumstances, I am liable to waiver of an immunity to waiver of duty – which in turn equates to saying that I may be liable to waiver of duty. Depending on your standpoint, the multiple levels of secondary jural relations seem either to need specifying to infinite precision or to be logically equivalent – in which case they would not need to be specified at all.
As with Dworkin’s right not to be lied to, I think there are a number of possible solutions to this puzzle. One is what you might call the “and no returns” approach. This would see the immunity of waiver I enjoy relative to PC Yellow and the criminal law elaborated into a general immunity, encompassing that immunity and all derivable immunities: as if to say, I have an immunity relative to you in the area of waiving the duty of obeying the criminal law, I have a composite disability relative to you in the area of waiver of this immunity and in the area of waiver of any higher-order immunities deriving from it, and I have a further immunity relative to you as concerns the waiver of the composite immunity correlating to that composite disability. This is a single sentence, but otherwise it’s not much of an improvement: we haven’t succeeded in parcelling up all those higher-order immunities into a single over-arching immunity. And, even if we had done, the Hohfeldian question would still be lurking: this immunity – can it be waived or not?
A more fruitful approach, I think, would be to say that, while it’s always possible to inquire about the powers or disabilities associated with a particular duty, liberty, liability or immunity – and once asked the question can always be answered – it’s not generally necessary to make the inquiry. An infinite (spiral) regress is always possible, but it only comes into being when you start to explore it. And – importantly – traversing the spiral regress generated by considering powers of waiver is something to be done in the real world, under specifiable conditions, not as an abstract exercise. This “real world” stipulation, I think, wards off both the mise en abîme feared by Steiner and the risk of the spiral collapsing into undifferentiated logical equivalence. Here’s an illustration. Let us say that an eccentric relative leaves me a small annuity in his will, on the condition that I visit his grave every May Day. The duty, correlating to a privilege held by Uncle Albert’s executor, is not waivable; if there is no visit, the executor will not pay out. This disability correlates to an immunity on my part; my duty cannot be affected by any variation of the terms of the will by the executor. I am not able to waive this immunity; I have a disability of waiver, correlating to an immunity to waiver held by the executor. In other words, I cannot agree to any variation of the terms of the will which the executor puts forward, and if I offer to agree any such variation the executor may not entertain the offer – he or she is immune to the suggestion. Further, the executor is powerless to waive this immunity, and this disability correlates to a further immunity on my part: it is not open to the executor to propose that henceforth, under certain circumstances, suggestions of agreement to possible variations in the will’s terms will be entertained, nor is it open to me to take any notice of such a proposal.
Let’s suppose, then, that I acquire the power to waive this last immunity, and the executor’s correlative disability with it. In this situation I would let the executor know that, under certain circumstances (which I would specify), I would endorse the proposal that suggestions of agreement to possible variations in the will’s terms would be entertained. What happens now? If my earlier intuition were correct, and a power of waiver would simply propagate back down a chain of immunities and disabilities, I could proceed fairly directly to asking the executor if I could cut out this year’s May Day observance and take the money anyway. This clearly isn’t the case: my willingness to endorse the above proposal (under specified conditions) creates the conditions for the proposal to be made, but doesn’t generate it; that’s up to the executor. If he or she wishes to make such a proposal, and if my stated conditions are congenial, my waiver of my immunity makes it possible for the executor in turn to waive his or her immunity and offer to accept my agreement to possible future variations of the terms of the will, should I give it. However, the waiver of my immunity does not make the waiver of the executor’s immunity necessary – and if the executor does in fact decide to waive his or her immunity, this can be done with a whole new set of strings attached. If both sets of conditions are satisfied, and if I wish to do so, I can then agree to any variation of the terms of the will which the executor puts forward – if he or she decides to do so, and if any new conditions attached to this operation are met.
I think that working through this example demonstrates that both the fear of an infinite regress and the fear of collapse into logical equivalence are overstated. It’s true that the spiral of correlative immunities and disabilities (or liabilities and powers) can always be given one more twist: in fact, to say that I can waive my immunity implies that I must have a power to waive my immunity to the waiver of an immunity to the waiver of an immunity to waiver of duty (and to say that I cannot would imply a disability, and so on). But once real world considerations are imported diminishing returns start to set in fairly quickly. It makes fairly good intuitive sense to talk about not being able to waive my immunity to any variation of my duty; it’s less obvious what a waiver of immunity to proposals that suggestions of agreement to possible variations in my duty might in future be entertained would look like, or when we might need one. The same logic applies when you look at the spiral from the outside in. While further twists of the spiral can always be generated, higher-order powers and immunities are always in a sense parasitic on lower-order ones, and can’t determine them: waiving a higher-order immunity may make it possible to waive the next one down, but does not make it necessary or likely. The spiral regress is populated by human actors with their own interests and bounded freedom of action; legal powers and disabilities constrain their actions but do not determine them.
