Defending Aggregated Legislative Intent

Abstract

Theories of aggregated legislative intent posit that the legislative intent of parliament is what a significant enough proportion of legislators intended (e.g., legislative intent is p if a majority intend that p). After all, many think the same way about democracy (‘votes reveal the will of the people’) and about courts (‘a court decision is based on judicial voting’). The existing literature on aggregated legislative intent, however, tends to make two undefended assumptions: (i) Informed Assumption: all legislators have policy intentions; and (ii) Group Intent Assumption: the existence of an aggregated intent entails the existence of a group intent. Despite these assumptions being subject to great scrutiny, they largely remain undefended. This paper defends the Group Intent Assumption and shows that aggregated theories can survive with a weaker version of the Informed Assumption.

In formulating a guiding principle for statutory interpretation, certain theories prioritise legislative intent. There are a few ways to conceptualise the relevance of legislative intent for this endeavour: modern intentionalists have taken the meaning of statutes to be derived from the communicative intentions of the legislature, ¹ while others might focus on the purpose and intended legal effects of the statute to determine its proper interpretation. ²

Nonetheless, there is much debate as to the coherence of legislative intent. For example, Shepsle says that “[l]egislative intent is an internally inconsistent, self-contradictory expression.” ³ Judges have also shared this skepticism, with a majority of the High Court of Australia having stated: “[L]egislative intention … is not an objective collective mental state. Such a state is a fiction which serves no useful purpose.” ⁴

In response, there is a considerable literature defending the idea of real legislative intent which very broadly can be categorised into shared intention and aggregated intent theories. Shared intention theories state that legislative intent is derived in some way from the shared intention of all legislators. Most prominent is the work of Richard Ekins, which focuses on the consequences of the shared intention of all legislators to make law for the common good (elaborated further in Section III). ⁵ In contrast, this paper defends and expands on theories of aggregated intent: any theory that states that the legislative intent for the meaning of a statutory provision is what a significant enough proportion of individual legislators intended. ⁶ For example, if a majority of legislators intended for a statute to have some effect, then that is the legislative intent for the statute. No notion of shared intent among all legislators is required. The intuition behind this view is that we conceptualise fundamental legal institutions in the same way—the ‘will of the people’ is revealed through the electoral process, and the ‘decision of the court’ is the aggregated decision of the court’s judicial members.

The problem addressed in this paper is that existing theories of aggregated legislative intent make two assumptions which have been largely left undefended: the Informed Assumption and the Group Intent Assumption. ⁷ The Informed Assumption states that all legislators have intentions for the meaning or effects of bills (beyond the mere intent to enact the text of the bill). The Group Intent Assumption states that aggregated intent is the same thing as ‘group’ (in this case legislative) intent. Both assumptions have been subject to great scrutiny, and this paper shows how the first can be weakened and the second can be defended. There are other objections against aggregation—such as with compromises, strategic voting, and mathematical impossibility theorems—but this paper will not attempt to canvas them given the already significant literature on these topics (although an extremely brief survey will be provided in Section I). ⁸

An initial methodological point and several caveats are made at the outset. Methodologically, while this paper is theoretical, the feasibility of many of its arguments depends on facts about actual parliaments. In that regard, most of the examples in this paper will be taken from both the UK and Australia, as they share a Westminster system of parliament with a relatively strong party system. In some cases, however, reference will be made to the US, as much of the existing literature grapples with issues arising from that jurisdiction. Nonetheless, the arguments provided here can be generalised to other parliaments or legislatures with similar properties to those identified here.

In terms of caveats, this paper concerns the existence of legislative intent and does not aim to make substantive normative claims. For example, it does not defend intentionalism (although sympathetic to it); ‘intentionalism’ being the theory that the correct interpretation of a provision matches either what the parliament actually intended or what the parliament is reasonably perceived to have intended. ⁹ Even if legislative intent exists, it does not follow that intentionalism is normatively attractive—perhaps what parliament intended should be ignored. Similarly, the paper does not argue that certain conceptions of parliament are normatively better than others (e.g., debates surrounding parliamentary sovereignty). Secondly, the claim is not that there always will be legislative intent beyond the choice of words on the page. Nevertheless, it will be shown that sometimes there might be intention as to the effects of those words or what they should mean.

Despite these caveats, the claim made in this paper is neither trivial nor uninteresting. Existing sceptics of legislative intent claim there are no plausible conditions under which legislative intent can exist or be identified. This paper shows that this is untrue and provides a foundation from which a defence of intentionalism can be constructed. In this sense, this paper shows the possibility that legislative intent can exist. This still has serious consequences for legal theory. Even if intentionalism is undesirable, if legislative intent exists beyond language choice, then non-intentionalists must explain what to do with this intent: Is it to be ignored? Is it to be a partial factor in legal interpretation? The existence of legislative intent has broader repercussions no matter one’s theory of interpretation.

The paper starts, in Section I, by introducing voting as a basic example of aggregation and provides a summary of some of the existing theories of aggregated legislative intent. The ideas of aggregated legislative intent have been primarily developed by public choice and positive political theories of legal interpretation, and I explain why the term ‘aggregation’ is used rather than those labels. In Section II, it is shown how these theories often leave the Informed and Group Intent Assumptions undefended. The rest of the paper shows how these assumptions can be supported. Section III addresses the Informed Assumption by showing that it is plausible that in some cases many legislators (although not all of them) would be informed through organisational knowledge and deference to party experts. Sections IV and V defend the Group Intent Assumption. Section IV does this by introducing minimalist theories of group intent that conceptualise large uncoordinated groups as capable of having group intent (e.g., a spontaneous riot or protest can communicate a message). Section V acknowledges that there can be a plurality of aggregated intent but only some are institutionally relevant to legal interpretation.

I. Aggregation and Legislative Public Choice

Traditionally, theories of legislative intent that utilise aggregation come from public choice schools. ¹⁰ In this section, I explain what aggregation is, summarise some of the work by public choice theories of legislative intent, and then explain why this paper uses the term ‘aggregation’ rather than ‘public choice’.

A. What is Aggregation?

The paradigm example of aggregation is that of voting. An aggregation method is a method of attributing intentions to a group based on the individual intentions of its members. ¹¹ A rule that attributes p as the group intention where the majority of individuals intend p is an example of such an aggregation rule (which will be referred to as ‘majoritarianism’ or ‘majority rule’ in this paper). All possible voting methods used in elections are aggregation methods. ¹² One should not conflate aggregation with majoritarianism, as it is only one type of aggregation rule; many countries use different types of aggregation methods in elections (e.g., first-past-the-post in the US and proportional representation in Australia). Hence, the aggregative explanation for how we can assert ‘the British people chose the Conservative government’ is that some proportion of the people voted for the Conservative government. ¹³

Typically, aggregation methods would require that a ‘significant enough’ proportion of members have the intent for it to be the aggregated intent. This paper assumes a majoritarian aggregation rule is significant enough but is ultimately neutral on whether majoritarianism is appropriate. There are complex debates in social choice theory and judgment aggregation—subject matters that compare various aggregation methods—to which this paper defers. Not specifying the appropriate aggregation rule is not an obstacle to theorising the two assumptions under investigation (Section II), just as the number of aspects of democracy can be theorised without specifying the appropriate voting method.

