I. Legal Certainty is Overrated

There is an important strand of scholarship that questions the value commonly attributed to clear and precise laws, instead praising the various desirable ends that can be served by laws that are indeterminate. Some question the political implications of placing such emphasis on certainty in legal writing, concerned that conceptions of the rule of law that emphasize legal certainty are hostile to progressive political agendas, or else assume an unduly narrow conception of individual liberty.Footnote ¹⁶ Others simply stress that indeterminate laws (can) serve a suite of beneficial purposes, which tend to be ignored.Footnote ¹⁷ Waldron, for example, argues that laws that employ contestable value predicates sponsor a degree of “thoughtfulness” by requiring the law's subjects to decide for themselves what the law requires.Footnote ¹⁸

This debate about the relative virtues of determinate and indeterminate laws has taken on new significance, given the rise of legal technologies which promise (however plausibly) to convert legislation into code. Demand for automated systems that execute code—for example, to replace or assist human decision-makers—has generated demand for legislation to be more code-like, drafted in a way that employs “if, then” logic with minimal need for human interpretation.Footnote ¹⁹ This nascent idea of “rules as code” has prompted some legal scholars to defend the indeterminacy of legislative texts, to highlight what might be lost if it became a mainstream approach to designing legislation. Mireille Hildebrandt and Laurence Diver, for example, argue that the indeterminacy of legislative texts is far more than an inconvenient barrier to law by algorithm.Footnote ²⁰ They describe the multi-interpretability of legislative texts as a protective feature, which allows for critical reflection and constructional choice at the point of application and hopefully guards against over- and under-inclusive laws. While these authors approach the issue of legal certainty from very different perspectives, they share a common sentiment: that a legal system comprising perfectly precise laws mechanically applied is not only impossible but undesirable; that “in some cases the craving for certainty matters less than the need for reflection, flexibility, reason and thought.”Footnote ²¹

We might question whether people really engage with legislation in this way, undertaking the kind of thoughtful and critical reflection that these authors describe.Footnote ²² But this is not directly relevant to the issue with which I am concerned, which is whether legislation should be reasonably knowable. Ultimately, I do not think that any of this scholarship defeats, or indeed attempts to defeat, that claim. It is rather a call to reconsider how much people need to know about the law and how that knowledge can be gained.

Thus Waldron challenges the assumption, that people can only be guided by law if it takes the form of a precise rule. He argues that standards can do the job, at least in certain circumstances. These “mobilize … the resources of practical intelligence possessed by the norm subject” to deliberate about what (for example) would be “reasonable” or “proper” in the relevant circumstances and adjust their conduct accordingly.Footnote ²³ This is a legitimate, even valuable, style of legislating—particularly in those areas of “law and life” where it is reasonable to assume that individuals’ conceptions of what those standards mean are likely to align.Footnote ²⁴ This, very briefly stated, is why Waldron rejects the idea that a statute that directs motorists to drive at a speed that is “reasonable and proper” leaves them unable to understand what is prohibited. But this does not deny that the law should guide. It is an argument about how the law can guide. It is also an argument about what guidance entails—essentially, that a person can said to be guided by the law in a meaningful sense, even if in working out what the law requires of them they have to undertake some independent reasoning of their own.

Further, Waldron's argument is largely confined to the question of whether legislatures should employ “moral” or “value predicates,” which is just one reason why a statute could (arguably) be described as difficult to know, and not the question with which I am concerned, of whether we should accept a system in which legislation is too voluminous, rapidly amended, intricate or dense for people to understand without an unreasonable expenditure of resources. As Waldron makes clear, his “reasonable speed” example would not be a sound way of legislating in every context.Footnote ²⁵ People likely expect that there will be some legal rule about how fast they can drive, and people have resources that they can draw upon to assess for themselves what speed is reasonable. Indeed, in many areas, the law roughly aligns with underlying moral norms, and one might argue that the need for certainty is less when they do. After all, one does not really need to know section 18 of the Crimes Act 1900 (NSW, Australia) to predict that they will be punished if they murder another. But legislation now ventures far beyond shared moral norms—and when it does, it is less reasonable to expect that vague legislation can convey what its subjects are required to do.

