Law and Disagreement
Waldron
Chapter Six – Legislators’ Intentions and Unintentional Legislation
1. LEGISLATIVE INTENT
A bill has been debated and a statute has been enacted. How now is it to be interpreted ? One initially appealing answer is that we should interpret the statute in the way the legislators intended, resolving any vagueness or ambiguity by finding out as much as we can about what the legislators had in mind. In this chapter, I shall discuss the latest attempt in analytic jurisprudence to revive the view that reference to legislators’ intentions should playa role in the interpretation of statutes. I use ‘revive’ advisedly. It is true that reference to legislative intent is reasonably common among judges and lawyers in America, and the appeal to ‘original intent’ is common too in the politics of American constitutional law. Philosophically, however, the idea of appealing beyond the statutory text to independent evidence of what particular legislators are thought to have intended has been subject to such powerful criticisms, most notably by Ronald Dworkin, that one is surprised to find it appearing again in anything other than a trivial form in respectable academic jurisprudence.
Yet in a recent book, entitled Interpretation and Legal Theory, Andrei Marmor makes an argument for the intentionalist thesis which is not only respectable, but novel and almost persuasive. Its originality consists in eschewing any argument based on democratic considerations. Marmor advances no version of the claim that judges ought to defer to legislators’ intentions because legislators are elected and acountable and judges are
120
not. His argument is based instead on a consideration of Joseph Raz’s theory of the authority” of law. The normal justification of X’S having authority over Y, Raz says, is that Y is more likely to follow the reasons that apply objectivity to his decisions if he follows the directives of X, than if he tries to work out the reasons himself. If this is true of X’s directives, Marmor reckons, it is likely to be true also of the intentions that lie behind those directives, even when such intentions are not fully disclosed in the directives themselves and have to be recovered by other means.
The application of this general argument to the case of legislation is obvious enough. A statute is conceived as the work of a lawmaker, X, and it is said to have authority just in case X has some sort of technical or moral expertise in the field that the statute addresses. X is the expert and so ordinary citizens will do better by deferring to his directives than by trusting their own judgement on the matter. Now, this expertise is disclosed in the first instance in the explicit words of the statute that X has written. It is however, unlikely to be exhausted by that. If his text turns out to be vague or infelicitous his subjects may have recourse to his expertise in other ways, by gathering clues as to the thoughts and purposes that were in his mind at the time the original statute was passed. The law’s authority, after all, is a matter of X’s authority, and that authority is defined dispositionally as the likelihood that others will do better by deferring to him. Thus, an argument for the authority of law based on the fact that the legislator is more enlightened than his audience may generate a case for appealing also to legislative intent as a further basis of enlightenment in determining what a citizen or official ought to do.
An argument of this kind depends, at least for its initial attraction, on a rather primitive picture of legislation. It gains its greatest plausibility’ by’ stressing the features that statutes have in common with simple commands. If we understand legislation along Hobbesian lines – ‘where a man saith, Doe not this, or Doe not this, without expecting other reason than the Will of him that says it’ – we may indeed be inclined to turn to the law maker’s intentions as a basis for interpreting the statutes that he has passed. For if there is something about his will that makes his original command authoritative, it seems natural to have recourse to that will – as far as one can ascertain it – if one is in need of further guidance in putting the command into effect.
121
I shall spare the reader any reiteration of the difficulties of this style of jurisprudence. The most important point is that Marmor does not take adequate account of the fact that modern statutes are not the products of single expert authors. They are produced by the deliberations of large multi-member assemblies whose claim to authority in Raz’s sense (if indeed they can make any such claim at all) consists in their ability to integrate a diversity of purposes, interests, and aims among their members into the text of a single legislative product. The modern situation, in other words, is not that of a person having authority, but (at most) of a group having authority, and of its having that authority only by virtue of the way in which it combines the interests and knowledge of its members in the act of legislating. In this situation, it is unwise to make any inference from the authority of legislation to the authority of anything said or any purpose expressed by particular members of the legislature that does not amount in itself to a legislative act. That, in outline, is the argument I shall make.
Someone may ask: What if the conditions under which a statute is passed are such as to make single-author jurisprudence more appropriate?
A particular law may have been conceived by just one legislator and his staff and passed ‘on the nod’, so to speak, by others in the chamber, in deference to his expertise in the area. In a case like this, surely it is not inappropriate to regard the particular legislator as the author, and to appeal to his intentions when we want to clarify what he has done. But whether this line of argument works will depend on whether we can plausibly attribute authority, not just legislative efficacy, to the single legislator in question.
We have to remember that theories of legal authority (and any concomitant theories of interpretation) are usually purchased wholesale not retail. The question for jurisprudence is seldom about the authority or interpretation of this or that statute in particular, but about the relation between authority and interpretation generally. We must ask, then, whether there is anything true in general about the way in which statutes are produced that makes appeal to legislators’ intentions a proper strategy of interpretation. How we answer that question will depend on what we take to be the most helpful general model of the legislator process, so far as theories of authority are concerned. My submission is going to be that, at that level, we will do better by eschewing any model that regards legislation as most commonly’ the intentional product of a single law-making author. We do better – as I have argued throughout this part of the book by focusing on the structures and proceedings of legislative assemblies.
The argument that follows will proceed in three stages. After some preliminaries, I will address the general issue of the intentionality of statutes. I will argue in section 3 that, under the conditions of modem legislation, it is often implausible to describe legislative acts as intentional acts, even though they take place in an intentionally-organized context. In section 4, I will show how Raz’s thesis about authority (or a plausible version of that thesis) might nevertheless apply to legislative acts, despite the fact that they are not conceived as intentional acts. Then, in section 5, I will show that the best arguments for the authority of statutes produced under these conditions are arguments which actually preclude any appeal to the intentions of particular legislators as a general interpretive strategy. I will end in section 6, with some further general comments on the significance of the considerations about legislation on which I have been basing my arguments.
122
2. LEGISLATORS NOT FRAMERS
First, I want to deal with some preliminary matters. Like Marmor, I take the sof this discuto be statutory not constitutional interpretation. The hardest case for an opponent of the intentionalist strategy – and thus the case that I have to address – is a statute passed recently, in a deliberative session of which there is an undisputed public record, by legislators familiar with the conditions of the society to which the statute is going to apply. If questions arise about the interpretation of a law enacted under these conditions, it is not implausible to suggest that they might be answered by looking to the public record of debates, committee deliberations, etc. to see what the legislators had in mind. Of course, such exercises are not always successful. Still, even if the judges get the legislators’ intentions wrong, or even if the legislators are for some other reason dissatisfied with the judges’ answer, the legislators can always enact another statute to impose the interpretation or purpose they prefer.