Models of aggregation were conceptualised to handle the type of disagreement that sceptics of legislative intent claim occurs in parliament. After all, elections still occur successfully in deeply divided societies. As the political scientist Krehbiel notes: “[aggregation] is interesting only when preferences are not perfectly homogenous.” ¹⁴ Governments in democracies are chosen due to elections, a ruling is taken as the court’s ratio because a majority decided that way, and a text is taken as passed by parliament through voting. It is fairly uncontroversial to accept that aggregation in these contexts does explain a group choice, preference, or desire.

Of course, for statutory interpretation, the relevance of aggregated intention is to interpret texts. As a non-legal example of how aggregation might affect interpretation, consider a marketing committee that crafts a short slogan. A small core team from that committee might be tasked with deciding on a slogan along with a marketing strategy. Upon presenting it to the wider committee, if the slogan was approved via a vote we might say that the intention of the committee for the slogan was whatever the core team’s marketing was.

In the legislative context, the Minister introducing a bill would be analogous to the core marketing team introducing their marketing strategy. For example, in the Australian case of Saeed, ¹⁵ there was a question whether the following provision excluded common law rules of natural justice from applying to a certain subdivision in the Migration Act:

51A Exhaustive statement of natural justice hearing rule

(1) This Subdivision is taken to be an exhaustive statement of the requirements of the natural justice hearing rule in relation to the matters it deals with. ¹⁶

The ambiguity here surrounded what “matters it deals with” meant, and the court decided that due to how s 51A interacted with s 57 of the Act, there was a textual argument for why s 51A was not exhaustive of the rules of natural justice. Without going into the details of the decision or s 57, what is important for our purposes is that after Saeed was handed down the parliament amended s 57 (although s 51A remains the same as above). The Minister moved the amendment explicitly as a response to Saeed and said that “the amendment will ensure that the procedural fairness requirements prescribed in the act will apply universally to all visa applications.” ¹⁷ Hence, there was a clear statement that the policy behind the current iteration of s 51A is for no common law rules of natural justice to apply. Since this vote won by a majority in parliament, on an aggregative account, the legislative intent of parliament for s 51A was what was elaborated by the Minister since that is what a majority agreed with. This is of course oversimplistic, and the paper delves deeper into its intricacies, but shows the general picture of how aggregated legislative intent operates.

B. Public Choice Theories of Legislative Intent

Aggregative notions of legislative intent have been mainly developed by public choice theorists. The general idea of public choice theory is to use assumptions and formal methods in economics to analyse political institutions, including parliament. ¹⁸ In the context of legislative intent, I refer to these approaches as ‘legislative public choice’. Three major areas in legislative public choice will be summarised to illustrate what it involves and how it has responded to attacks on aggregation.

In the Migration Act example above, there was a single policy strategy for the bill. However, often numerous factions in parliament will have different motivations for passing a bill. Some object that the preferences of legislators are hard to identify since the text of a bill is the product of the compromise of competing factions with very different motivations. ¹⁹ One area in legislative public choice investigates how to identify aggregated intent in the face of compromises. ²⁰ As an example of work addressing this, Rodriguez and Weingast posit that intentions can still be ascertained from the statements of pivotal legislators. ²¹ ‘Pivotal legislators’ are key legislators who endorse moderate rather than strong versions of a bill. ²² Because pivotal legislators neither want the bill to fail nor want the strongest version of the bill, they have a strong incentive to articulate arguments for the moderate position and in doing so also articulate how other coalitions view the bill. ²³ Thus the clearest picture of the various policies behind a bill can be obtained by investigating the speeches of pivotal legislators.

The solution to the compromises above assumes that legislative history is an accurate record of the thoughts of legislators. A second major area in the legislative public choice literature involves defending the relevance of legislative history. ²⁴ Some object that legislative history is purely ‘cheap talk’ which does not truly signal the beliefs of legislators. ²⁵ As an example of this work, McCubbins, Noll, and Weingast adopt a premise from the economics of signalling that actions are informative when taken by an “informed person” who pays some cost for that action. ²⁶ While some speeches are inevitably cheap talk, it would make no sense for all debate to be cheap talk if legislators are rational; otherwise, legislators would just vote without debate. ²⁷ Legislators pay a cost in taking the time to debate and so the action must be informative in some sense. It cannot be said that what is informative about debates is only that it is a public relations exercise, as it would be more efficient for rational legislators to avoid debate, vote, and then publicise well-written rhetoric for why they voted. ²⁸ Instead, an explanation for the floor debates is that they do genuinely allow legislators to deliberate on policy. ²⁹

Lastly, a third major area in legislative public choice defends the possibility of legislative intent despite the fact that cycling of group preferences can occur. ³⁰ Cycling occurs where individuals might have consistent preferences but the group preference, i.e., the product of aggregation, has the following structure: the group prefers A over B, B over C, but also prefers C over A. ³¹ Notice the inconsistency between the group preferring A to C but also C to A. As an example of a response to cycling, Farber and Frickey note that existing work on aggregation—namely the median voter theorem—shows that cycling is avoided where preferences of legislators can be arranged on a single conservative to liberal scale. ³² In such cases, the winning preference is that of the median voter—in this example, the politically moderate legislator. They further argue that roll call data has shown this single scale from conservative to liberal is a defensible empirical assumption in the US. ³³

Cycling is typically considered a type of mathematical problem for aggregation (while I have explained cycling in English, specialist work represents it in the formal mathematical language). It is recognised that other well-known mathematical issues with aggregation exist. ³⁴ Correspondingly there is a large literature that defends aggregation against such attacks. ³⁵ For example, the political scientist Gerry Mackie states: “[t]he predictions of the chaos model [that cycling is inevitable] fail in human subject experiments, are perhaps impossible to test in natural settings, and utterly lack realism.” ³⁶

This paper will not retread this enormous and technical debate beyond making two observations. The first observation is that these mathematical objections should not worry lawyers unless we adopt broad scepticism towards all kinds of institutions. Since mathematical problems apply to every type of aggregation, these are issues not only for legislative intent but elections and courts as well (since they too operate by aggregation). ³⁷ Kenneth Arrow, of Arrow’s impossibility theorem, said:

Can social choice theory [the mathematical study of aggregation] be useful in criticizing the election procedures or legislative or judicial decision making? The failure to meet the criteria is a legitimate criticism. But since it is universal, it does not serve as a sharp distinction. ³⁸

The second observation is that the work defending aggregation noted above offers a way out for lawyers who do not want to be entirely skeptical about the feasibility of our major governmental institutions. ³⁹

These examples are a limited summary of legislative public choice but will be enough to establish the problems with existing theories in Section II.

C. Aggregation and Public Choice

The reason this paper uses the term ‘aggregation’ rather than ‘legislative public choice’ is that one can accept that aggregation determines legislative intent without buying into the wider economic framework of public choice theory. Certainly, a large part of public choice theory is the examination of different types of aggregation rules. However, public choice theory often commits itself to many classical economic assumptions about institutional actors; for example, the assumption that legislators are rational and always maximise the utility of certain goods—notice the focus above on what incentivises rational legislators. ⁴⁰ This paper focuses on establishing the link between aggregation and legislative intent and stays silent on broader commitments to individual rationality.