Nor do I think the rules of code critics refutes the idea that law should be knowable. Rather, I take these scholars to argue that knowability must be achieved in a way that leaves space for discretion and adaptation at the point where law is applied. Hildebrandt, for example, accepts that the rule of law requires “closure”: a clear and binding answer regarding what the law means and requires.Footnote ²⁶ The point is that this closure should be provided by a court, stepping in to decide which of the multiple possible interpretations of a legal text is legally binding, rather than legislative texts themselves. There is a point at which multi-interpretability may undermine the knowability of legislation,Footnote ²⁷ but I also think there is a meaningful distinction between knowability and certainty—and that a standard of reasonable knowability can accommodate interpretive disagreement. As I noted at the outset, this would not seem to require iron-clad certainty about how the law will apply in every case. More intriguing, and pertinent for my purposes, is the emphasis that Hildebrandt and other scholars place on judicial review. The next section examines this idea.

II. Contestation will Suffice

While it is commonly assumed that legislation should be knowable, it is rarely said that this is the only quality that good legislation or a good legal system should have. Instead, reasonably knowable legislation is usually said to work in tandem with legal procedures that allow people to contest the way that legislation has been applied.Footnote ²⁸ But Vincent Chiao has recently made a more ambitious claim: that contestation will suffice. Chiao's premise is what he calls “hyperlexis”Footnote ²⁹—a state in which the sheer volume of legislation outstrips the capacities of human actors to know its content in advance. He presents this as a problem for the rule of law that, along the lines I sketched at the outset, is traditionally understood to require that the content of the law be knowable in advance. His conclusion is not that the rule of law has failed, but that we should shift emphasis from one aspect of it to another, abandoning a “planning conception” of the rule of law that focuses on “law's ex ante function in guiding behaviour” and embracing “a contestatory theory” instead—which, he says, is not compromised by hyperlexis.Footnote ³⁰

But the ability of the law's subjects to contest it is surely compromised by hyperlexis too. It is notorious that the complexity of legislation hinders the ability of legal subjects to identify legal causes of action, and increases the costs of pursuing them, which may discourage them from doing so at all.Footnote ³¹ Those costs reflect, among other things, the considerable training and time required to enable even legal experts to understand complex legislative texts. This afflicts all cases, including potential applications for judicial review. More than thirty years ago, Bruce Dyer wrote that “judicial review is usually only a practical option for corporations and wealthy individuals who have substantial sums at stake.”Footnote ³² There are some potential exceptions, including those who have pro bono assistance, or are lucky enough to receive legal aid, and those “in such desperate circumstances” that they are willing to represent themselves, undeterred by “the threat of an adverse award of costs.”Footnote ³³ There are also “new administrative law mechanisms” in many legal systems, which tend to be more accessible than courts and do not place the same burden on legal subjects to call government to account.Footnote ³⁴ While the volume of determinations assessed by these institutions tends to dwarf that of those reviewed by the courts,Footnote ³⁵ it remains an open question whether and to what extent they contribute to the culture of contestation under discussion here. This would require consideration (among other things) of the way these mechanisms fit into the constitutional framework and the scope of their remit (many, for example, are focused on the merits of decision-making rather than its legality). Either way, these mechanisms are largely absent from most discussions of the rule of law, which focus by and large on judicial review. But there is no legal system in the world in which the courts have the resources to review a substantial proportion of the innumerable decisions now made under statute. For example, Robert Thomas and Joe Tomlinson recently reported that “around 12 million social security decisions per year are made” by the UK Department for Work and Pensions.Footnote ³⁶ To give this some context, the Ministry for Justice reported that a total of 2490 applications for judicial review were lodged across the four quarters ending March 2023.Footnote ³⁷ This represents all such applications in England and Wales and not just those relating to the social security. Figures such as this lend empirical weight to the conceptual claim that “litigation and adjudication are always the law's Plan B. Plan A is its subjects should be guided by the law without more.”Footnote ³⁸

The emphasis on contestation is, moreover, misplaced given the nature of many decisions now made under statute. Statutes often empower executive decision-makers to grant or deny rights and benefits, depending on whether the applicant satisfies the legislatively prescribed eligibility criteria. If those rights or benefits are granted, the beneficiary will surely not contest. What is important here is that the person can know the rights or benefits that they may be available, and the steps that they must follow to obtain them. Imagine that our legal subject fails to apply for a social security benefit because they do not know they may be entitled to it. In those circumstances, the ability to contest the law would serve no purpose, for there would be nothing to contest. Many more of the decisions made under statute are of limited significance, at least when viewed in isolation. It is unlikely that a person would go to court because they have been refused a permit to cut down a tree in their backyard or park their car on the street outside their house. We would surely not describe a legal system as healthy if people were not able to find out in advance whether they were statutorily entitled to do these things without an unreasonable expenditure of resources or a trip to court.