None of this applies in constitutional adjudication. Despite its popularity among conservative politicians, the appeal to the intentions of the framers of the US Constitution is ludicrously implausible. Those who make the appeal disagree about who counts as a framer. Is it the drafters of the text, the members of the Philadelphia Convention, the ratifiers, or the people who elected the ratifiers? Not only that, but whoever the framers are taken to be, their actual intentions are largely a matter of conjecture and at least as controversial as any less historical form of ‘interpretation’. We know ‘practically nothing’, for example, about what went on in state legislatures during the ratification debates. Moreover, even if we knew the framers’ intentions, it would be unwise to rely on them. The
123
framers lived two centuries ago and, despite their revolutionary virtue, they were utterly unacquainted with the conditions of modern politics and society in America. They lived in a loosely federated set of sparsely populated post-colonial white supremacist states, whose economies were based on the exploitation of African slaves in agriculture, and whose politics were confined to those who owned slaves, women, or property. For the purposes of the Raz/Marmor argument, these men would not even have understood, let alone have had a better grasp than us of, the conditions for the subsistence of a continental superpower as a free and constitutional republic, under conditions of ethnic diversity, democratic equality, and post-industrial crisis. There may be a case for regarding the text of the Constitution as a stable and indispensable framework to house the volatility and vicissitudes of modern politics. But that stabilizing function has more in common with coordinative authority – which Marmor acknowledges can generate no justification for an appeal to original intent – than with the authority of moral expertise on which deference to framers’ intentions is more usually based.
I take it, then, that we are to discuss the role of legislative intent in the interpretation of ordinary statutes under modern conditions. I shall assume, in addition, that when we talk about legislation, we are envisaging the work of a Congress, a Parliament or a state assembly: that is, a body comprising a number of members (usually in the hundreds) of various political persuasions, elected as representatives by the people of the state to whom the legislation is going to apply. I shall assume, too, as I did in Chapter Four, that the legislators are a diverse body of people, drawn from different groups in a heterogeneous and multicultural society. I assume in other words that there is very little in the way of shared cultural and social understandings among them beyond the rather stiff and formal language that they address to one another in their legislative debates. I shall assume finally that the legislature possesses fastidiously defined procedures to determine how a piece of legislation is introduced, debate, and passed, what it comprises (that is what its final authoritative text is), and when it takes effect. These procedures amount to the constitution of the legislature, and reference to them will form part of the rule of recognition for the legal system.
Though these assumptions are obvious enough in the modern world – and no doubt wearying to my readers at this stage! – they turn out to be crucial for my argument against appeals to legislators’ intentions. Accordingly, I will spend some time examining their grounds in section 6.
124
3. UNINTENTIONAL LEGISLATION
It is sometimes said, as though it were a truism, that anything we interpret must necessarily be conceived as a product of someone’s intention. The point is taken to be an elementary one about what counts as a linguistic artefact. Wave action on sand or the rustling of the wind might produce marks or noises that look or sound like texts. But one cannot treat them as texts – it is said – and one cannot set about discerning their meaning, unless one attributes their appearance to the intentions of some intelligent author. Stanley Fish puts the point this way:
[I]n order to hear sense in arbitrarily produced sounds or marks we have to hear those sounds and marks within the assumption that they have been produced by some purposeful agent; that is, we have to hear them as not arbitrarily produced, even if to do so we must attribute purpose and intention to the waves or to the wind or to the great spirit that rolls through all things.
The positing of intention comes first on this account; the positing of meanings second.
In exactly the same way, many legal theorists think it obvious that a statute must be conceived as something which has been produced intentionally; and they infer from this that the only possible object of interpretation in the law is the meaning envisaged by those whose purposes constitute the intentionality of the provision under consideration. Thus, imagine that a piece of paper duly certified in accordance with proper parliamentary procedure appears in front of a judge with marks on it that look like the following:
Vehicles in the Park Act 1993. (1) With the exception of bicycles and ambulances, no vehicle shall be permitted to enter any state or municipal park. (2) Any person who brings a vehicle into a state or municipal park shall be liable to a fine of not more than $100.
According to Stanley Fish and others who take the view that I am discussing in this section, if the judge views this as a statute and begins the laborious business of working out, for example, whether its provisions apply to the proverbial jeep on a plinth, he must be treating the marks on the paper as the product of someone’s intention.
125
Fish concedes that it is a further question whether we should search the legislative record in order to find intentionalist clues to the solution of the jeep-on-a-plinth problem. Acknowledging legislative intentionality, he says, is not the same as deferring to legislators’ intentions, especially when the latter are not the intentions disclosed immediately in the text of the statute itself. We may end up answering our question about the jeep by reading the statute in a way that would have surprised its authors. Still, the crucial point, according to Fish, “ is that one cannot read or reread independently of intention, independently, that is, of the assumption that one is dealing with marks or sounds produced by an intentional being, a being situated in some enterprise in relation to which he has a purpose or a point of view”.
The fact that ‘intention’ and ‘meaning’ are heavily theory-laden terms does not, according to Fish, undermine the point that interpreting a text is always a matter of determining what the intentional being who produced it meant. It is always, he says, a matter of discovering the meaning of which the statutory text was intended to be the expression.
Though this position may seem uncontroversial with regard to novels and poems, it is confused – and significantly confused – in regard to modem legislation. Legislation, I have assumed, is the product of a multimember assembly, comprising a large number of personsof quite radically differing aims, i, and backgrounds. Under these conditions, the specific provisions of a particular statute are often the result of compromise and line-item voting. It is perfectly possible, for example, that our imagined Vehicles in the Park Act, considered as a whole, does not reflect the purposes or intentions of any of the legislators who together enacted it. Let me spell out this possibility. With regard to section 1 of the Act, the legislators might have been divided on the exception for bicycles (which I shall call B), on the exception for ambulances (A), and on the inclusion of state parks as well as municipal parks (S). Suppose the legislators divided into three equal factions on these issues as in Figure 3.
Faction 1 Faction 2 Faction 3
B B B
A ~A A
~S S S
Figure 3.