II. Two Undefended Assumptions

Despite the richness of legislative public choice, it often relies on two undefended assumptions. The first is the Informed Assumption: all legislators are informed about the intended effects of bills they vote on. For example, we noted above that Farber and Frickey argue there is aggregated policy intent when legislators are arranged from a conservative to liberal scale. However, this assumes that for every statute, legislators have thoughts on the conservative or liberal effects of bills. Also note that built into the premise of McCubbins, Noll, and Weingast’s work on signalling is that the actions are informative when there is a cost to the “informed person” who acts. ⁴¹ Lastly, Rodriguez and Weingast rely on pivotal legislators who have incentives to clearly advocate for their position, which assumes that they have a well-thought-out position. ⁴²

The second assumption is the Group Intent Assumption: once an aggregation is identifiable it entails that there is a group intent. The existing literature argues that there is an aggregated winning intent or preference that can be identifiable through the median legislator, pivotal legislator, or legislative history. Nonetheless, there is no further argument that if some preference is the aggregated preference, it therefore counts as genuine group intent.

There has been much pressure on these assumptions. With the Informed Assumption, it is drafters who write bills, so most legislators are not intimately acquainted with them. ⁴³ Further, legislators are provided with so much information that it becomes impossible for them to process all of it. ⁴⁴ There is thus an argument that many legislators might actually be ignorant about the intended effects of the bills they are voting on. If large numbers of legislators are uninformed, it is unclear if there are any individual intentions to aggregate.

The Group Intent Assumption has been challenged by what I call the ‘groupness’ and ‘pluralist’ objections. The groupness objection accepts that aggregating intent might be possible but rejects that it follows there is a group (or legislative) intent. For example, a majority of lecturers in a law school might prefer black to other colours, but it does not seem appropriate to say, “The law school prefers black.” Analogously, even supposing a majority of legislators intended for a bill to mean m, why should m be taken as the group’s meaning? As Waldron comments: “ordinary citizens … are bound … 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.” ⁴⁵

The pluralist objection was most prominently put forward by Shepsle and Dworkin, who claim there are too many different thought processes that might have gone into a ‘yea’ or ‘nay’ vote. ⁴⁶ A legislator might have voted for a text because of their personal tastes, or a considered value judgment, or what they think their constituents want. If whatever is aggregated is the legislative intent, then it turns out that there is a multitude of legislative intentions (including those of personal tastes and ambitions).

It is noted that the Group Intent Assumption might not be necessary for a theory of legislative intent. Cross has argued that even if all that we have is a set of various legislators’ intentions which is not attributed to a single legislative body, which he calls “disaggregated intent,” he argues that the United States Constitution requires judges to consider that set of intentions. ⁴⁷ This is tantamount to an endorsement of a notion of legislative intent but a rejection of the Group Intent Assumption. I will not consider this option, however, as I argue there are enough resources in the modern philosophy of group action to justify the Group Intent Assumption.

III. The Informed Assumption

It is conceded that not all legislators are informed all the time. Such an assumption is admittedly untenable. However, a weaker alternative to the Informed Assumption is defended, namely the ‘Partially Informed Assumption’: Many (not all) legislators are informed about the general policy issues of the statute or key provisions (or both).

Political scientists and organisational theorists propose that large organisations such as governments do have methods of processing information. I further argue that there is likely to be a culture of deference in parties towards bill experts where there is a history of strong party discipline. As will be seen in Section IV, this weaker proposition is sufficient for legislative intent.

As a preliminary remark, there are ways to carry out aggregation even with very few legislators who are informed: for example, using the aggregation model of Terzopoulou, Endriss, and de Haan. ⁴⁸ To oversimplify their method, they evaluate a set of complete and consistent propositions—in a statutory context, this would be the candidate interpretations of the statute—based on how many members assent to each proposition in that set while ignoring members with no intentions. ⁴⁹ Sets of propositions—or the set of candidate interpretations—with the most support would then be the group intent. ⁵⁰ Even further simplified, this means aggregation rules are only used on those who are informed. For our purposes, the basic point is that there are aggregation methods that have been constructed which can deal with uninformed legislators. A partially informed parliament is still necessary to be established, however, as the aggregated intent of three informed legislators while ignoring 200 legislators who did not think about the bill is unlikely to count as genuine legislative intent.

A. Organisational Knowledge

Organisations can be structured in such a way as to aid in the distribution of knowledge. ⁵¹ Specific to governments, Jones and Baumgartner argue that a theory of government information processing must explain how issues are selected and how information about those issues is understood. ⁵² They make several suggestions as to how governments might be structured to allow for this.

In parliament, the task of issue selection is offloaded to committees and senior politicians who set the agenda and narrow down the possible decisions that need to be made. ⁵³ Hence the policy options include not every single possible solution to a problem, but only the options that the committees and politicians lay before parliament.

For information comprehension, Jones and Baumgartner posit that groups use the same heuristics that non-expert individuals use to digest complex issues. One example is to use indicators to represent complex trends and events (e.g., GDP to represent economic growth). ⁵⁴ Other heuristics include thinking of issues from a political “left-right” dimension. ⁵⁵ Additionally, structural elements of parliament also play a role: while issues are complex and multidimensional, “party leadership’s recommendations are often used to reduce the complex space to a single choice dimension.” ⁵⁶ Depending on the country, the presence of explanatory notes, explanatory statements or speeches, histories, or summaries for legislators also constitutes an important part of simplifying information for dispersal. For example, the speech of the Minister post-Saeed (Section I) would have allowed legislators to recognise that the intent of the amendment was to confirm that s 51A exclude common law rules of natural justice.

This does not guarantee that every legislator will be informed but allows that interested legislators can obtain an idea of the intended policies behind a bill.

B. Deference

Granted that there might be legislators who do not research policy issues despite the availability of the informational resources discussed above, there is good reason to think they would defer to experts on a bill. Consider a legislator who votes on a bill but has done no research. Why vote when they could abstain? There are roughly four possibilities.

1. (a) They were instructed to do so by a party leader.

2. (b) They vote with whatever the party votes for.

3. (c) They vote based on misinformation or limited information.

4. (d) They vote randomly with no regard to consequences.

Both (a) and (b) are what I call deference options. These seem the most likely in strong party systems like the UK or Australia. ⁵⁷ Even in a country with less party discipline, like the US, empirical research suggests that many legislators vote based on legislative history and explanations rather than the text itself, and these documents are provided by those involved with the drafting (indicating deference to their expertise). ⁵⁸ With deference, the claim is that legislators form an intention similar to ‘I intend that this bill does whatever the bill experts intend for it to do’. This deferral is not necessarily something that is at the forefront of legislators’ minds but is a presumption that legislators make when they intend to toe the party line—one cannot toe what does not exist. ⁵⁹ Texts do not appear from thin air; someone thought there was a reason to put those words on a page.