Of course, it is important on a symbolic level that the governed can confront the government as their equal. And judicial review may have a normative impact beyond the individual cases decided.Footnote ³⁹ The claims are familiar: if government actors are aware that the decisions can be challenged in court, they will strive to comply with the law; if the public knows that they can bring such a challenge they will have greater faith in government. But these claims are very difficult to prove.Footnote ⁴⁰ Some say that it is impossible to accurately determine whether judicial review is having these effects;Footnote ⁴¹ others maintain that we can at best use empirical evidence to “speculate about the conditions and factors which mediate the influence of judicial review judgments on administrative behaviour.”Footnote ⁴² The evidence that does exist is decidedly mixed.Footnote ⁴³ Some studies indicate that judicial review has meaningfully enhanced administrative decision-making outside the courtroom in certain contexts, or instilled public faith in the legal system; others conclude that judicial review has had no effect beyond the resolution of individual cases; others again believe its broader effects are profoundly negative (for example, teaching administrators how to “review proof” their decisions without actually improving their procedure or substance). At the very least, the impact of judicial review varies considerably between jurisdictions and decision-making contexts.Footnote ⁴⁴ This is not to deny the pivotal role that judicial review plays, but rather to place it in perspective—a perspective that makes it plain that contestation is challenged by complex legislation too.

Let us assume that people can readily contest the way legislation has been applied to them after the fact, its complexity notwithstanding. Does this solve the problems posed by unknowable legislation? This, of course, requires us to consider what those problems are. Chiao claims that contestation “is capable of vindicating familiar features of the rule of law.”Footnote ⁴⁵ He describes the overarching objective of the rule of law as “taking the edge off human power.”Footnote ⁴⁶ True, this is a familiar concept, often captured in the old distinction between the rule of law and the rule of men. The idea is that the rule of law structures and confines the power of those who happen to wield it, thereby reducing the risk that it will be wielded arbitrarily.Footnote ⁴⁷ And one can easily grasp how publicly promulgated laws that people can readily contest serve this end. Even if legislation is not reasonably knowable in advance, it does provide an objective touchstone against which the legality of government action can be tested after the fact. But, as Chiao notes, what he calls “the planning conception” of the rule of law (which relevantly demands that law be reasonably knowable to the people) serves a range of more specific instrumental and normative purposesFootnote ⁴⁸—and many of them cannot be vindicated by contestation alone.

Raz, for example, says law must be reasonably knowable in advance because otherwise it cannot perform its defining purpose: to guide human behavior. This guidance function cannot be achieved by contestation because law cannot guide a person's practical reasoning if they do not know it exists.Footnote ⁴⁹ This requires ex ante knowledge of the law, of some degree and kind.Footnote ⁵⁰ Fuller emphasizes that legislation must be reasonably knowable in advance because this is the only form of law that respects the moral agency of its subjects, by affording them the dignity of choosing whether or not to obey. Thus unknowable laws are not only bad because they are unlikely to work, but because they will ultimately fray the relationship of reciprocity between law-makers and their subjects that enables legal systems to endure.Footnote ⁵¹ Contestation might contribute to these goals, but it cannot achieve them on its own. A legal system in which people are unable to know what they are allowed to do without going to court—faced with the prospect of being told, after the fact, that their actions were in fact unlawful, or else doing nothing at all—does not respect their moral agency in the way that Fuller described. It is Bentham's Dog Law, writ large.Footnote ⁵²

And so, what Chiao presents as a mere shift in emphasis within the accepted interstices of the rule of law really requires us to compromise some of its most important commitments. He does not deny that law needs to be reasonably knowable in order to work properly, or that people should be afforded the dignity of being able to know the law that binds them. Rather, he says that, at some higher level of generality, the rule of law can be vindicated by contestation alone. But contestation does not achieve the things that reasonably knowable laws do, and it does not fix the problems posed by unknowable laws.

III. People Only Need to Know Rules

Like Chiao, Rubin is anxious to ensure that our theories of legislation bear some resemblance to reality—and are reasonably consistent with what, as a matter of politics, we expect government to do. But Rubin's project is far more ambitious. In short, he seeks to show that unknowable legislation is not really a problem at all. He argues that the requirements of certainty, predictability and so forth usually associated with the rule of law are not useful standards for assessing legislation in contemporary administrative states. Insofar as people need to know the law, they can and should get the information that they need from sources besides legislative texts.