126
Successive majority voting on these various questions would produce our familiar statute – B & A & S- even though that combination corresponded to nobody’s preference.
Someone may respond: surely, after the amendments are in, the legislators will vote on the bill as a whole, in its final form (B & A & S). Thus even if they initially disagree, the enacted statute will at least reflect the intentions of a majority at that last stage, taking into account their awareness of what was politically possible. But this is purely an artefact of our particular parliamentary procedures – that we have a ‘third reading’ debate and division, for example, after the amendment stage. One could imagine a legislature that proceeded quite differently. There might be a preliminary discussion during which all the issues likely to provoke a division were identified. General debate would ensue, at the conclusion of which members would feed their votes on the various issues into a machine which would produce the statute in its final form on the basis of the voting and promulgate it automatically to judges, officials, and the population at large. The possibility of such a machine was considered by Richard Wollheim in his famous article ‘A Paradox in the Theory of Democracy’:
“I now want to. . . envisage Democracy in terms of a certain machine which operates according to [the following] method or rule. The machine – which we may for convenience call the democratic machine – operates in a discontinuous fashion.
Into it are fed at fixed intervals the choices of the individual citizens. The machine then aggregates them according to the pre-established rule or method, and so comes up with what may be called a ‘choice’ of its own. Democratic rule is said to be achieved if throughout the period when the machine is not working, the most recent choice of the machine is acted upon.”
Now, theorists of public choice are familiar with various tangles of cyclicity that such a machine might get into. Let us assume, however – as Wollheim did – that these are not a problem for the cases we are considering. The point of this thought-experiment (which is not the same as Wollheim’s use of it) is that it enables us to envisage a piece of legislation
127
that cannot be conceived as, in Fish’s words, something produced ‘by an intentional being, . . . situated in some enterprise in relation to which he has a purpose or a point of view’.
Of course, this is just a model: in the real world, statutes are never produced exactly like that. But also, in the real world, statutes are never produced exactly as the product of one person’s coherent intention. The interesting question is which picture is more helpful for our thinking about the intentionality of statutes under modern legislative conditions. Given the large part that is played by compromise, logrolling, and last minute amendments in contemporary legislation, my money is on the machine.
Fish may make another response. He may say: even if it is the product of Wollheim’s machine, still the text of our statute is quite unlike the cases mentioned earlier of marks made by waves in the sand or noises in the wind.
He is right. There are three important differences. (1) The machine does what it does on the basis of the legislators’ inputs, and those inputs are fully intentional. (2) The machine is programmed to produce the particular signs (words) that it does (under particular input conditions) because those signs are taken to have certain meanings by those who program it. It is the programmers’ intention, for example, that the English words ‘with the exception of bicycles’ be produced by the machine in the event that a majority of legislators press button ‘B’. This is their expectation about how the machine’s output – whatever it is – will be understood by those to whom it is sent: they will read it as a set of English sentences. And it is this expectation that the legislators share as they input their votes. (3) The whole system (the machine itself and the conventions surrounding its use) has been set up intentionally, pursuant to certain democratic procedural aspirations. The entire exercise, then, is an intentional one, and that is enough to distinguish it from waves in the sand.
But none of that is sufficient to establish Fish’s conclusion that the text must be taken to embody a particular intention, for example, an intention to
128
ensure that vehicles other than bicycles and ambulances do not enter state or municipal parks. For neither that intention nor anything like it figures among the intentional elements 1-3 that we have just identified. Accordingly, when the statute is read for the first time by an official or a citizen (or for that matter by a legislator interested to know what he and his colleagues ‘decided’), Fish is not entitled to say that the aim of the reading must be to determine what somebody meant.
Max Radin once wrote that it is the job of a legislature to pass statutes not form intentions. A legislature is an artificial actor, and the passage of a statute is its action: indeed we refer to statutes as acts of Congress or Parliament or whatever. But though we use the language of agency in this way, we must not be misled by an obsessive analogy with the actions of natural persons into searching for a legislative equivalent for every event or state associated with action in the psychology of individual agents. Few would say, for example, that legislatures (as opposed to particular legislators) have motives as well as intentions associated with particular acts; few would be willing to put the kind of weight on the motive/intention distinction that we apply in the individual case. So why insist on a correlate for intention at all? Why not simply say that the act of a legislature is an artificial resultant of the acts of individual legislators, structured and related to one another through certain procedures, decision-functions and perhaps machines? Why not say that while each of the latter actions – the individual actions – is of course the product of an intention, the resultant action- the act of the legislature itself – need not be?
Part of the confusion tempting us to ascribe intentions to legislatures has to do with the concept of meaning. It may be thought that the very fact that legislation is a speech-act – the production of a meaningful utterance or text – means that it cannot be regarded as non-intentional. What is it after all to say ‘No vehicle shall be permitted to enter any state or municipal park’ but to utter a string of sounds with the intention to produce a certain effect
129
or response in one’s audience by virtue of their recognition of that very intention? Even if one were to eschew the idea of a single utterer for cases like this and concentrate just on the meaning of the sentence, it remains the case that our most plausible accounts of sentence-meaning make that notion a function (albeit a complex one) of individual intention.
I do not think that considerations like these drive us towards any notion of legislative intentionality. I have noted already – item 2 above – that even a Wollheim machine would have to be programmed and operated on the basis of certain assumptions about tmeanings of various signs and sounds. A legislator who votes for (or against) a provision like ‘No vehicle shall be permitted to enter any state or municipal p’ does so on the assumption that – to put it crudely – what the words mean to him is identical to what they will mean to those to whom they are addressed (in the event that the provision is passed). He can entertain that assumption only on condition that the words are meaningful elements in a language – that is, only on condition that there is a community in which it is well known that members of that community commonly use such words to produce a certain effect or response in their audience by virtue of the audience’s recognition of that very intention. That such assumptions pervade the legislative process shows how much law depends on language, on the shared conventions that constitute a language, and on the reciprocity of intentions that conventions comprise. But though they indicate a place for referring to intentions – the intentions of language-users as such – in any comprehensive account of what is going on, they provide no warrant for the view that, simply because a particular piece of legislation has a linguistic meaning, it must embody a particular intention attributable to a language-user .
4. AUTHORITY
It is one thing to play with the possibility of unintentional legislation, along the lines of my thought-experiment involving the Wollheim machine. But what becomes of the authority of a legislature if its statutes are not conceived as intentional acts? How can a legal provision be regarded as authoritative if the legislature employs a Wollheim machine to generate its laws? To address this question, we have to look again at the notion of authority.
