Introduction
John Austin famously claimed that law is the command of the sovereign backed by threats of sanctions. Few, if any, now think Austin was right: laws are not plausibly seen as commands. Nonetheless, the idea that legal systems threaten citizens with sanctions has not only survived Austin but is taken by most legal and political philosophers as “reasonably uncontroversial,”Footnote 1 “unsurprising,”Footnote 2 and as portraying “a large part of how law operates.”Footnote 3 Despite its wide acceptance, this idea has never been carefully developed, let alone criticised. This paper takes up this task. My main claim is that contrary to what is usually thought, typical legal systems rarely threaten citizens.
I start by clarifying what it means for typical legal systems to threaten citizens and showing that to discuss this claim we need: (i) a list of types of actions performed by typical legal systems that could plausibly be seen as threats; and (ii) a sound account of threats. I offer (i) in the first section and (ii) in the second. My account of threats, however, is deliberately incomplete; it only contains elements that bear on the paper’s core discussion. The third section is where I advance my main claim. There I show that our legal systems rarely satisfy a necessary condition of conditional threats. The conclusion presents some implications of my main claim to core discussions in legal and political philosophy.
1. The Threats of Legal Systems
The idea that typical legal systems threaten citizens is not yet as clearly articulated as it should be. To avoid ambiguities, and to ensure that the idea is philosophically interesting, I’ll stipulate that:
To say that a legal system threatens citizens means that legal authorities, by performing some actions, conditionally threaten citizens in the name of the system.
According to this stipulation, assessing whether typical legal systems threaten citizens involves nothing more than assessing whether the relevant actions performed by legal authorities in the name of the system count as conditional threats. Therefore, to proceed with the discussion we will need: (i) a list of actions performed by legal authorities in the name of the system that could plausibly be viewed as conditional threats; and (ii) an account of conditional threats. I provide the list in this section and a partial account of conditional threats in the second section.
Two caveats: first, the discussion solely concerns conditional threats that legal authorities make in the name of the legal system. I’ll assume that an authority makes a conditional threat in the name of the legal system only if the relevant announcement (or speech-act, if you like) is made in the exercise of an authority’s role and the legal authority is legally permitted and empowered to make it. This qualification excludes from consideration those actions of legal authorities that cannot be plausibly seen as actions of the legal system.
The second caveat is that I’m only interested in those actions that authorities of typical legal systems perform in the name of their respective systems. For the present purposes, typical legal systems are those legal systems that currently exist in western democracies and those that closely resemble them.
Different types of legal authorities perform different types of actions in the name of the system. When presenting a list of the actions relevant for the discussion, it thus makes sense to consider the different types of authorities and their respective actions separately, if only for presentation purposes. In my list below, I include actions performed by law-makers, law-enforcers, and law-appliers.
1.1. Law-Makers
The idea that legal systems conditionally threaten citizens is most commonly associated with the activities of law-makers. They are, after all, those who perform (and have the power and permission to perform) the following actions:
(t1) issuing laws that prescribe sanctions or enforcement mechanisms.
(t2) issuing laws that criminalise conduct.
(t3) issuing laws that authorise coercive enforcement mechanisms.
Most legal and political philosophers who have claimed that legal systems conditionally threaten citizens have used the actions in t1, t2, or t3 to illustrate their claims. Examples where this happens are myriad. Gerald Gaus, for instance, claims that “[w]hen the state issues a law, it typically commands citizens to act and, further, threatens them with fines and imprisonment unless they obey.”Footnote 4 James Edwards, in his turn, reports that many philosophers of criminal law think that “to criminalize ϕing is, inter alia, to threaten to punish those who ϕ.”Footnote 5 And Kenneth Himma claims the following in relation to t3:
[t]he authorization of coercive enforcement mechanisms, of course, constitutes the characteristic manner in which legal systems make threats as a means of inducing certain acts or omissions. The authorization of coercive enforcement mechanisms is, thus, analogous to the gunman’s threat of violence.Footnote 6
1.2. Law-Enforcers
The police are a typical example of a law-enforcement body. At face value, it is plausible to view the police as conditionally threatening citizens in the name of the system when one of their members enforces the law. ‘Stop, or I’ll shoot!’, uttered by a police officer, is probably the paradigmatic example where a conditional threat seems to be issued by a law-enforcer in the name of the system. By uttering these words, the police officer does at least one thing: announces that an enforcement action—shooting—will be performed unless the order is obeyed. Thus, there seems to be some room for claiming that law-enforcers conditionally threaten citizens by announcing that unless citizens ϕ, enforcers will perform an enforcement action.