Rubin states that “legislation is an independent category that does not necessarily possess the normative force or metaphysical “kick” of law.”Footnote ⁵³ The more precise claim appears to be that most legislation does not create rules of conduct that the ordinary member of the public must obey, on pain of sanction.Footnote ⁵⁴ Rubin thinks that statutes are better described as “internal government instructions”:Footnote ⁵⁵ “public policy directives that the legislature issues to government implementation mechanisms”—that is, the courts and, more frequently, the executive branch.Footnote ⁵⁶ Even legislation that ostensibly creates rules of conduct for the public (like a statute creating a criminal offence) is more accurately described as a direction to a government entity to enforce a rule (for example, to charge, try, and punish a person who has committed that offence).Footnote ⁵⁷ Once we jettison the notion that all legislation is a way of communicating commands to the people, Rubin argues, the arguments for demanding that it be clear and comprehensible to the public fall away.Footnote ⁵⁸

Rubin also questions how we can and should make legislation knowable to the people. I take him to accept the underlying claim that the public need to be able to plan their lives within the framework of the law. However, he says that it is wrong to think we can or should achieve this by drafting legislation in a way that makes its legal effect knowable to the people. The first reason for this is that it is never the text of the statute itself, and always the actions of the “implementation mechanism,” that determines the rights and interests of the individuals. In most instances, it will be an exercise of executive power that determines these. The second reason is that legislation may well be highly indeterminate (or, as Rubin calls it, intransitive), conferring an open-ended power on the executive to decide how best to pursue the relevant policy objective. There is nothing inherently wrong with intransitive legislation. The legislature may well choose to legislate in this way if it decides it is the most effective way to instruct the executive to achieve the relevant end. And where legislation is highly intransitive, the member of the public will find out next to nothing about their legal position from reading its text.

If the legislation is highly prescriptive (or, as Rubin calls it, transitive), the person could theoretically find out what they need to know from reading its text, because the statute will stipulate in some detail the rules that the executive has to apply to a case such as theirs. But of course, the person may struggle to understand the text—for example, because it is too technical. And there is no compelling reason, Rubin argues, to compromise the efficacy of the administrative state by striving to draft statutes such as these in a way that ordinary people can understand. If the legislature wants the people to understand the legislation and how it might affect them, it should direct the executive to help them to do so, using the various tools at its disposal (“announcements, educational programs, interpretations, advice, threats, inspections, subpoenas, regulations, informal hearings, and a variety of similar techniques”).Footnote ⁵⁹

This is a powerful argument, which suggests that most writing on the rule of law tells us very little about the standards we should expect of vast swathes of contemporary legislation. We can all grasp the futility in commanding that the public follow a rule but keeping that rule secret, the injustice in punishing someone for failing to obey a rule that they could not reasonably be expected to know. But, Rubin asks us, does the public need to understand legislation that does not create rules of conduct that they must obey? Powerful as it is, I think Rubin's account too readily assumes that contemporary legislation does not make rules that the public has to obey, and too quickly assumes that knowledge of such rules is all that is required to plan one's life.

What do statutes do? Any answer to that question must be heavily qualified because statutes serve innumerable purposes and take myriad different forms. Yet, we can describe general trends. Creating rules of conduct that the public must obey is no longer the dominant function of statutes (if it ever was). But criminal lawyers in many jurisdictions have noted the proliferation of statutory offences, and the rapidity with which they are amended.Footnote ⁶⁰ Many of these are now found outside the statutes where one would find the traditional prohibitions against murder and theft—for example, in antitrust laws, corporations legislation, environmental protection legislation, local planning laws, workplace health and safety legislation, legislation that governs when people practice as doctors or lawyers or sell food or alcohol or hairdressing services, and so forth.

As that general list of subjects implies, rules of conduct are regularly used as a way of achieving broader policy objectives, whether they are backed up by criminal or civil sanction. These rules often operate in conjunction with a legislative requirement to obtain a certain kind of permission from an executive decision-maker. For example, section 33 of the Australian Customs Act 1901 (Cth) prohibits any person from moving certain goods without the requisite authorization granted under that Act. These rules differ from the classic example of the penal statute because their purpose is not really to stop people from engaging in certain activities because the legislature has decided that they are wrong, but rather to ensure that people only engage in certain activities in the way that the legislature chooses. Thus the overarching aim of section 33 of the Customs Act, we might reasonably presume, is not to stop people from moving goods full stop, but to ensure that they only move goods with the statutorily specified permit, which will only be granted when the person and the goods satisfy certain statutory specified criteria, thereby going someway to ensuring that only certain goods are moved by only certain kinds of people and in certain ways. But the result for the would-be exporter who fails to follow the law is roughly the same: they are penalized if they move goods without first satisfying the requirements that the legislature has chosen.