130
In what follows, I shall concentrate on the theory of authority presented by Joseph Raz, for this is not only the leading conception, but the one on which Marmor’s argument is based. It may not, however, fully capture the claim that legislation makes on our respect: I have tried to describe this further and deeper claim in Chapter Five. So far as that deeper claim is concerned, there is no particular difficulty about legislation which is the product of a Wollheim machine. The difficulty lies solely with the Raz-ian conception of authority; so, for the sake of argument, this chapter focuses almost exclusively on Raz’s conception.
The thesis about authority that Marmor invokes from Joseph Raz talks about ‘the normal way to establish that a person has authority over another person’. It is persons who have authority, on this account, not organizations or laws. In a subsequent chapter of the book, Raz asks whether anything changes when we move from a one-on-one situation to an issue involving the authority of the state. However, the change that seems to interest him there is the fact that ‘political authorities govern groups of people’ rather than particular individuals.37 He does not appear to be interested in the ramifications for his thesis of the fact that political authorities like legislatures are groups of people. The closest Raz comes to considering the issue is in the following passage that Marmor cites from an article published earlier than The Morality of Freedom: ‘[A] directive can be authoritatively binding only if it is, or is at least presented as, someone’s view of how its subjects ought to behave.’ Marmor glosses this as ‘the idea that only an agent capable of communion with others can have authority over them’, (He uses it to argue against Ronald Dworkin’s view that the norms judges should take as authoritative are norms they themselves produce by a process of interpretive construction.)
The fallacy in Raz’s thesis (or at least in Marmor’s presentation of Raz’s thesis) about the normal justification of authority lies in forging too tight a connection between authority and authorship. On the Raz/Marmor account, authority cannot be systemic: it must be attributable to an author.
Their claim seems to be that the reasons of ordinary citizens can be affected in the way that authority affects reasons only if there is some identifiable person (natural or artificial) who has the authority in question. Thus Marmor assumes that a statute, S, can have authority only if the following is true about the person, X, who issued it: namely, that any person Y to whom the statute is addressed does better by following X’s view of the reasons that apply to Y’s conduct than by attempting to work out those
131
reasons for himself. For a statute to have authority, on this account, is for its author to have (or to have had) a view which is superior in some regard to any view likely to be formed by the person to whom the statute is addressed.
I want to suggest, however, that the situation in which statutory authority (in Raz’s sense) is traceable back to personal authority is a special case. In other cases, it is possible to talk about the authority of the statute tout court and develop a version of the normal justification thesis Which refers only to the likelihood that person subject to it will do better by following its provisions. What I have in mind is something along the following lines (which I shall label ‘thesis []]I because I will want to refer to it later):
[J]: A statute (or any text) S has authority over a person Y only if person Y is likely better to comply with the reasons which apply to him by following the provisions of S, than if he tries to follow those reasons directly.
Now, [J] may be true on account of something about the views or the intentions of the person X who authored or voted for S. But that need not be the reason for its truth: S may satisfy [J] despite the fact that S has no author, or satisfy it in virtue of features that make no reference to any person’s expertise .
Marmor of course does not consider [J] – for I have just invented it.
However, in his discussion of Dworkin’s arguments, he seems to suggest that whenever anything like [J] is true, we must either locate an individual X, whose view on the matter in question is taken to be superior to Y’s, or we must pretend that there is some such individual. This insistence on producing or imagining a singular X is never defended (except by reference to Raz’s formulations). I suspect, however, that it is due to an inability to
132
conceive of how Y could do better by following the provisions of S than by following his own reason, unless S was an expression of someone else’s view (someone who was smarter or more knowledgeable than Y was). So let me explain how that might be possible.
First, a trivial example. I am very bad at keeping track of credit card transactions. I use my card, several times a week, but I do not keep the receipts or write down the amounts, so I am never sure how much has been charged to my account. Any attempt I made to work it out for myself would probably be inaccurate. Fortunately, Chase Manhattan Bank provides an ‘8001 number for me to call. I punch in my credit card account number, and the last four digits of my Social Security number to verify my identity, and the disembodied ‘voice’ of a machine will say something like this:
Your current balance is FOUR hundred and thirty FIVE dollars and thirty cents. Your available credit is FIVE thousand FIVE hundred and sixty FOUR dollars and forty cents.
Now I am under no illusion that there is anyone at the bank who has worked all this out and is waiting to communicate it to me, any more than I think the ‘voice’ I hear is that of some individual at Chase headquarters, sitting patiently by the phone waiting for people like me to call. The deliverance I have just quoted is nothing but the product of a machine responding on the spot to my touch-tone input, to various inputs from retailers, and to its record of my credit limit. No one has worked out what my balance is: there are just the noises recorded above. Indeed the machine may not even be storing a record of my balance; apart from the sending of my monthly account, it may be computed only as a response to my telephone call. Still, the machine’s output is authoritative for me in terms of thesis [J]:
I do better by relying this text for information about the outstanding balance on my credit card than I would by relying on my own forgetful computations.
I know of course that someone has intentiprogrammed the machine to produce outputs (in response to certain inputs) such that people who treat its sounds as English sentences will have reason to rely on the information such sentences conventionally convey. That is like the conditions 1-3 for the Wollheim machine that I mentioned in the previous section. But I can concede all that without saying (or pretending) that anyone is intentionally conveying to me the message that my outstanding balance is $435.60. That m)r reliance on the machine’s output makes reference to someone’s intentions (in programming the machine) does not make the machine, for me, an intentional system.
133
In the case of the bank’s computer, there are good reasons for me to respect the machine’s output more than I respect the results of my own computations: the bank has an interest in keeping accurate records and no interest, as far as I can tell, in concealing the true state of my account from me. But legislatures are not banks. What reasons might there be for respecting the deliverances of a Wollheim machine?
I will consider three lines of argument: (A) the utilitarian argument, that a Wollheim machine might aggregate individual preferences in the way required by the applicable social welfare function; (B) the Condorcet Jury Theorem, that the view determined by a Wollheim machine to be that of the majority has a probability of being correct that is greater than the probability that anyone of the views being counted is correct; and (C) the Aristotelian claim (which we considered briefly in Chapter Four) that a multitude of persons, by a process of deliberative synthesis, may forge a better view than even the wisest of them could come up with on his own.