Options (a) to (c) can be aggregated over. Option (d) would be indicative of an uninformed legislator but is highly unlikely both from the perspective of public choice theory and empirical studies. Rationally, public choice theory posits that legislators attempt to maximise their interests through predicting the likely costs and benefits of their vote. ⁶⁰ Voting randomly would be the least likely strategy to do this. Alternatively, if we want to avoid assuming rational legislators, empirical studies in the US show that legislators need to perform competently and be team players within their party to gain respect. ⁶¹ While this research was carried out in the US, it would be surprising if these are not also desirable qualities in other parliamentary settings. Hence, voting randomly is unlikely to further a legislator’s ambitions.

Establishing these deferential intentions requires an identification of the bill experts in the drafting process. To give an example of how this might work, consider Page’s survey—based on anonymised sources—of how drafting typically occurs in the UK’s Office of Parliamentary Counsel. ⁶² The start of the process occurs when instructions from a Minister’s department (or several departments) are provided to Parliamentary Counsel and the relationship is one where departments are treated as clients. ⁶³ Instructions come from departments that can be of “varying quality,” with some containing detailed policy plans whereas others express vague intentions. ⁶⁴ Notably, because the departments are the clients, drafting is subject to those instructions. ⁶⁵

There are thus three possible candidates for the bill expert: Counsel, departmental members, and Ministers. There is sufficient involvement and cooperation among all parties that individual intentions are likely to exist among them which can themselves be aggregated. Consider the following case study by Page, which is a composite of real instances of drafting. Counsel was given instructions to make it illegal for unqualified people to “become mechanics.” ⁶⁶ The underlying purpose was to make it unlawful to carry out dangerous repair works by those unqualified to do so. Counsel was concerned that defining a ‘class’ of mechanics was incoherent, as this was not a club that one could be precluded from joining. ⁶⁷ Instead, Counsel advised that the more appropriate language would be to license the occupational activity of being a mechanic. ⁶⁸ This change of language was consequential, as it meant that the bill would not capture dangerous repairs that were undertaken without pay. ⁶⁹ Further, this proposed amendment ended up requiring existing department policy documents to be modified quite heavily. Nonetheless, Counsel was quoted as saying: “[the government] took it quite well actually. They are used to it.” ⁷⁰

The above case study shows deep policy discussion and cooperation between Counsel and government. Given this level of coordination, it might be conjectured that in some ideal cases, there might be widespread agreement on the intended effect among experts. In other cases, despite some disagreement, all three groups are sufficiently involved that sufficient thoughts on the policy effects would exist that can be aggregated. Due to the sufficient closeness between the experts and the bill, it is unlikely that experts would be uninformed. Even if there were some uninformed experts, it would be a small enough minority that the aggregated intent could still be considered group intent (see Section IV for more discussion of this). Thus within the body of experts itself there can be the aggregated intent of Counsel, Ministers, and the department(s), which is deferred to by members of the party who introduced the bill. ⁷¹ Essentially, two levels of aggregation occur: once at the drafting stage and then again at the final vote in parliament. ⁷² It should be noted that this notion of deference is not what Vogenauer calls an “agency model,” where legislative intent is always offloaded to the bill experts. ⁷³ The expert’s policy intentions are only relevant to the extent that legislators defer to them—if there is evidence in a specific situation that legislators ignore the expert’s policy intentions, then the expert’s intent might not be the legislative intent.

Problems might still arise where not all Counsel and departments work on the whole bill. For example, committee drafters in the US have noted that different parts of large omnibus legislation are often drafted by different groups working in isolation. ⁷⁴ The lack of inter-group cooperation is not insurmountable. If there are three drafting groups A, B and C that draft Parts I, II, and III of a statute respectively, then consistent with our theory of deference we may assume that legislators defer to the intentions of A for I, B for II, and C for III.

Because the parts are written in isolation, this might mean that statutes are not always consistent, but even individuals do not have perfectly consistent intentions and desires. ⁷⁵ People often have conflicting intentions—they might want to both lose weight and eat fast food. This does not mean that those intentions do not exist. The resolution of these inconsistencies depends on one’s theory of interpretation (all theories of interpretation need to deal with indeterminacy or inconsistency in law), and as noted in the introduction, this paper establishes that legislative intent can exist but does not investigate its normative consequences. ⁷⁶

C. An Example

The Australian Migration Act provides statutory authority to the Commonwealth to aid countries that help Australia process asylum seeker claims as follows:

(3) To avoid doubt, subsection (2) is intended to ensure that the Commonwealth has capacity and authority to take action, without otherwise affecting the lawfulness of that action. ⁷⁷

A bill amendment was moved by Senator Leyonhjelm to clarify that ss 198AHA (2) & (3) do not allow persons to take actions that would be illegal in Australia when processing those claims. ⁷⁸ Senator Brandis (a senior government senator) assured the parliament that the “amendment is entirely unnecessary” as “[n]o law can authorise what is, independently of that law, still an illegality under Australian law.” ⁷⁹ Senator Kim Carr stated that the Labor Party (the opposition at that time) “accepts the assurance of the government that this amendment is unnecessary.” ⁸⁰

As noted above, legislators would likely defer to bill experts. In this case, senior politicians from both major Australian parties expressed their opinion on the intent behind the provision. Ultimately, the bill amendment was negatived. ⁸¹ Hence, there is strong evidence that the Senate did not intend for immunity to apply, as there is strong evidence that those who voted against the amendment would have accepted what the senior senators stated as the party position.

There is a remaining question of how the Senate’s intent would interrelate with the lower house since Australia has a bicameral system. In terms of deference, it will be an empirical question as to who is treated as the experts by the various houses. It is likely that the drafters, department, and legislators in charge of the bill would take that role even for the upper house. However, where new information is brought to light, such as in the example above, legislators might additionally defer to the leading members of the respective houses. In terms of the actual aggregation, two options present themselves. One is to treat both lower and upper houses as one big house and aggregate across both. ⁸² Another possibility is to aggregate the houses separately. A rule must then be developed to deal with inconsistent understandings between the two houses. Deciding between the two methods turns on one’s normative and institutional understanding of bicameral parliaments. ⁸³

D. Agenda-Setting

The theory of deference places much emphasis on bill experts (including Ministers introducing the bill). The Minister who introduces the bill hence sets the agenda. A feature of agenda-setting is that the winning outcome might depend on the options available, which are determined by a committee or the Minister. ⁸⁴ Hence the meaning of a statute is highly influenced by a small subset of parliament. ⁸⁵ The alleged problem, as Shepsle puts it, is this descends into a “rule of seniority.” ⁸⁶

There are several responses to this. First, this is a normative argument that senior members should not be able to manipulate the agenda such that a majority intent can be found that suits their wishes. As noted in the introduction, this paper is not a normative defence of legislative intent, and perhaps that is just the reality of how parliaments operate. Secondly, it is not clear why this critique only applies to legislative intent and not hierarchies in party systems generally. On this critique, any bill introduced by a senior party member should be normatively suspect and by extension, all democracies with party hierarchies should be normatively suspect as well. Thirdly, agenda-setting of an informal kind also occurs in elections. Enelow and Hinich comment that if all issues were under consideration during election periods, democracy would be unworkable. ⁸⁷ Instead, voters—who are typically not full-time students of politics—focus on the issues and candidates put forward by parties. To some extent, for a truly operational democracy (and parliament), not all issues can be discussed, and a limited number of actors must set the agenda.