Beyond this, there is a plethora of statutes that do not create rules of conduct that the public must obey. But does it follow that the public need not know these—that people can plan their lives in a meaningful sense without the reasonable ability to know what these laws are and how they might affect them? Our conception of what it means to plan one's life within the law must incorporate some sense of what the law does, of the nature of the legal system in which one lives. The ability to plan is only exhausted by knowledge of rules if one adopts a very narrow conception of what that means: to live without being punished by the state. That conception of planning is particularly impoverished in contemporary administrative states, in which the state does so much more than merely prohibit and punish.

In saying this, I do not deny that arbitrary punishment is one of the greatest dangers against which the rule of law guards: first, by providing individuals with the knowledge that they need to conduct themselves in a way that will not attract the coercive power of the state and second, by ensuring that government will not exert its coercive force in circumstances not clearly authorized in advance. It is an acute example of treating people as objects rather than agents, “vulnerable to the arbitrary will of another in conditions of dependency.”Footnote ⁶¹ This is destructive of the relationship of reciprocity between legal subjects and law-makers that is required to sustain respect for law itself. But other kinds of laws can treat people in this way too, and cause similar harm. That harm may well be of a lesser degree than subjecting someone to arbitrary punishment, but over the course of the innumerable interactions between the people and the state that are now mediated by legislation, it may also accumulate.

Take the many statutes that confer contingent rights—contingent, in the sense that people can only obtain them if they satisfy the statutorily prescribed criteria and often applied in accordance with the statutorily prescribed steps. A person who fails to do this will not face any sanction: they will simply fail to obtain or validly exercise the relevant right. Thus the person who wishes to work or permanently reside in a country in which they were not born must first show that they are eligible for the relevant visa; the person who wishes to create a corporation or file for bankruptcy must follow the steps specified in the relevant legislation; the person who wishes to stipulate what will happen to their assets or their children upon their death or to transfer their house to one of those children while they are still alive will only achieve these goals if they comply with the substantive and procedural requirements set out in the relevant Acts and regulations. Consumer protection legislation is another prominent feature of the contemporary legislative landscape, which often confers rights built upon, but significantly different from, those previously recognized at common law, which only arise in certain circumstances and must be exercised in certain ways.

These examples show that many of the goals that people within our legal system may wish to pursue are themselves created by law. One can readily imagine a different kind of legal system where there are no corporate entities with a separate legal personality; no legal status of bankruptcy; no option for individuals to specify what happens to their property upon their death.Footnote ⁶² But in many legal systems, these are legally constructed options that people may choose to pursue, and legislation defines how they must do so. Imagine, for example, a legal system in which a person's property is automatically transferred to their first-born child upon their death when that child turns 21 years of age—or, in the absence of any such child, to the state. In such a legal system, the very concept of making a will would not make sense; stipulating what would happen to one's property upon death would not be a goal that anyone could reasonably pursue. Cormacain gives the example of recent legislation from the United Kingdom, which stipulates that all adults will be deemed to consent to the donation of their organs upon their death unless they opt out.Footnote ⁶³ This does not create a rule that members of the public must now follow, but it does constrain the ability of people to plan their lives (or indeed deaths) in the way they choose.Footnote ⁶⁴

There are also a great many statutes in contemporary administrative states that confer benefits—particularly social security benefits, or “welfare.” These benefits are, again, typically contingent, in the sense that legislation will prescribe when they arise and what must be done to obtain or exercise them. There is room for debate about the similarities and differences between the “administrative state,” the “welfare state,” and other styles of state in between, but the provision of social security is such a pivotal feature of many contemporary legal systems—both in terms of the size of the legislative apparatus required to do so, and the number of administrative actors employed to administer that scheme—that it should feature prominently in any account of how contemporary legislation functions.

Statutes could be designed differently. They could confer unconditional rights or benefits on everyone. They could oblige government actors to distribute those benefits without requiring members of the public to proactively apply for them. Advocates of the universal basic income have argued for just that—for reasons which include the fact that the legislative scheme required to sustain it would be far simpler.Footnote ⁶⁵ This style of legislating could work well with respect to benefits, but less so with regard to rights, which by definition must be exercised by the right-holder—and a person who does not know that they have a right, or how to use exercise it effectively, is unlikely to do this. Some knowledge of the law would also be required to exercise the right to opt out. In any event, many statutes presently place the burden on the subject to find out about them and take action accordingly, and while this remains the case, there is good reason to require that they be reasonably knowable to the people.