None of these arguments is watertight. But in common with Marmor and Raz, I do not want to make a watertight case for the authority of such legislation. The claim by Marmor that is under consideration is only that if (and to the extent that) there is reason to accord authority to a statute, there may also be reason to accord authority to the intentions of some or all of the members of the legislature which passed it, even when these intentions are not disclosed in the text of statute itself. My claim in section 5 will be that even this inference of Marmor’s is invalid, at least to the extent that the authority of a statute (such as it is) is based on argument A, argument B, or argument C.
Before moving to that, let me sketch out the three lines of argument in a little more detail.
A. The Utilitarian Argument. Historically, the utilitarian case for democracy has been that a majoritarian decision-procedure at the level of public choice can sometimes model an aggregative social welfare function at the level of utilitarian morality. For both cases – that is, for both a majoritarian decision-procedure and a Benthamite social welfare function – the fact that a course of action promotes the satisfaction of some preference counts in its favour, and when it becomes apparent that not everyone’s preferences can be satisfied, one adopts the course of action that satisfies as many’ of them as possible. Thus as democrats, we follow the will of the majority; and as utilitarians, we try to promote the greatest aggregate
134
happiness. The two may amount to roughly the same thing if individual votes are a reliable guide to the conditions of individual happiness.
Now of course there are all sorts of difficulties with this equation. It presupposes that individuals are voting their own happiness; it assumes they are reliable judges of that; and it avoids altogether the question of the intensity of their preferences. Moreover, for the case of a representative legislature (as opposed to a directly democratic one), a number of epicycles have to be added to ensure that votes cast by each legislator accurately represent the proportion between various preferences of his constituents.
Still, even if these difficulties cannot be perfectly overcome, it is easy to imagine that a Wollheim machine, registering and aggregating a large number of individuals expressed preferences, might produce an output whose utilitarian credentials were much better than any rough-and-ready computation that could be made by the individual citizen. If an individual knows that his own computations of social utility are even less accurate than the machines, and perhaps biased also by wishful thinking in his own favour, it might be wiser for him to forego self-reliance and rely instead on the somewhat less fallible outputs of the machine as his guide to what the general happiness demands. In these circumstances, anyone who believes that the truth about what ought to be done is utilitarian will have reason to regard the outputs of the machine as authoritative in the sense of thesis [J].
B. Condorcet’s Jury Theorem. Under the previous heading, we assumed that individual legislators were asking themselves different questions as they voted: each was asking ‘What do I want?’ (so that legislator W was asking ‘What does W want?’ and legislator X was asking ‘What does X want?’).
The same “would be true if each legislator was asking himself, ‘What are the preferences of my constituents?’ We imagined that the Wollheim machine would take each person’s preferences and aggregate them using a social welfare function to produce, for the first time in this process, a conclusion about the general good. Often, however, legislators take themselves to be asking the same question: ‘What is best for everyone, or for society at large, or what is most just?’ Often, each takes himself to be expressing a view about the general welfare. When this is the basis of individual legislators’ votes, is there any reason to believe that a Wollheim machine will produce an authoritative output, in the sense of thesis [J]?
There is; and the argument is a matter of simple arithmetic. Suppose that some voters are addressing a single question with two answers, one correct and one incorrect (for example, the)r are jurors deciding whether someone did or did not commit a robbery). Assume that the probability of
135
each voter choosing the correct answer is greater than 0.5, that is, that each is more likely to get it right by thinking about the matter and then voting his best judgement than if he simply tosses a coin. Then provided the votes are cast independently of one another, the probability that the answer chosen by a simple majority of them will be the correct answer is somewhat greater than 0.5, and it rises toward the limit of certainty as the size of the group of people voting increases.
This theorem, due to the Marquis de Condorcet, has its limits. The same reasoning yields the conclusion that if the average competence of the individual voters dips below 0.5, then majority competence tends towards zero as group size increases. In other words, the theorem provides a ground for the authority of the outcome if individual voters are, on average, more likely than not to be correct in the alternative they choose; but it provides reason for doubting the authority of the outcome if on average the voters are more likely than not to choose incorrectly. As we saw in Chapter Three, Condorcet himself believed that, historically, average individual competence tended to decline as group size (for example in a legislature) increased. But of course there is no reason to think that this truth (if it is a truth) of political sociology exactly cancels out the arithmetical effect we have been discussing.
In any case, Condorcets theorem indicates a reason for according authority to the output of a Wollheim machine in at least some circumstances. If a large number of legislators address themselves scrupulously to some objective issue and if they each have a better than even chance of being right, then the ordinary citizen would be wise to trust the decision
136
generated by a majoritarian machine out of their individual votes. Indeed, on Condorcet’s reasoning, each individual will do better by observing the authority of the machine-produced legislation than he would by trusting the reasoning that led to his own individual vote. This, as we shall see, is fatal to ‘s case. But before we consider that, let us look at the third of our arguments for legislative authority.
C. Aristotelian Synthesis. Condorcet’s result is maddeningly mechanical.
Forget truth, objectivity, and justice: the same arithmetical reasoning tells us about the prospects that a majority of balls pulled at random out of an urn will be black, given that more than half of the balls in the urn are black.
Moreover, Condorcet’s result takes no account of discussion, deliberation, and persuasion – the very processes that are likely to produce a legislative record to which Marmor’s intention-seeking judge can appeal. The third line of argument I want to consider concentrates a little less on the Wollheim machine and the arithmetical blindness of its operations, and a little more on the real-world prospects that a multi-member legislature might come up with some result that satisfies condition [J]. Like the previous argument (and unlike the utilitarian theory), this one also assumes that legislators are addressing themselves in good faith – as they deliberate and vote – to a single question with an objective answer.
The argument this time is due to Aristotle, and it is the one to which we referred in Chapter Four. When Aristotle asks in Book III of the Politics whether political power should be in the hands of the many or in the hands of a few men (or even one man) of extraordinary wisdom and virtue, he entertains the possibility that there is more to be said for the multitude than might at first appear:
For the many of whom each individual is not a good man, when they meet together may be better than the few good, if regarded not individually but collectively, just as a feast to which many contribute is better than a dinner provided out of a single purse. For each individual among the many has a share of excellence and practical wisdom, and when they meet together, just as they become in a manner one man, who has many feet, and hands, and senses, so too with regard to their character and thought. Hence the many are better judges than a single man of music and poetry; for some understand one part and some another, and among them they understand the whole.