3. Just step sideways
This is satisfactory as far as it goes, but I don’t think it gets to the heart of Steiner’s worry about Simmonds’ formulation. In my example, a stack of immunities and disabilities followed by a single power of waiver would not lead to the waiver of the first immunity automatically, regularly or (in practice) very often at all; in practice, it would be of very little moment whether the fifth or sixth twist of the spiral was populated by a disability or a power, given the extreme unlikelihood of any fifth- or sixth-level power of waiver actually enabling a first-level waiver. But this result – the waiver propagating back up the spiral – would be possible; the original immunity would not be unwaivable. To formulate an immunity which literally could not be waived, one would need to follow the spiral regress, essentially, to infinity: stopping at the 5th or 10th or 100th iteration would leave open the possibility that the waiver of an n-times-parasitic immunity would propagate all the way back up to the immunity which we originally wanted to protect.
I can see two possible approaches to solving this problem. One would be to appeal to the “real world” approach and dismiss the question as badly-framed. Hohfeld’s jural relations have their own logical precision and purity – the argument would run – but they are jural relations first and last, abstractions moulded to the proportions of real-world problems. A 100th iteration of immunity/disability ball-passing is unimaginable in a real situation; as Steiner says, we just haven’t got the time (although we have got the people – it only takes two). However, what this implies – contra Steiner – is that to treat the spiral regress as a mechanism capable of generating a hundred or more iterations is to get it wrong. An unwaivable immunity, on this logic, is an immunity which is effectively unwaivable: as if to say, “we agree that the next question will be answered in the negative, for as long as either of us has the motivation to ask the next question”. What this formulation brings out is how firmly Hohfeld’s relations are rooted in the model of relations between two people, and specifically agreements between two people. An unwaivable immunity is fundamentally an agreement; as such it cannot be enforced (“I demand that you make this immunity unwaivable by joining me in answering the next question in the negative”) unless it has first been agreed – in which case what is being enforced is not an agreement but observance of a prior agreement.
Another possible approach takes us back to the hierarchy of officials Steiner envisaged as an alternative to an infinite regress. Infinite regress is a besetting problem for theories of the law. Where, after all, do laws come from? Plainly, laws are made by authorities legally endowed with the power to make law. But how did this power arise? It must have been created by an act of law-making; this itself must have been carried out by some higher authority, itself endowed with the power to make law… and so implausibly on.
One way to avoid this infinite regress is to declare the regress to be finite, essentially capping it off at a fixed point. The command model enables us to cut the knot fairly crudely, simply declaring that the state – or the sovereign – is the final source of the law’s legitimacy and hence ultimately takes precedence. Hans Kelsen’s theory of the Grundnorm (‘basic norm’) can be seen as a similar manoeuvre on a more theoretical level. To quote a paper by Neil Duxbury (which, on a personal note, was the first work of legal theory I ever read):
Every legal norm ‘must be created by way of a special act … not of intellect but of will’ – the will of not just anybody, but of a person or body legally authorized to create the legal norm. That authority is itself conferred on that person or body by another legal norm .. which must itself be created by way of an act of will issuing from a person or body whose law-creating capacity is authorized by yet another legal norm. And so on, until we reach the basic norm. Whereas we can explain the reason for the validity of any legal norm by saying that it is attributable to the will of a person or body whose action is authorized by another legal norm, this explanation cannot be applied to the basic norm. The basic norm is not an enacted norm. ‘It must be presupposed,’ Kelsen elaborated in 1960, ‘because it cannot be “posited,” that is to say: created, by an authority whose competence would have to rest on a still higher norm. This final norm’s validity cannot be derived from a higher norm, the reason for its validity cannot be questioned.’ Because it is not an enacted norm, moreover, it ‘cannot be the meaning of an act of will’; rather, ‘it can only be the meaning of an act of thinking’ – the consequence of ‘presuppos[ing] in our juristic thinking the norm: “One ought to obey the prescriptions of the historically first constitution.”’