One might say that it is not necessary to know these benefits exist or how to obtain them, in that a person can freely pursue their chosen goals without exercising these rights or enjoying these benefits. A person can choose to attend university or have a child without accepting the financial support of the state to do so. But it is a different proposition to say that there is no harm done if a person chooses to attend university or have a child (or perhaps more realistically, chooses not to do these things) while reasonably ignorant of the statutory benefits that were available to help them pursue these goals. That assumes that our moral agency does not encompass the ability to choose whether to exercise the rights or obtain the benefits that have been provided for by law, which is a curiously hollow conception of moral agency to employ in an administrative state. It ignores the fact that these rights and benefits may well have been provided by the legislature due to evidence that it is difficult to achieve certain goals without them. If nothing else, it hinders the efficacy of the legislation, as people cannot exercise rights or obtain benefits if they do not know that they exist, or the legislative prescribed procedures that they must follow to do so—and we should assume that government has legislated these rights and benefits because it wants them to be used.

A similar argument has been made by the private law theorist Lisa Austin, who has argued that the rule of law ideals of accessibility and predictability have rightly shaped the development of the common law of property.Footnote ⁶⁶ Many of these laws confer powers on individuals, such as the power to assign an interest in land. These laws must be reasonably knowable, Austin argues, but not because people would otherwise fail to obey them—for it makes no sense to talk about a failure to obey a power. Rather, this makes the exercise of these legal powers possible. As Austin phrases it:

This facilitates the freedom to secure “a temporal horizon for action,” which “transforms abstract agents into individuals with concrete identities and abstract choices into individual life plans.”Footnote ⁶⁸ The same could be said of the rights contingently conferred by statute and, I think, the contingent benefits that so many contemporary statutes provide. People cannot obtain these rights or benefits without following the legislatively prescribed instructions. Enabling them to do so makes it possible for the law to serve its apparent purposes and, in turn, the people to pursue their own chosen plans across time.

If we do not expect legislation that confers rights and benefits to be reasonably knowable to the public, then the person who fails to exercise or obtain them because they were reasonably ignorant has no one to blame but themselves. What kind of responses might this generate in our person confounded by the unknowable law? Take, for example, a person who fails to obtain a particular benefit or exercise a particular right because they did not know it existed. This would probably generate some sense of injustice (“those with more education or the money to pay a lawyer would have been able to take advantage of this, but I missed out”). Perhaps a sense that the law is not being made to further the public good (“what is the point of conferring a right or benefit, but then making it so hard to find out how to access it?”) Perhaps ultimately, there is a sense of alienation: that they are merely an object upon which the law operates, and not a participating agent in the legal system. All of this is destructive of the relationship of reciprocity between legal subjects and legal institutions—of respect for law itself. These are the same concerns that animated many people who wrote about the rule of law and the problems posed by unknowable rules, and once we refine them to better reflect the function of contemporary legislation, we see that they are still compelling.

Besides the categories of legislation that I have described, which either do impose rules of conduct that the public must obey or confer contingent rights or benefits upon them, there are undoubtedly statutes that do not directly affect the public. There are many statutes, like those directed at specialist regulators or corporations, or those concerned with the internal structuring or budgets of executive departments or agencies,Footnote ⁶⁹ which we might say do not touch upon the lives of “ordinary people.” We might therefore accept that this category of legislation need not be reasonably knowable to the people. But this would be a far more modest claim than Rubin's, that contemporary legislation need not be knowable at all.

It may also require further refinements. As Cormacain makes clear, a model that accepts different levels of complexity depending on the intended audience of the statute faces a problem because we can never confidently predict who will engage with a statute or the level of expertise they will have.Footnote ⁷⁰ The assumption that executive actors are better placed than “ordinary people” to cope with very complex legislation is also being tested by the evolution of the administrative state and contemporary legislative practice. In some jurisdictions, many positions are now filled by workers on non-ongoing contracts who cannot easily claim deep knowledge of the legislative schemes they administer.Footnote ⁷¹ In others, there is evidence that “departments with large numbers of front-line staff delivering public services tend to have a higher proportion of junior staff.”Footnote ⁷² And the capacities of all executive actors are tested by the frequency of legislative change.Footnote ⁷³ There are also many other harms that unknowable legislation may pose, which are often left behind by focusing on the idea of “the rule of law”—for example, the ways in which complex legislation may undermine the democratic process, as the people are left unable to properly scrutinize the performance of the legislators they are constitutionally responsible for choosing. For present purposes, it suffices to say that there are good reasons why legislation should be knowable to the people, even if this does not create rules of conduct that they must obey. But does this necessarily require that people be able to understand legislative texts? This is the final claim to which I now turn.