137
What lies behind this is the idea that a number of individuals may bring a diversity of perspectives to bear on the complex issues under consideration, and that they are capable of pooling these perspectives to come up with better decisions than anyone of them could make on his own.
Suppose, for example, that the Athenian ecclesia is considering whether to mount an invasion of Sicily. One man may know the geography of the Sicilian coastline; another may know the character of the island’s inhabitants; a third may have experience of amphibious expeditions to other regions; a fourth may be acquainted with the current state of the Athenian fleet; a fifth may be a bitter veteran of past military catastrophes; a sixth may know what is to be gained from the expedition; and so on. No one person may have all this information and experience himself. The most rational way to make the decision, then, is to pool the knowledge of the various individuals.
That is why Aristotle takes it as the mark of man’s political nature that he has been endowed with the faculty of speech. Each can communicate to another experiences and insights that complement or qualify those that the other already possesses, and when this happens in dense interaction throughout a community, it enables the group as a whole to attain a degree of wisdom and practical knowledge that surpasses even that of the most excellent individual member.
Now, there are two ways we can model what goes on in this pooling of knowledge and experience: (i) individual synthesis and (ii) group synthesis.
i. Individual Synthesis. Each legislator listens carefully to the views of each of the others, and forms them into a synthesis which is then reflected in his voting. If he possesses a modicum of rationality and comprehension, the view that he ends up holding should count as authoritative, at least in comparison to the views of any persons who have not been exposed to a diversity of perspectives in this way. In addition, the view of a majority of such legislators would have whatever authority stems from the fact that it appears to embody the most persuasive of the various syntheses achieved in this way.
ii. Group Synthesis. Alternatively, if the issue is a complex and subtle one, an authoritative synthesis may emerge at the level of group action without necessarily emerging at the level of individual understanding at all. This is the model of public deliberation – sometimes referred to as ‘the market place of ideas’ – presented by John Stuart Mill:
Truth, in the great practical concerns of life, is so much a question of the reconciling and combining of opposites that very few have minds sufficiently capacious
138
and impartial to make the adjustment with an approach to correctness, and it has to be made by the rough process of a struggle between combatants fighting under hostile banners.
Here the suggestion is that people simply fling their experiences and opinions into the public forum and, whether others understand them precisely’ or not, they will have their effect, and the truth will emerge by a sort of ‘invisible hand’ process, analogous to that by which Adam Smith thought efficiency would emerge from the operation of the market. The difference between this model and the individual synthesis model is that Mill does not presuppose that each person’s contribution is being carefully taken on board by every other person and synthesized by him with his own view before he sends his modified opinion out to engage with the views of others in a similar way. According to Mill’s model, quite incommensurable ideas may yet have a dialectical effect on one another, so that something better emerges in the discussion, even though the ‘adjustment’ between the various views has not been made by the deliberate synthetic activity of any ‘single mind’.
If anything like this happens, then once again there is a reason to accord to the output of such a process the sort of authority intimated in thesis [J], even though there is no such reason for according authority to the particular views of any individual participant.
5. THE AUTHORITY OF THE GROUP
I hope it is clear now what my argument against Marmor’s thesis is going to be. It is not simply that useful legislation can be conceived to emerge non-intentionally, through the impersonal or mechanical processes we have been discussing. I also want to argue that in the case of each of these processes, any reason we have for according Raz-ian authority to the resultant legislation is also a reason for discounting the authority of the views or intentions of particular legislators considered on their own.
There is an intriguing point about the rule of law here. The model of authority used by Raz and Marmor presupposes that there is one person, X, whose knowledge or situation is such that any ordinary citizen, Y, should trust the directives of X rather than his (Y’s) own reasoning. But it is an important part of our normative concept of legal authority that legislators themselves should be bound by the laws they enact. On the
139
Raz/Marmor approach, it is impossible that a law could have authority vis-a-vis a legislator, for it makes no sense to say that X has better reason to trust the directives of X than he has to trust his own reasoning on the matter that X is addressing. By contrast, the three arguments I have presented have the attractive ability not only to satisfy thesis [J] concerning the normal justification of authority, but also to make it clear why laws should have authority over the legislators who vote for them.
Let me now go through the arguments in turn, to discuss their implications for the respect we should accord to the views, hopes, intentions, and other non-canonical opinions of tindividual legislators.
A. The Utilitarian Argument. On this account, each individual legislator is voting either his own interest or the interests of his particular constituents. But the authority of the resultant legisis based on its claim to represent the general interest. The reasons that led the legislator to vote as he did, then, are not the same reasons that there are for respecting the result:
the result should be respected because it embodies a conclusion about social utility, whereas the former claims to be counted on the ground that it embodies individual or sectional utility. The ground of the law’s authority, therefore, is not reflected in the views or purposes of any particular legislator.
Maybe in very simple cases we can say that the general interest just is the shared interest of a majority of voters. In a straightforward pork barrel case, for example, the issue may be whether to benefit the East-Coasters with a new highway or to benefit the West-Coasters with a new dam (when we cannot do both). If the preponderance of individual preferences favours the latter, it might be thought that we can take the selfish purposes of the West-Coast voters as our best guide to the implementation of the socially-favoured policy. The social decision is about who to benefit; once that has been decided, we might as well take the interests of the victors as our best guide to what the general utility requires.
But a case has to be staggeringly simple to have this feature. If the legislative choice is complex, if benefits and costs are indirect as well as immediate, or if the general interest dictates any form of compromise, then the pork-barrel authority of the individual voters in the majority (such as it is) simply evaporates.
140
B. Condorcet’s Theorem. Condorcet’s theorem amounts precisely to the claim that one will do better by following the majority decision than by following the judgement of a given legislator. What is more, ceteris paribus the wiser the legislators are on average, the more reason there is to follow the majority decision than to follow the views of any legislator chosen at random.
Suppose, however, that one is not picking one’s legislator out at random, but choosing a member of the majority. (This after all is Marmor’s suggestion as to which legislators are the ones whose intentions we should defer to.) Surely Marmor could argue that the fact that a given legislator, L, is a member of the successful majority, shows (by Condorcet’s theorem) that one will do as well by following L’s view as by following the view that emerged by majority voting.