We can safely say that this is not entirely satisfactory, since Kelsen himself ultimately abandoned this line of thought (or, Duxbury argues, subverted it by developing tendencies within it); his final conclusion was that the basic norm should be thought of, not as a norm attributable to an act of thinking, but as a fictional norm attributable to the will of a fictional authority. This is a more subtle and interesting point than it looks – particularly when we take into account that, at least some of the time, Kelsen used ‘fictional’ to denote that something not only did not but could not exist, owing to internal contradictions – but I won’t investigate it here.
My current point is that both Kelsen’s basic norm and Austin’s sovereign – considered as capstones topping off an otherwise infinite regress – are arbitrary and unsatisfactory solutions, but solutions to a genuine problem. The problem is not, in Steiner’s terms, “a game whose rules include a stipulation that, at the end of any round, either player is entitled to demand a further round” – as we’ve seen in the context of Uncle Albert’s will, in practice this isn’t likely to cause any difficulties. The problem – both for the legitimacy of legal authorities, and for Simmonds’s unwaivable immunities – is a game in which, at the end of every round, the player must ask for another round. In both cases the question is unanswered at the end of each round, and it’s a question that needs an answer – whether it’s where the legitimacy of law-making authorities comes from or whether an immunity genuinely cannot be waived.
For the first of these cases of infinite regress, a much more satisfactory alternative is offered by Hart’s rule of recognition. Hart’s deceptively simple proposition is that any legal system includes a criterion by which laws can be recognised as ‘legal’, and which is acknowledged and upheld by the practices of officials within the system. Instead of referring upwards to a higher authority (itself dependent on a still higher authority), this approach effectively refers sideways. The question posed is not whether an enactment derived from a legitimate authority, but whether the authority in question was engaging in what was recognised as the activity of law-making within that legal system, including observation of the rules and criteria applicable within that system. The regress stops after a single step; the question of whether, for example, the constitution of the present law-making authority took place in accordance with the criteria then applicable is of purely historical interest – unless that question forms part of the criteria to be applied within the current system, in which case it will in effect already have been asked.
Can the infinite spiral regress associated with unwaivable immunities be dealt with similarly? I think perhaps it can. I suggested above that an effectively unwaivable immunity – as distinct from an immunity which is unwaivable by definition – could be modelled as an agreement that the immunity should be treated as unwaivable, generating a disability of waiver whose correlative immunity was in turn treated as unwaivable, and so on: “we agree that the next question will be answered in the negative, for as long as either of us has the motivation to ask the next question”. This is a “let’s not go there” model of unwaivable immunity, essentially. Perhaps all that is needed to formalise this practice – and bridge the gap between ‘effectively unwaivable’ and ‘unwaivable by definition’ – is a generally recognised rule, and a practice of classification through which it can be determined whether the rule applies. In other words, perhaps when we say that an immunity is unwaivable we are not saying that the derived nth-level immunity to waiver carries a disability of waiver correlative to an n+1th-level immunity, and so on; perhaps we are saying that we can rely on this immunity being treated as unwaivable (by the “let’s not go there” method), because we know that it falls into the class of immunities which we have an established and publicly recognised practice of treating as unwaivable. The infinite regress doesn’t evaporate quite as dramatically as in the previous example – it’s still meaningful to say that I have no power to waive immunity to changes in my standing relative to the criminal law, for instance, and to ask what such a power might look like. Navigating the spiral regress ceases to be necessary, though, which is the desired effect.
Does this class of immunities to be treated as unwaivable, or this practice of recognising immunities as unwaivable, correspond to anything in the real world? Fortunately for me (and for your patience), I think it does. One way of modelling the difference between the criminal law and most (all?) other branches of the law is, I think, precisely the unwaivable immunity with which we started: the immunity to being placed above (or below) the law. In other areas of the law – areas which approximate more closely to Hohfeld’s model of a two-person agreement – it is an open question whether a duty can be waived and (if not) whether the immunity correlative to the disability of waiver can itself be waived. In the criminal law the answer to both questions can only be No. This is one aspect of the uniformity of the criminal law, which can be considered one of its defining virtues: equal subjection to the criminal law can be seen as a civil right, a key element of citizenship.
This is speculative and fairly hasty stuff, but I think it’s worth thinking about. To recapitulate, if we did resolve the spiral regress in the way I’m suggesting, it would mean that there was at least one recognisable area of legal practice which operated on the basis of duties and derived immunities being unwaivable – and did so without inquiring too deeply into the waivability of higher-level immunities, as the answer could be assumed to be negative. The criminal law seems like a good candidate.