IV. Legislative texts need not be knowable

Recall Rubin's argument that there are other, better ways of assisting people to understand how legislation may affect their legal position, besides striving to draft legislative texts in a way that is comprehensible to the general public. This, I think, is the best defence of legislative complexity. It suggests that we can have our cake and eat it too: the legislature can design legislation in a way it thinks is necessary to achieve a sophisticated array of policy goals, unhindered by any requirement to make that legislation generally knowable to the people; the public can still acquire the knowledge that they need to plan their lives within the framework of that law.

This seems theoretically sound. Most theorists agree that mere coincidental conformity with legal rules is insufficient. In order to say that a person has obeyed a legal rule, its status as a legal rule has to have some bearing on that person's reasoning. If they do x in perfect ignorance of the fact that the law requires x, they cannot really be said to be obeying the law. The reasons for this stem from notions of legal authority, and what it means to wield it.Footnote ⁷⁴ But, as Sevel has shown, the law's subjects need not acquire this knowledge from reading primary legal sources. They could be said to obey the law, in a meaningful sense of that term, even if they form their understanding that the law requires them to act or not to act from some other source— or example, a legal advisor.Footnote ⁷⁵

A different frame of analysis is required to accommodate the fact that legislation now does far more than create rules of conduct that the public must obey. I have argued that unknowable legislation is problematic, even if that legislation merely confers rights or benefits rather than creating rules of conduct that the public must obey. The reasons for that are, briefly summarized, that otherwise people are unlikely to know they have those rights, or are entitled to those benefits, and hence the law will not work; that people are thereby hindered from planning their lives within the framework of the law, in a meaningful sense; that this may in turn diminish faith in the law and its institutions, as people are made to feel like objects upon which the law operates rather than participating agents in the legal system. At the outset, I made clear that I do not seek to show the precise type or degree of knowledge that is required to satisfy the “knowability imperative” that I defend. But it seems feasible that the problems I have identified could be averted in a system where people, though unaware that any particular statute exists or of what it says, are nonetheless able to find out what their legal rights and benefits are. For example, the executive could provide information sheets to the public, explaining those rights and benefits in terms that are far easier to navigate and understand than the legislative texts themselves—just as Rubin suggests.

But executive guidance is not a benign solution. At the very least, it is not without consequence. It implicates a raft of public norms and principles, and the way legislation is designed. The reasons for this turn on a more nuanced understanding of the form of contemporary legislation, and the constitutional framework in which it operates.

Rubin made no claim about the dominant form of legislation—though he does focus on highlighting the ubiquity of “internal” and “intransitive” legislation and defending them against common rule of law critiques. Following in his footsteps, highly intransitive legislation is often said to be the norm.Footnote ⁷⁶ That may in fact be the case in the United States, where it appears that much of the legislation enacted by Congress is devoted to establishing administrative agencies and conferring their powers in broad terms, rather than specifying how they are to act. One could reasonably assume that primary legislation would take this form, and even argue that it should. As Rubin put it, “We are socially committed to the view that government agencies possess technical expertise, data-gathering ability and problem-solving capacity.”Footnote ⁷⁷ Conversely, legislatures seem to lack the time and resources that would be required to “micromanage” all the complex spheres of activity now regulated by the state.Footnote ⁷⁸ But a growing body of empirical evidence suggests that they are attempting to do so.Footnote ⁷⁹ In the United Kingdom and Australia, for example, the volume of primary legislation has expanded dramatically—perhaps not to the same extent as secondary legislation, but considerably so. Much of that legislation is highly prescriptive, specifying in crushing detail how and when the executive may act rather than directing the executive to pursue broadly stated policy goals in general terms.Footnote ⁸⁰

When legislation is intransitive and confers considerable discretion on the executive branch, the executive may issue policy statements that explain to the public how that discretion will be used. This effectively fills in the gaps left by the legislature,Footnote ⁸¹ and it is widely regarded as a desirable thing for the executive to do. But this is a different kind of guidance from that provided when the executive translates highly transitive statutes to make them easier to navigate and understand. This may be as simple as breaking down a statute that runs to thousands of pages into bite-sized pieces, so users can more readily identify the part that applies to them. Or it may comprise the translation of dense and technical statutory provisions in a style that is easier for lay people or industry to understand, perhaps bolstered by the use of practical examples. At its highest level, it may involve creating automated tools that purport to generate individualized “advice” about how complex legislation applies to a given individual or set of facts, based on the inputs of the relevant user. All three are evaluative: they require the executive to identify the relevant legislative content and, to varying extents, repackage it in different words—or, in the case of automated systems, code. And as with any translation, there is a real prospect that meaning will be lost or changed.