This argument would be faulty, for two reasons. First, the Condorcet theorem concerns ex ante probabilities and turns on the multiplicity of ways in which a majority can be constructed. Once a result has been arrived at by majority voting, the fact that L was a member of the particular majority that supported the result confers no additional authority on his judgement beyond the individual competence with which he began.
Secondly, the only way to determine whether L is a member of the majority – for the purposes of this argument – is by looking at whether his view matches the precise terms of the decision that has been arrived at by voting. It is only with regard to that issue that individual and majority competencies have been established. To the extent that his view varies from this, nothing can be inferred as to its reliability. If it is even subtly different, then the numbers may differ – and the whole gist of the Condorcet result is that small differences may make a big difference, so to speak. This difficulty, moreover, cannot be evaded by insisting that we consult L’s intentions only when they correspond exactly to the issue that was voted on. The whole point about appeals to legislative intent is that evidence about L’s intentions can cast extra light on the interpretive difficulties that courts are facing only to the extent that L’s intentions are somewhat different from those represented conventionally by the ipsissima verba of the statute.
Let me put this last point in a way that takes it slightly outside the mechanics of the Condorcet framework. All suggestions about consulting the intentions of legislators are suggestions about consulting views, purposes and ideas that are not directly disclosed in the text of the legislation
141
itself. (Else what would be the point?) But it is only the text that has been voted on, and it is only relative to that text that we can talk about a majority view. Relative to any other understanding, the identification of a view as being held by the majority is always tendentious, and it is unlikely that its identification as such will ever be less controversial than the issue on which its alleged authority is supposed to cast light.
C. Aristotelian Synthesis. Once again the immediate argument is obvious. Aristotle’s doctrine of the wisdom of the multitude is precisely the view that a group decision is sometimes more to be trusted than the judgement of its most knowledgeable and distinguished member. The latter, then, should defer like a good citizen to the decision in which he participated; and if he should defer, then surely every member of the group should defer. The opinions and intentions, then, that he is subordinating to the text, can hardly be taken by us as a basis for supplementing it.
Substantively, the point here is that a particular legislator’s intentions may reveal only a partial view of the synthesis embodied in the legislation. To interpret the legislation in the light of such intentions would be to distort an integrated whole by biasing it towards one of its parts.
Admittedly, this does not take account of the ex post situation. For after the deliberation and the pooling of knowledge and experience have taken place, it may be true that some (or perhaps even everyone) of the legislators is better informed on the issue than he or anyone else would be coming to it cold, with the benefit only of his own unaided judgement. That is the implication of model (i), of individual synthesis, which we considered towards the end of section 4. It may be thought, moreover, that such individual syntheses have to emerge at some stage or other in the process, otherwise voting by individual members would not ever reveal the wisdom of the multitude. John Stuart Mill’s model (ii) might be all very well as a description of the informal emergence of veridical beliefs in a culture.
But it is one of our basic assumptions about a legislature that its decisions are taken by explicit processes of voting. There are no invisible hands here; there are nothing but real individual hands on the voting buttons.
All the same, an awareness of the second model helps us realize how narrowly circumscribed the first may be for the purposes of statutory interpretation. For if a judge appeals beyond the text of a statute to the intentions of particular legislators, he will be appealing to things said or done during the course of the bill’s passage. At that stage, the decisive synthesis may not have emerged or crystallized in individual minds, and we may be dealing with what can only be described pro tern as ‘the reconciling and combining of opposites. . . by the rough process of a struggle between combatants fighting under hostile banners’. At that stage, to
142
take the inscriptions on any one banner as indicative of the ultimate synthesis underlying the statute may be most unreliable. If, on the other hand! one waits for the final synthesis to emerge, one will be waiting in effect for the text of the final statute – So even if model (i) is the correct image of Aristotelian synthesis, it gives little comfort to anyone who wants to establish the authority of legislators’ intentions expressed otherwise than in the chapter and verse of the text that embodies what the legislature ultimately decides.
6. LEGISLATURES
I want to conclude with some remarks about the assumptions I have been making about the character of modern legislation. I assumed – as I have assumed throughout the book – that any legislature worth discussing is a large multi-assembly, comprising hundreds of persons with diverse views, affiliations, and allegiances. I assumed too that such a body would need formally specified procedures in order to determine what counted as its decisions. And I have assumed implthat any bills it considered and any statutes it enacted would be complex in their contents and accordingly complex in the process of their passage. These assumptions tend to undermine any talk about ‘the intentions of the author’ of a statute. To the extent that there is one author, it is the legislature considered as a body and as distinct from the individual members (or any subset of the individual members) that it comprises. So far as that ‘author’ is concerned, all we have to go on are its formally specified actions; there is no question of our being able to discern or attribute to it any thoughts, intentions, motives, or beliefs beyond that. I argued in section 3 that we would do best to abandon all talk of such intentionality when we are considering such bodies. A slightly more conciliatory way of putting the same point is that we abandon all talk of legislative intentions apart from the intentionality that is part and parcel of the linguistic meaning (i.e. the conventional or in Gricean terms, ‘sentence-meaning’) of the legislative text itself. Beyond the meanings embodied conventionally in the text of the statute, there is no state or condition corresponding to ‘the intention of the legislature’ to which anything else – such as what particular individuals or groups of legislators said, wrote, or did – could possibly provide a clue.
The point, in other words, is this. There simply is no fact of the matter concerning a legislature’s intentions apart from the formal specification at the act it has performed Certainly, the specification yielded by the decision-procedure (that is, by the legislature’s constitution) will be of an
143
action under a certain intentional description, for legislation is, as I said earlier, a speech-act. The intentional description will be that such-and-such words were used with their conventional English meaning. That, however, is all there is to say about the institution’s intentions. In the case of a natural person, we can go beyond that. We can always ask, ‘Well what did you have in mind, what were you thinking of, when you did (or said) X?’ if the meaning of X is unclear. The person’s answer, though not necessarily definitive of the contents of his intention at the time, at least provides us with a clue; for there is some link in the case of a natural person, though certainly not a direct link, between intention and occurrent thought. In the case of a legislature or a corporation, there is no such clue, for there is nothing amounting to the institution’s occurrent thoughts.