IV. The Groupness Objection to the Group Intent Assumption

Supposing that many, but not all, legislators do have individual intentions, it is still not obvious that aggregating those intentions leads to ‘group’ intent. This is the groupness objection. This section discusses minimalist theories of group intent that explain how mental states can be attributed to large unstructured or semi-structured groups. It will be shown that such theories can be applied to parliament.

A. Group Intent and Explaining the Groupness Objection

Why should we think that groups have intent? Philosophical theories of group intent attempt to make sense of statements such as “We want to watch the next Denis Villeneuve movie” and “Manchester United wants to win the treble.” Examples of such statements also abound in law: we say that “The British people preferred the Conservative party to Labour” and “the Supreme Court decided for the plaintiff.” Some of these statements even appear in legal provisions: international legal obligations of nation-states depend on what a state ‘knows’ in some cases. ⁸⁸

Arguments for the existence of group intent are typically composed of variations of a ‘Togetherness Claim’ and a ‘Mental Distinctiveness Claim’. The Togetherness Claim states that there are some events that involve collectives as opposed to individuals.

Togetherness Claim: There are collective events, characterised by agents acting ‘together’, which are different from individual events where agents act individualistically. For example, two friends walking together in a park is a different type of event from two strangers walking along the same path in a park. ⁸⁹

The next claim is that what is distinctive about collective events is the mental states of the individuals. In the walk example, both the friends and the strangers might have very similar physical actions (all walking in the same direction in the park). What must be different is the intentions of those individuals while walking. As an example, Bratman posits that what is mentally distinctive is that both friends intend that ‘we go on a walk’, both friends intend to do so because they know the other also intends to do so, and there is common knowledge of all of this. ⁹⁰ The claim can be summarised as follows:

Mental Distinctiveness Claim: A major difference between collective and individual events is the mental states of the individuals. ⁹¹

Whatever is mentally distinctive about the individuals in the collective event is referred to as the group intention (e.g., the shared intention to go for a walk). Note that the Mental Distinctiveness Claim does not presume, although it does permit, that group intentions are a different type of intention from normal individual intentions. Neither does the claim presume, although it does permit, that there is some ‘superagent’ that is the group mind. ⁹² It is possible, e.g., with Bratman’s view, that the group intention is just a collection of intentions in individuals’ minds that references the intentions or actions of other individuals.

While the general details of this analysis are plausible for coordinated groups like friends walking, it is seemingly harder to make these claims for larger, uncoordinated groups. Ekins elaborates and also suggests aggregative accounts cannot handle the groupness objection:

Several people seated in a park may all run for the same shelter in the event of rain, each intending for his part to shelter there from the rain. Here, there is no group intention and no group act.… The central objection to summative accounts [i.e., pure aggregative accounts] is that they fail to distinguish coincident intention from jointly held intention. ⁹³

Ekins has broadly proposed a similar view as Bratman to understand legislative intent, that is, an intent derived from mental states shared by all members of a group. ⁹⁴ In Ekins’ case, he argues that legislators share the goal of deliberating and crafting bills for the common good, and this process of deliberation allows legislative intent to be identified. ⁹⁵ The benefit of Ekins’ strategy is that this maintains the structure of the Togetherness and Mental Distinctiveness Claims: there is a collective event where legislators act together to pass bills, and what is mentally distinctive is some shared intent to deliberate and craft laws for the common good.

While it might be argued that these shared intent models ultimately incorporate an element of aggregation (i.e., the aggregation rule for shared intent models is a unanimity rule: the group intent is what everyone agrees on); ⁹⁶ shared intent models differ from what this paper calls ‘aggregation models’ in two important ways. The first is that aggregation models will generally treat a unanimity aggregation rule as uninteresting. The point of using aggregation is to allow for disagreement. The second is that shared intent models typically go further than just unanimity. The mental state that members must have generally is some strong interdependent mental content—e.g., common knowledge about the intent. When I use the term ‘aggregation model’ I use it in the sense of being interested in more than just unanimity and, as will be explained in this section, not needing interdependent mental content.

This paper does not attack shared intent models and merely raises aggregative models as an alternative. I shall not retread criticisms of shared intent, but the basic gist is that these models do not explain non-cooperative elements of parliament very well—legislators often do not act together in a coordinated manner, especially on the policies behind bills. ⁹⁷ While Ekins does try and resolve this by relying on what a “well-formed assembly” ⁹⁸ would understand the proposals to be, this has been criticised by others as lacking in detail as to how interpretations of statutes are identified or relying on extremely idealistic notions of parliament. ⁹⁹ This is not to say those criticisms are successful, but they give reason to explore views of legislative intent not so reliant on cooperation.

In this section, I explore minimalist accounts of group intent that show how the Togetherness and Mental Distinctiveness Claims can still apply in cases of larger uncoordinated groups. Minimalist accounts are those that do not require highly interdependent mental states among group members for group intent. ¹⁰⁰

B. A Minimalist Version of the Togetherness and Mental Distinctiveness Claims

Minimalist theories argue that shared intent models cannot be the only way to generate group intent, as the requirement of shared intent is too stringent to capture ‘loose groups’. ¹⁰¹ Loose groups are those that either lack a codified organisational structure or have unreliable or incomplete communication channels between their members: e.g., mobs, social movements, a sizeable military unit, robbers carrying out a complicated bank heist, etc. ¹⁰² Members of these groups often do not have shared intentions except for very general ones like “we are part of the same movement.” It is possible, consistent with our discussion in Section III, that some of these members are also uninformed as to certain aspects of the group’s actions.

To illustrate, consider the Stonewall riots which occurred after a police raid of a gay and lesbian bar. ¹⁰³ A participant of the riots later recollected how it was neither planned nor coordinated:

We all had a collective feeling like we’d had enough of this kind of shit. It wasn’t anything tangible anybody said to anyone else, it was just kind of like everything over the years had come to a head on that one particular night in the one particular place, and it was not an organized demonstration. It was spontaneous. ¹⁰⁴

Minimalist theorists argue that the Togetherness Claim can still be made in examples like these; there are events other than those of very coordinated groups which can still constitute a collective event. In the Stonewall riots, the participants did not know each other and there was no discussion on why they were rioting. Yet, there is still a way in which the riots were a collective event given the ‘collective feeling’ of the participants. We might contrast the riots with a large drunken brawl in a pub. Some patrons might defend themselves, not knowing why they are being attacked. Others are trying to run out of the pub. Some might be so drunk they join in the fight for no reason. The brawl is far less of a collective event and perhaps is not one at all (not collective in the sense of missing a ‘togetherness’ element).

On what is mentally distinctive about these loose groups, minimalist accounts tend to posit that there can be group intent where

1. (a) There is some general collective goal, or an event involving a collective.

2. (b) The members intend to participate or contribute in some way to the goal or event.

3. (c) Common knowledge is not necessary (not everyone needs to know that they all share the same goals and plans). ¹⁰⁵

Applied to Stonewall, there was a general understanding that this was a protest or riot (feature—(a)) and the rioters intended to engage in the protest (feature—(b)). There is no need for everyone to have common knowledge as to the exact point of the protest or what each member is doing (feature—(c)).