But a qualification is needed. Law-enforcers sometimes have the legal duty to perform a law-enforcement action, and sometimes they only have discretionFootnote 7 to do so. As I hope it becomes clearer later in the paper, it is worth distinguishing the following two actions for the purposes of discussion:
(t4) announcing that unless citizens ϕ, law-enforcers will perform an enforcement action that they have the duty to perform.
(t5) announcing that unless citizens ϕ, law-enforcers will perform an enforcement action that they have discretion to perform.
Law-enforcers could be seen as conditionally threatening citizens by performing yet another kind of action in the name of the system. In perhaps all typical legal systems, those who disobey law-enforcers’ orders are liable to a sanction or to another unwelcome measure. For instance, generally those who resist an arrest are liable to a sanction. Besides, resisting an arrest (or another enforcement act) may trigger a permission for an enforcer to use force. For that reason, one might think that by making a statement such as ‘You are under arrest!’ in a context where it is both a punishable offense to disobey orders of authorities and there are norms permitting and empowering authorities to sanction those who disobey their orders, the law-enforcer implicitly communicates that if citizens disobey the enforcer’s order, a sanction or enforcement mechanism will be employed. It seems, thus, that one could claim that law-enforcers conditionally threaten citizens by:
(t6) issuing orders in a context where it is a punishable offense to disobey the orders of legal authorities and where there are norms permitting and empowering authorities to sanction those who disobey their orders.
1.3. Law-Appliers
Judges are the paradigmatic law-appliers. It is not easy to find examples in the literature where judges are portrayed as conditionally threatening citizens.Footnote 8 But there is some room to construe them as doing so. In passing a sentence, judges typically require an offender to pay a fine, or meet the terms of a probation order, or perform community services, and so on. On top of that, judges typically demand (even if tacitly so) suitable behaviour in court.
As with the orders given by law-enforcers, the requirements made by law-appliers typically take place in a particular context. Namely, in a context where those who disobey their orders commit an offense which makes them liable to sanctions or enforcement mechanisms, and where there are authorities empowered and permitted to bring about the sanctions and enforcement mechanisms (the relevant authorities may or may not be the law-appliers themselves).
For that reason, one may be inclined to think that, when made in this particular context, law-appliers’ demands are followed by the tacit message that further consequences will be applied (by the judge or someone else) in the event of misbehaviour. Therefore, one could hold that, in addition to law-enforcers, law-appliers could also be viewed as conditionally threatening citizens by performing the type of action described in t6.
One could also think that law-appliers conditionally threaten citizens by announcing that they will (or may) perform some law-applying actions. Law-appliers apply laws that prescribe sanctions and enforcement mechanisms (those laws that law-makers issue when performing the action in t1). It could be thought that a judge conditionally threatens a citizen by, e.g., announcing that if someone on probation misbehaves, the judge will—by applying a given law—sentence them to jail.
The same qualification added earlier in relation to law-enforcers’ actions is relevant here. Sometimes judges have a duty to apply these laws; sometimes they have discretion to do so. For the purposes of discussion, I’ll consider the following two actions separately:
(t7) announcing that unless a citizen ϕs, the law-applier will apply a law that prescribes sanctions or enforcement mechanisms that they have a duty to apply.
(t8) announcing that unless a citizen ϕs, the law-applier will apply a law that prescribes sanctions or enforcement mechanisms that they have discretion to apply.
That is the list. I now turn to conditional threats.
2. Conditional Threats
Most philosophers accept that A conditionally threatens B only if:
-
(1) A communicates to B that unless B does y, x will (or may) be brought about.Footnote 9
Most, if not all, also accept that a successful account of conditional threats must be capable of distinguishing conditional threats from conditional warnings, especially from conditional warnings about one’s own prospective action. The distinction is important because while threats carry a presumption of wrongfulness (and a burden of justification) warnings do not—or so it is often assumed.Footnote 10 In fact, some have even argued that a distinction between these two categories of speech acts is only worth drawing when relevant for debates about the permissibility of such forms of communication.Footnote 11
With this in mind, my main goal in this section will be to present a further necessary condition of conditional threats that suffices to distinguish them from conditional warnings in a way that is consistent with the assumption that threats, but not warnings, carry a presumption of wrongfulness. This necessary condition will be crucial for section 3, where I argue against the view that legal systems often conditionally threaten citizens. To do so, I’ll develop a point originally made by Kent GreenawaltFootnote 12 —and accepted by philosophers such as Alan WertheimerFootnote 13 and William Edmundson.Footnote 14
Greenawalt thought that a criterion that distinguishes conditional threats from conditional warnings about one’s own prospective action (henceforth ‘conditional warnings’) could be extracted from the analysis of a hypothetical example. Before presenting the example, let me just add a caveat about terminology. Greenawalt uses ‘manipulative threats’ and ‘warning threats’ to refer to what I am respectively calling ‘conditional threats’ and ‘conditional warnings.’ This difference in terminology might give the impression that, contrary to what I’m suggesting, Greenawalt was not concerned with the distinction between threats and warnings, but with the distinction between different sub-categories of threats. However, this terminological battle leads us nowhere and doesn’t do justice to the fundamental issue at stake: that Greenawalt was ultimately concerned with distinguishing ‘coercive’ from ‘non-coercive speech’ and clearly presented what he calls ‘warning-threats’ as an instance of the latter. Provided that the sort of speech-act Greenawalt calls ‘warning-threats’ and I call ‘conditional warnings’ are the same phenomenon (which, by stipulation, they are)Footnote 15 and are instances of ‘non-coercive speech’ (which is precisely what Greenawalt claims), nothing substantively hangs on whether Greenawalt saw ‘warning-threats’ as a subcategory of warnings or as a sub-category of threats.