What are we to say, then, about the particular thoughts and hopes of individual legislators? Do they bear no relation at all to the legislature’s action or intentionality? The answer is complicated. The intentional speech-acts of the legislature are constitutional functions of the intentional voting-acts of the individual members; but what matters here is simply the intentionality of ‘yea’ or ‘nay’ in relation to a given text, not any hopes, aspirations or understandings that may have accompanied the vote. A legislature is impotent – incapable of action – unless there is a rule for aggregating or combining the votes of its members: the principle of majority-decision, for example. But it does not need, in addition, any rule for combining their thoughts, hopes and understandings into something that would count as the thoughts or purposes of the institution. Its intentionality – such as it is – is quite secure beyond that. Once again, the same point can also be stated in a more conciliatory form. Of course individual legislators have thoughts and intentions and of course those can be loosely associated with the human authorship of the statute. The point is that there is no way of going beyond that sociological platitude to settle any interpretive dispute. People only appeal to legislators’ intentions when there is a disagreement, for example, in a court, about what purpose (or whatever) to attribute to the statute. If the two competing views (or any pair like them) are represented among the non-canonical intentions of the legislators, then there is no way any one such view can be singled out as authoritative for the purpose of settling the dispute.
The defender of legislative intent may make one last reply. Surely – he will say – it matters whether the view we have in mind corresponds to the intentions of the majority or the minority in the legislature. Surely the majority view is – by reason of the rules that allow it to prevail – the more canonical.
144
I think this is based on a deep misapprehension about the political authority of a legislature and about the principle of majority-decision. A statute passed in Parliament is an Act of Parliament, not an act of the majority party. Certainly, it claims whatever authority it has in the community on the basis of that non-partisan characterization (and I am thinking now of the deeper sense of authority explicated, above, in section 6 of Chapter Five). If we think, for example, that ordinary citizens supporting a minority party are bound to respect legislation sponsored by the majority, it is because they owe that respect to the legislature, and to the procedures and institutional forms that constitute it, not because they owe it to the majority as such. Even if the legislative and elective procedures are majoritarian, the citizen’s allegiance is to the principle of majority-decision not to the (members of the) majority.
Partly this is a matter of the complexity of modern legislation, of the multiplicity of views and considerations that must be brought to bear in debates and committee stages, and of the contribution that may be made by members of minority parties to the content of a bill through rhetoric, deal-making or amendment. Our discussion of Aristotelian synthesis here and in Chapter Four captures, I think, how important this can be, and how it undermines any simple appeal to the views of a few privileged members as providing us with a canonical legislative intention.
But above all it is a recognition – once again – that the elementary circumstance of modern politics is plurality and that the form of legislation, as of all collective decision-making, is e pluribus unum. The authority of a law is its emergence, under specified procedures, as a ‘unum’ out of a plurality of ideas, concerns, and proposals, in circumstances where we recognize a need for one decision made together, not many decisions made by each of us alone. The ‘unum’ does not abolish the plurality, nor is it insensitive to it. It is one decision we need, not necessarily one personality, and so it is not merely as a matter of logic that we should refrain from attributing mental states to the legislature. By the same token, however, there is no justification for privileging the mental states of any faction in the legislature as canonical with regard to the decision that has been made by the whole. The decision is made in the name, and for the sake, of the entire community, and one hopes it has been made in a way that encourages
145
rather than precludes a plurality of contributions from a variety of directions. What the decision is – what we have done – is the text of the statute as determined by the institution’s procedures. Those procedures make us one in action, and their identification of something as the text of a statute makes us one as the authors of a deed. Before that, however, and beyond that, we are many, and no further status as a part of the unum can be attributed to anything else that any of us says or thinks.
The other side of this coin is that we should not underestimate the difficulties in the way of a body of legislators understanding one another as they proceed through the stages of law-making. I said in section 2 (it is something I also elaborated in Chapter Four) that we must assume we are not dealing with persons who are transparent to oanother, or who share a comprehensive body of common understandings. Legislators will come to the chamber from different communities, with different ideologies, and different perspectives on what counts as a good reason or a valid consideration in political . The only thing they have in common, in their diversity and in the welter of rhetoric and mutual misunderstanding that counts for modern political debate, is the given text of the measure currently under consideration. That is constituted by the conventions of the shared official language as the only landmark, the only point of reference or co-ordination, in a sea of possible misunderstanding – and even then it is fragile enough and always liable to fly apart on account of the fragility of shared meanings. The point is that the text of the statute, carefully drafted, proposed and amended in accordance with the most formalistic procedures, has a canonical status in legislation that is different in kind from any common view or shared sense of purpose that one might discern in the committee rooms or in the parliamentary corridors. The latter are always tendentious, always likely to be the exclusive province of a few like-minded legislators whose cozy understandings tend to defeat the purpose of a solemn gathering of all the nation’s representatives. By appealing to such views, the courts take the side of what was ‘implicit’ or ‘understood’, and that may be harmless enough in a small homogeneous legislature. But it strikes me that, in a multicultural society, legislators are entitled to insist on the authoritativeness of the text and nothing but the text as the only thing that one can be sure has been at the forefront of each members legislative endeavours.
One final point. I hinted at the beginning of this paper that some American judges make a practice of appealing to certain statements on the legislative record (a formal committee report, for example, or the unchallenged statement of a bill’s manager) which do not traditionally count as part of text of the statute they are considering. It seems to me that Ronald
146
Dworkin has this exactly right when he says in effect that the judges are developing a practice of recognizing such statements as acts of the legislature and that legislators are responding to that recognition by producing statements that are intended to be taken in that way. Nothing I have said is incompatible with these practices: they represent in effect a gradual modification of the legal system’s rule of recognition from the judges’ side, and, as far as the legislature is concerned, they represent a gradual modification of its constitutive procedures.
To the extent that these practices are well-established, they may require us to modify our view (perhaps quite subtly) about what counts as the text of our statute. But they do not require us to deny that our best guide to what counts as that text are the reasons (related to the e pluribus unum considerations I mentioned a moment ago) for identifying certain things and not others as acts of the legislators in their collective capacity. Most importantly for the purposes of this chapter, such practices lend no weight at all to Marmor’s argument that the intentions of a legislator might have authority over and above their status as disclosed in whatever turns out to count in fact as the authoritative text of the legislation itself.
What I hope to have shown, then, is that the force of Marmor’s argument is narrowly confined to those very rare cases where legislation is produced by an individual author whose knowledge and expertise provide the same reason for respecting his intentions as for respecting the text he has produced. More positively, what I hope will follow from this is a greater recognition in jurisprudence of the conditions relating to plurality that are the very stuff and circumstance of our politics.