The extent to which the public relies on executive advice about the law is often overlooked. Others have noted that the complexity of contemporary law forces the public to rely increasingly on expert advice, but they tend to refer only to the advice of lawyers.Footnote ⁸² Large corporations and other sophisticated actors will no doubt employ teams of lawyers to help them understand how to comply with complex legislative schemes, and the proliferation of statutory offences has increased the need for legal advice in important ways. But few members of the public will consult a lawyer to help them understand the innumerable statutes that regulate their daily lives: whether they are entitled to unemployment benefits, or how to get a permit to park a skip bin outside their house, or whether they are entitled to return a television that has stopped working for a refund. That is partly because the occasions on which a person's life is shaped by legislation are so frequent that no one could plausibly obtain legal advice every time, partly because the cost of legal advice would probably outweigh the value of the potential benefit and partly, I think, because the executive now provides a lot of information about this kind of thing for free.

Rubin would probably argue that this is exactly as it should be. And of course it is not an entirely new phenomenon. It is unlikely that members of the public routinely read legislation in the past, particularly given it has only recently been digitized and made available online.Footnote ⁸³ It is likely that people have always relied, to varying extents, on other sources of information—including the advice provided to them by the executive branch. But two things do seem different now. One is that the reliance on alternative sources of information is now unavoidable. Legislation is now more accessible, but even those who are minded to read it will likely lack the skills and resources they need to identify the relevant parts, or to understand it, and few will be able to assess for themselves whether what the executive has told them about that legislation is accurate. The second is the disruptive influence of technology. The executive is not providing all of this information in person. It is providing automated systems—apps, tools, calculators, chatbots—that assist users to navigate complex legislative schemes.

Like the automation of executive decision-making, the automation of information is likely to have profound implications, most of which we can only guess at, not least because there is very limited information publicly available about how these automated systems themselves are designed. If nothing else, this is likely to contribute to the demand to make more legislation code-like, which I explained in Part I.

A legal system that forces the public to rely upon the executive's interpretation of the law—or, we might say, relies upon the executive to fulfil the crucial function of making the law reasonably knowable—finds itself in a difficult position. This is a system in which the executive wields extensive de facto authority to say “what the law is.” But in many common law systems, the executive's interpretation of legislation carries little to no legal weight. The courts do not defer to those interpretations and members of the public who rely upon them will have limited legal recourse if it turns out that the executive's interpretation of the legislation was wrong, or it changes its mind. This creates a real risk of unfairness, if not arbitrariness: the executive's word has force de facto, but not de jure.Footnote ⁸⁴ If nothing else, it simply diminishes the utility of the guidance that the executive can provide, for that guidance is always defeasible as a matter of law.

The courts could adopt a different approach, concluding—out of concerns of fairness to those who have relied on the executive's advice, or that the law should speak with one consistent voice—that they should defer to the executive's interpretation of the law, or fashion some doctrine of administrative estoppel that gives it legal force. The courts have done something like this in the United States, at least in the past—particularly via the doctrine of Chevron deference.Footnote ⁸⁵

But Chevron was recently overruled, in the case of Loper Bright Enterprises v Raimondo. Footnote ⁸⁶ It is beyond the scope of this article to venture an argument as to where that was right or wrong. It suffices to say that judicial deference to the executive's interpretation of legislation is constitutionally contentious. It jars with the fundamental constitutional idea that the executive is subordinate to the legislature and should not be able to conclusively decide the limits of its own powers. Courts in many common law jurisdictions have recently emphasized that judicial control over questions of statutory meaning is essential, not only to ensure the supremacy of parliament, but to guard against the fragmentation of the law; to ensure that everyone is bound by the same law as interpreted by an independent court.Footnote ⁸⁷ Widespread reliance on the executive's interpretation of legislation puts the legal system between a rock and a hard place: accept that doctrine does not reflect the way people really interact with legislation, and risk unfairness to those who rely on what the executive says; or compromise the constitutional idea that the executive is subordinate to primary legislation. The point is to pause and reflect on the directions in which these trends are leading, on the potential implications of accepting and adapting to legislation that is too complex to be reasonably knowable, rather than resisting it.