As an example of a specific minimalist account, consider Kutz’s notion of participatory intentions. Kutz defines the collective goal or group intent as the “product” of the overlap of intentions to participate in some group activity or action. ¹⁰⁶ In terms of “participatory intentions,” a patron of the Stonewall pub might have started throwing bottles at the police with the hope that others would join. ¹⁰⁷ Since the group intent is the product of participation, if enough people started engaging in that activity it would constitute overlap and then the group intent to protest would arise. ¹⁰⁸ It is necessary, to ensure that this is a collective event, and establish the Mental Distinctiveness Claim, that there is some mutual expectation among group members that they are engaging in the same activity. ¹⁰⁹ However, Kutz maintains this hope might only be a very weak expectation. ¹¹⁰ Thus it is sufficient that the initial protester hoped that others would join and that those who joined were responding to those preceding them.

As noted above, what is mentally distinctive about collective events might just be some set of intentions in the minds of individuals (e.g., “I intend to throw things at the police with other people”). Further, this paper is open to group intent being reducible to those intentions. In which case, to assert ‘group intent that p exists’ is just short-hand for asserting ‘some aggregation of people intended to participate in some event because of, or in order to, p’. Combining minimalism with reductionism allows ‘the Stonewall group rioted because they were unhappy with police treatment of gay people’ to be true by virtue of the uncontroversial fact that a lot of patrons of the Stonewall pub intended to riot because of their displeasure with police action.

This reductionist and minimalist strategy is thus a simple claim that factually there are loose groups that possess some collective dimension, and what is distinctive about these loose groups is that their members intend to participate in some event. This reductionist strategy does not render group intent meaningless. Just because a collective event is really just an event involving individuals, does not mean that those individuals act in the same way as they would in a purely individual event. In Ekins’ example of random individuals running for the shade, there were no overlapping participatory intentions to act in the one event. In contrast, in cases like riots or mobs, there are participatory intentions to act. Hence, a minimalist-reductive account of group intent still distinguishes between loose collective events from entirely individual events.

C. Identifying Content

It is important for the purposes of legal interpretation to identify the content of the group or legislative intent (i.e., that the legislative intent is for a bill to mean p). So far, I have described when loose group intent exists, but not how to determine the content of that intent. I propose that among minimalists, Kutz’s overlapping participatory intentions, discussed above, is helpful in identifying such content—the content of loose group intentions is determined by the overlapping participatory intentions.

Problematically, Kutz seems to endorse what I shall call a ‘lowest common denominator’ account of overlap: the overlap is the minimal content in common among all differing participatory intentions. ¹¹¹ To show why this is an issue, a lowest common denominator account means that the presence of those who are misinformed, ignorant, or dissenting would destroy, or twist, what would intuitively be understood as the content of loose group intent. For example, with the Stonewall riots, there was evidence that the musician Van Ronk was involved because he had anti-police sentiments and not necessarily because of deep commitments to LGBTI advocacy. ¹¹² The lowest common denominator account would mean that the presence of this single person is enough to show that the Stonewall riots were only about anti-police sentiments and not gay and lesbian oppression since that is what all rioters had in common. This would lead to a rather strained interpretation of this historic event in the LGBTI liberation movement.

The same can be said for other loose groups. Consider the bombing of a military target where two out of a hundred in a military division participated in the event but did not realise they were attacking a real enemy as opposed to a war games exercise. The lowest common denominator account prevents us from saying the army division intended to bomb the military target. It is these kinds of groups—the Stonewall riots and military divisions—that minimalist theories purport to explain but get very wrong if they use the lowest common denominator account. In essence, the lowest common denominator account forces us back into unanimity—the minimal content everyone shared—to theorise loose groups that are not meant to have shared or unanimous intent. This is an odd theoretical mixture.

The above examples show that a theory of overlap and content must allow for ignorance or dissent; otherwise, it prescribes odd group intent to the Stonewall riots and our hypothetical military exercise. Aggregation rules that allow for partial information (see Section III) would very easily satisfy these constraints of allowing ignorance and not requiring unanimity. These aggregation rules would also explain why the minority of dissenters or ignorant members in the above groups did not affect what was overall mentally distinctive about those loose groups. Hence, on a minimalist account, an aggregation rule is a very natural method of conceptualising overlapping content and still retaining the groupness of that content—the content is not aggregated over any random number of people but over a group of people who intend to participate in the same event.

The aggregated content, however, can differ from how some might naturally describe the group content. To explain what I mean by this, consider that a legislator who votes might plausibly be taken to intend to ‘participate in the lawmaking act with the desire that this law will have effect e’. If there is enough overlap, then we can say a majority of legislators vote with the intention to ‘participate in the lawmaking act with the desire that this law will have effect e’ (call this the ‘aggregate description’ of content). It is plausibly argued that these individuals can only intend to participate with the desire of certain effects and cannot intend to make the laws themselves as only parliament—the institution—has the ability to make laws (not just participate) with definite effects. ¹¹³ Individual legislators have no such power. However, the natural way that lawyers describe legislative intent is that parliament ‘intends to pass the law with effect e’ (call this the ‘natural description’ of content). Notice that the participatory intention and the desire have disappeared from the natural description. The parliament is not participating in making a law with other people, it is making the law itself. Why this matters is that it is odd on a reductionist account, as introduced in Section IV.B, that the natural description of content is different from the aggregate description of content: How does the reduction happen if the content is different?

Recall the reductionists’ view of group-level events in Section IV.B: ‘The group intends p’ is a shorthand for ‘Some aggregation of people participate in an event because of or in order to p’. The same strategy can be applied to content. The natural description of content is also just a shorthand for, or perhaps a metaphorical way of describing, the aggregate description of content; an assertion of parliament intending to make law is only a convenient way that lawyers speak about the fact that many legislators participated in lawmaking with the desire for some effect to become law. On a reductionist minimalist account, only facts that meet the aggregate description exist and are literally true. Nonetheless, as noted above at the end of Section IV.B, reductive minimalism is not redundant, as it still separates collective events from individual events. The aggregate of many legislators participating in lawmaking is itself still a genuine collective event. The job then turns to a theory of legal interpretation to explain whether the facts described by the aggregate descriptions are relevant in telling us what the content of law is. I explore this in Section V.

It is of course possible for an aggregation account to dislike this kind of reduction and instead offer an emergent account where the natural description fact supervenes on the aggregate description fact (supervenience meaning that changes in aggregate descriptions affect the natural description). However, the natural description fact does not reduce to an aggregate description fact on the emergent view. ¹¹⁴ Just as the existence of a beach depends on sand particles but the property of ‘being a beach’ is quite different from the property of ‘being a sand particle’, the fact that the parliament intends to make law is some kind of emergent phenomenon from many legislators participating in lawmaking. This puzzle of natural description facts being different from aggregate description facts is then just a special case of explaining emergence which affects many other systems as well.