Terminology apart, here is the example which Greenawalt uses to distinguish conditional threats from conditional warnings and which I’ll call ‘Abortion’ for convenience’s sake:
[Abortion]: Tim wants to make love with Vicki.… Tim decides that success will depend on frightening Vicki. He knows that Vicki has recently had an abortion, that she cares a great deal about her parents’ opinion of her, and that her parents regard abortion as a terrible wrong. Tim threatens that he will inform Vicki’s parents unless she makes love with him, and she complies.… Tim is neither a devoted opponent of abortions, ready to expose all those who have them, nor a believer in parents knowing everything their mature children are up to. Tim would not even think of reporting the abortion to Vicki’s parents in the ordinary course of events, and further would not even think of that as an apt response if, absent an initial threat, Vicki had rebuffed his amorous advances.Footnote 16
And here is an important point Greenawalt makes after presenting Abortion:
(a) If Tim had for some reason only been able to communicate his wish to make love but not the threat, his reasons for telling her parents would never come into existence.Footnote 17
Greenawalt then takes what he says about Abortion and extends it to conditional threats in general:
(b) I shall use the term “manipulative threats” to refer to threats of action when the actor would not suggest or engage in the action were it not for the threat itself and its linkage to a demand.Footnote 18
What Greenawalt says is not entirely clear. In fact, what he says in (a) seems to be different from what he says in (b). In (a), Greenawalt seems to be claiming that the existence of Tim’s reasons for telling Vicki’s parents about her abortion depend on the announcement, i.e., Tim’s reasons are what I will call ‘announcement-dependent.’ The passage also implies that Tim had no prior, announcement-independent, reason for telling Vicki’s parents about her abortion. Which suggests that, for Greenawalt, an announcement is a threat only if the threatener does not have an announcement-independent reason to carry out the announced consequence and the announcement creates new, announcement-dependent, reasons for the threatener to carry out the announced consequence. More schematically, Greenawalt could be construed as claiming that A conditionally threatens B only if:
(2a) A has an announcement-dependent reason to bring about x (or to have x brought about) and A has no announcement-independent reason to bring about x (or to have x brought about).
In (b), however, Greenawalt doesn’t seem to be making a point about the threatener’s reasons. The phrase ‘would not suggest or engage’ could be interpreted as indicating that the threatener’s motivation or conditional intention to engage in the threatened action depends on the communication of the threat, i.e., that the threatener has become motivated or adopted a conditional intention at least partly because the threat was issued. Greenawalt’s description of Abortion is consistent with this interpretation. In fact, this reading has some uptake in the literature. For example, while discussing Abortion and Greenawalt’s view, Ben Sachs claims that “before Tim is able to communicate the conditional intention to tell Vicki’s parents of her abortion, he does not have that conditional intention.”Footnote 19 Sachs’ main point is that when an announcement is a conditional threat, the communicated conditional intention (i.e., the intention to tell Vicki’s parents about the abortion unless she has sex with him) must not have been adopted by the announcer before announcing it. Thus, for Sachs (and according to this reading of Greenawalt), A conditionally threatens B only if:
(2b) A does not have the conditional intention to [do x unless B does y] before making the announcement.
Despite obvious differences, both (2a) and (2b) call our attention to an important point about the distinction between threats and warnings. The point is that by issuing a conditional warning about one’s own prospective action, the announcer primarily—and, in some cases, exclusively—aims to inform the addressee of something that the announcer had prior reasons to bring about or was committed to making happen. Warnings, in this sense, just alter the addressee’s “perception of her environment,”Footnote 20 her epistemic position. Threats, on the other hand, do not just alter the threatenee’s epistemic position, they are also “situation-altering,”Footnote 21 in the sense that the communication of the threat “sets in place the conditions for the unpleasant consequences of failure to comply”Footnote 22 with the threatener’s demand.