D. A Minimalist Account of Legislative Intent

We now have all the tools to construct a minimalist notion of legislative intent. To summarise the minimalist account of group intent that has been discussed:

1. (i) The existence of loose groups: Some groups have loose features—they are not entirely coordinated—and yet seem like they involve collectives rather than individuals working on their own. A minimalist account of groups or group intent asserts that loose groups do in fact involve true collectives.

2. (ii) The Togetherness Claim: What makes loose groups collective is that they still involve multiple participants contributing to an activity or event—this makes the event a collective one.

3. (iii) The Mentally Distinctive Claim: What is mentally distinctive about such loose groups is that those participants intend to participate or contribute to the activity or event, even if there is no common knowledge or understanding or plan about the activity or event. The mentally distinctive nature of the loose group can be considered the group intent.

4. (iv) Aggregated Content: We can identify the content of what is mentally distinctive for the group based on the participatory intentions of individuals. The method most coherent with the minimalist program is to determine that content based on aggregation.

There is then a question whether the description above of loose groups just reduces to the individuals or whether it is an emergent phenomenon. As noted in the previous sections, I am open to both, although I think the reductionist account is simpler in that it does not need to deal with emergence and yet can still separate collective from individual events.

I will apply these features to the parliament in turn. First, parliaments have features of looseness when it comes to the intended effect of bills. Parliament is not quite as loose as a riot or social movement but is certainly not highly coordinated either when it comes to policy development. Let us say that a group decision or activity is ‘structured’ (as opposed to loose) the more communication and agreement there is between its members. There is a scale of looseness in terms of how many activities of a given group are loose or structured. A spontaneous riot is highly loose since it started without any agreement on rioting. On the other hand, where every decision of the group needs unanimous approval, this would be highly structured. However, there can be examples in the middle which have some structured and some loose features (we might call them semi-loose). Roy and Schwenkenbecker’s example of a specialist military option falls under this category; normally such units have a clear line of command and communication, but in specific missions such communication might not be available. ¹¹⁵

Parliaments are a semi-loose group. The aspects of parliament that are highly structured are the agreed-upon procedures by which bills are debated in parliament. ¹¹⁶ However, these procedures do not apply to the deliberations and policies behind the specific content of bills—in theory, all legislators are free to do as little or as much research or engagement as they please. In fact, it would be very surprising if parliament was considered a structured group in the context of policies; parliaments are typically comprised of at least two parties which do not aim to agree with each other and which do not necessarily communicate with each other. While it was noted that deference is a serious aspect of party politics, it is not mandated by parliamentary rules. So parties might have quite a bit of structure to them, but this structure is lost when extended to the parliament as a whole. Hence there is looseness in parliament in regards to the content or policy of bills.

Second, the passing of a law is a collective event since there is an event that legislators are participating in together—the parliament passes a law, not individual legislators. Thus, the Togetherness Claim is established.

Third, what is mentally distinctive about loose collective events is overlapping participatory intentions. Legislators have overlapping intentions in passing laws in a certain way (although they might not agree on the exact way) and as a result, the Mental Distinctiveness Claim is made. Hence, we can say there is some version of group or legislative intent.

Fourth, as further argued above, the most natural account of identifying the content of large loose group intentions is an aggregative one. So to determine the content of legislative intent, we aggregate over those participating—just like we would with the protests.

To illustrate these points concretely, consider again the Migration Act example from Section III.C—on whether people processing visas offshore can carry out acts illegal in Australia. All the senators have the intention to participate in the lawmaking event of the Australian Senate by virtue of being in the hall and voting. The question then is whether there are any overlapping intentions of the senators beyond the mere passing of the bill. As noted above, due to both government and opposition leaders claiming that the provision would not permit illegal acts, and by virtue of deference, we might say that a majority of the Senate have intentions that s 198AHA(2) would not allow for actions that are illegal in Australia to be carried out. Hence on the Togetherness Claim, there is some collective event that was occurring when the Australian Senate was passing s 198AHA(2). What is mentally distinctive about this event is that there are overlapping intentions to pass a bill, specifically s 198AHA(2), with the desire that it does not allow for actions illegal in Australia to be carried out by foreign actors. Since the best way to determine content in a semi-loose group is aggregated content, the legislative intent here is to not allow officials who process offshore visas to take actions which would be illegal in Australia.

To reemphasise the points above, it is not asserted that overlapping participatory intentions are metaphysically unique—e.g., there being a group agent. The proposal is simply that such mental states separate certain collective events from mass individual events. The relevance of such an event to interpretation is discussed next.

V. The Pluralism Objection to the Group Intent Assumption

A second objection to aggregated legislative intent being genuine group intent is that this leads to all kinds of legislative intent, since legislators have all kinds of motivations for voting for a bill. This is the pluralism objection. It is argued that all participatory intentions do contribute to different types of legislative intent, but only some are institutionally relevant to the interpretation process.

A. Institutional Relevance

I propose that all aggregated intentions count as real group intent—even personal ambitions. If a large number of legislators passed a law because they were lobbied by big corporations, then there is legislative intent to pass a law and to support big corporations.

At this point, the question is one of relevance: Which legislative intentions are relevant to interpretation? Just as an individual person can have complex and varied thoughts on any issue, so can a group of people. An attendee of a staff seminar might think that the arguments were unpersuasive but the powerpoint slides were nicely made. Both are real thoughts, but only one of them would be relevant to the subject matter of the talk. The same occurs with parliament. The specific notion of relevance in the context of legal interpretation is a type of institutional relevance which depends on one’s conception of parliament and the judiciary: Specifically, whether there are institutional reasons to aggregate over certain types of participatory intentions. For example, consider representative democracy. People have varied reasons for voting for a certain government—from genuine policy concerns to impressing a romantic interest to being misled by voting cards. In a representative democracy, there is an institutional reason for the government to identify the policy preferences and interests of the citizens and not other interests.

In relation to legal interpretation, the relevance depends on what institutional reasons there are for courts to identify what legislators think. This would depend on one’s theory of parliament and courts. No doubt under some interpretive theories, what legislators think will never be institutionally relevant. As noted in the introduction, this paper does not defend any one theory of parliament (among other constitutional principles). The paper does, nonetheless, show that there are several conceptions of parliament where there can be institutional relevance of aggregated intent.

To show how institutional relevance operates, an intentionalist and a non-intentionalist example are provided where personal ambitions are irrelevant. One popular intentionalist view is that parliamentary sovereignty entails that the role of parliament is to communicate messages about the law and the role of a judge is to decode that message. ¹¹⁷ In that parliamentary role, however, its authority is to change legal effects through communication; personal ambitions are not relevant to such a role. To identify what is communicated, a plausible intentionalist argument is that an individual legislator’s communicative intentions are relevant as it is legislators for whom citizens vote. Hence democratic concerns make the overlapping communicative intent about policy effects institutionally relevant.

For a non-intentionalist example, Greenberg argues that the interpretation of statutes is determined by how the actions of parliament impact the moral rights and obligations of citizens. ¹¹⁸ For example, a parliament’s law to drive on the right-hand side of the road makes it morally obligatory to do so for the safety of others. Nonetheless, he still states: “On the Moral Impact Theory, all of the linguistic and mental contents associated with the legal texts are among the factors that are potentially relevant to our obligations.” ¹¹⁹ He further argues that grounds like democracy and fairness might provide a partial moral reason to identify communicative intentions, although this might be trumped by other normative considerations. ¹²⁰ Legislator’s intentions are thus partially relevant, although not fully determinative, of what the law is. Either way, neither democratic nor fairness norms would likely make the personal ambitions of legislators relevant, although they might make their policy intentions relevant.

Aside from precluding personal ambitions, institutional relevance also provides a more fine-grained account of the type of aggregated intent relevant to legal interpretation. For example, Goldsworthy distinguishes between what a parliament communicates and how it expects a law will be applied. ¹²¹ Goldsworthy argues that communicative intentions trump specific applications because the rule of law and separation of powers leave it in the hands of judges to determine how to apply law. ¹²² Application is not the role of parliament. On the other hand, an intentionalist strongly committed to the democratic legitimacy of legislators might prefer applications, since that is how legislators intended for the law to operate. Hence, the idea of institutional relevance also provides a method for deciding what level of policy intent is relevant to interpretation.

To show how this might operate in practice, suppose a parliament in the 1960s produced a statute that communicated a prohibition of discrimination when made based on sex, gender, and race. This was the communicative intention. Further suppose that since it was the 60s, parliament did not think it would apply to the LGBTI community and there was clear evidence that this was the case. ¹²³ This was the application intention. As noted above, Goldsworthy would argue based on rule of law and separation of power arguments that communicative intentions are institutionally relevant to legal interpretation but not application intentions. In this case, Goldsworthy can argue that the idea of ‘discrimination’ communicated entails protection of the LGBTI community even if the legislators did not want that consequence. On the other hand, the strong legitimacy intentionalist argues that application intentions trump communicative intentions. In this case, for democratic reasons, such a statute would not protect against discrimination toward the LGBTI community, since there were clear application intentions not to include them (while conceding that such a law might be morally imperfect). Note that in this discussion both communicative and application intentions exist; the legal controversy is which ones are institutionally relevant—Goldsworthy and the strong legitimacy intentionalist differ on that account.

B. Dworkin’s Two Critiques

The notion of institutional relevance still faces two criticisms that Dworkin put forward in Law’s Empire.

The first is that this is a pyrrhic victory for the intentionalist. If a theory of institutional relevance is required, then legislative intent “must be answered in political theory, by taking up particular views about controversial issues of political morality. So the speaker’s meaning theory [or intentionalist] cannot make good its presumed claims of political neutrality.” ¹²⁴ Perhaps Dworkin is correct; this paper does not attempt to defend any specific type of neutral-intentionalism, just that some notion of legislative intent can exist. It might very well be that intentionalists cannot be politically neutral. Nonetheless, even in this paper, the engagement with political morality is fairly minor. The conceptualisation of parliament can be seen as a sociological task of observing which existing norms in the relevant constitutional system pertain to parliaments rather than a deep engagement with moral and political philosophers.

Dworkin’s second critique is that one shouldn’t consider the hopes and desires of legislators. He argues that the hopes of legislators “very often do them no credit”—e.g., a legislator might hope that the effect of an environmental bill will be very narrow due to their political ties to large corporations. ¹²⁵ Dworkin claims these “selfish ambitions … have no place in any acceptable theory of legislative interpretation.” ¹²⁶ The argument, it seems, is not just that there are selfish ambitions but that they might be tied very strongly to the intended effect of bills. At the core of the objection is that it is impossible to separate policy (e.g., the narrowness of the bill) and non-policy (e.g., monetary motivations) intentions; otherwise, an intentionalist can just ask judges to ignore the non-policy intentions.

As a preliminary note, this non-separability claim affects elections as well: non-policy issues will also often influence votes for a politician (e.g., the sexual orientation or race of the politician). ¹²⁷ Hence if non-separability ruins legislative intent, it also ruins the ability for politicians in a representative democracy to identify what interests they are representing.

More substantively, a few responses can be made to the claim of non-separability. First, it does seem like the policy and non-policy intentions of Dworkin’s selfish legislator can be separated. The legislator hoped that ‘The Act be construed narrowly’ because ‘I have big corporate sponsors’. One can just adopt the clause occurring after ‘that’ and ignore the non-policy content after ‘because’ (for the institutional reasons discussed above). Second, even if policy and non-policy intentions are inseparable, it is not clear that selfish legislators are an issue for intentionalists. If there is evidence that a bill was passed to gain favour with certain corporate lobbyists, an intentionalist might very well be licensed to interpret the law as pro-corporation. Intentionalists, and this paper, make no comment about the morality of the content of legislative intent. Just as private individuals can have bad intentions, the same might be said about politicians and parliament.

VI. Concluding Remarks: Consequences for Legal Interpretation

This paper has aimed to show that there is reason to believe that many legislators do have intentions as to the effect of bills (even if it is deferential) and that the aggregate of such intentions can be considered genuine legislative intent. Legislative intent is a product of the participatory intentions of legislators just as we might say that there is a group intent behind protests or marketing teams. This is the same explanation for why we might say that there is the ‘will of the people’ or that the Stonewall riots were related to gay and lesbian liberation.

One feature of this theory is that it was not claimed that there is legislative intent on every provision and every single word. Nonetheless, a couple of examples were provided where this theory of legislative intent still has practical consequences for legal interpretation assuming intentionalism: the case of Saeed in Section I and the Migration Act debate in Section III. Given this concession, it might be queried whether aggregated legislative intent can truly support intentionalism if it only exists some of the time. Although I do not defend intentionalism in this paper, I maintain that accepting the theory of legislative intent in this paper is not fatal to further work on intentionalism. First, where there is a clear distinction between textual meaning and legislative intent, intentionalism will still lead to a different result. For example, recall in Saeed (see Section I) that there was a question as to whether s 51A excluded common law rules of natural justice. Even before Saeed, a Minister had noted “the purpose of this bill [to introduce s 51A] is to make it expressly clear that particular codes in the Migration Act do exhaustively state the requirements of the natural justice or procedural fairness rule.” ¹²⁸ Nevertheless, their Honours in Saeed noted that no matter how clear a Minister’s statement or an explanatory memoranda, a textual reading must be preferred, thus leading to the decision that s 51A was not exhaustive. ¹²⁹ On the aggregative account, it is likely the explanation by a key government minister to which members of their party defer. Hence an intentionalist using this account would think that Saeed was wrongly decided, while a textualist would not.

Second, neither intentionalists nor textualists should expect that ordinary meaning would often depart from legislative intent; a person wishing to communicate through writing to large audiences would often rely on ordinary and public meaning. ¹³⁰ Hence an intentionalist might be justified in theorising that most of the time, if there is no legislative intent beyond text, the parliament just intends to enact the ordinary meaning.

Third, it is perfectly coherent to posit weaker forms of intentionalism which only make claims about the priority of legislative intent rather than its frequency. ¹³¹ What this means is that intentionalism only stipulates that where legislative intent exists, it should be prioritised as the correct interpretation. This version of intentionalism does not predict, as an empirical matter, that there are often individual intentions in play on the effects of every provision.

In essence, while the theory of legislative intent advocated in this paper is limited, it is still potentially useful for questions of legal interpretation.