To illustrate, compare Abortion with another case proposed by Greenawalt, which I will call ‘Snitch’:
[Snitch]: Tim, disturbed by the fact that Vicki is selling drugs in her apartment, decides that he will inform the police if that is necessary to get her to stop. He tells her that she had better stop or he will do so. Afraid of being arrested, she stops.Footnote 23
Tim’s announcement in Snitch looks like a warning rather than a threat. For Greenawalt, the reason why this is so is that “[i]n contrast to the manipulative threat, the prospect of the [announced] action here has reality apart from Tim’s utterance of the threat to take it.”Footnote 24 This suggests that Tim’s reason, motivation, or intention to inform the police doesn’t depend on his making the announcement.Footnote 25
At face value, this distinction between conditional threats and conditional warnings extracted from Greenawalt’s discussion is appealing—and so are (2a) and (2b). In fact, we can infer from how people talk that they implicitly grasp this distinction between threats and warnings and use certain phrases to portray themselves as issuing warnings. For example, sometimes warnings are preceded or followed by expressions that suggest that the warn-er has announcement-independent reasons or has already adopted the intention to do what they have warned. Expressions such as ‘I’m just telling you,’ ‘my hands are tied,’ ‘if you do this, I will have to do that’ are examples. A professor who, in the course of explaining the marking criteria, announces to their students ‘Unless you answer all questions in the exam, I’ll have to fail you’ is certainly just warning the students rather than threatening them. That the professor’s announcement does not ‘set in place the conditions’ for their failing the students explains why this is so.
Despite their initial plausibility, there are reasons to reject both (2a) and (2b). Consider the following scenario:
[Kidnapper]: K kidnaps L’s son. K tells L that L’s son will be released from captivity only if L does the following: approaches M and threatens to break M’s arm unless he gives up his wallet; breaks M’s arm in the event he doesn’t give up his wallet. L believes that complying with K’s directives is the only way to save his son from captivity. L finds M and announces: ‘Give me your wallet, or I’ll break your arm.’
L has certainly threatened M. But neither (2a) nor (2b) would allow us to say so. That is because L’s reasons and intention to break M’s arm have ‘reality apart’ from L’s announcement. After all, prior to making the relevant announcement, L had reasons and the intention to do whatever was necessary to save his son from captivity, including breaking M’s arm if M didn’t give up his wallet. Thus, Kidnapper shows that (2a) and (2b) are too demanding: a conditional threat can be issued even when the threatener has announcement-independent reasons or has adopted the conditional intention to perform the announced consequence.Footnote 26
A possible way to circumvent the kidnapper objection would be to simply drop the requirement that the threatener must lack announcement-independent reasons (or the intention) to perform the announced consequence. We could, along these lines, hold that A conditionally threatens B only if:
(2c) A has an announcement-dependent reason to bring about x (or to have x brought about).
This amendment might suffice to circumvent the kidnapper objection. Here is how: if we look closely at Kidnapper, we can see that L does have an announcement-dependent reason to break M’s arm. After all, by making the announcement, L has indeed acquired a reason to break M’s arm so as to stick to his word. This reason is certainly weak and most likely outweighed by other reasons that L had. However, it is still an announcement-dependent reason. The kidnapper objection, therefore, seems not to work against (2c).
Though (2c) can overcome the kidnapper objection, it seems to suffer from the opposite problem of (2a) and (2b): while (2a) and (2b) were too demanding—and underinclusive as a result—(2c) is too permissive, and therefore overinclusive. To see why, let us come back to Snitch, a case Greenawalt, but also Ben Sachs and others,Footnote 27 have taken as a clear example of a warning rather than a threat.
In Snitch, Tim tells Vicki that she’d better stop selling drugs in her apartment or he will inform the police. Because Tim had decided to snitch in order to stop Vicki from selling drugs before making the announcement, Greenawalt (and others) assume that “the prospect of the [announced] action here has reality apart from Tim’s utterance.”Footnote 28 In other words, it is assumed that Tim’s announcement does not give rise to new, announcement-dependent, reasons to snitch on Vicki. But this assumption is mistaken. By making the announcement, Tim seems to acquire at least one (perhaps weak) announcement-dependent reason: a reason to stick to his word so as to save his face, as it were.
This is not a peculiarity of Snitch. This kind of announcement-dependent reason to save one’s face emerges in other examples that we might want to classify as warnings rather than threats. Think, for instance, of the earlier example of the professor who announces that they will have to fail students who do not answer all the exam questions. By uttering these words, the professor does acquire an announcement-dependent reason not to go back on their word (and maintain their authority and reputation). But it would be forceful to see the professor as threatening the students.
Greenawalt (and others) seem to have unveiled an important and plausible difference between threats and warnings. In some sense, warnings merely affect the target’s epistemic position, whereas threats are ‘situation-altering’ in a more robust sense. However, as may be clear from the discussion, it is not easy to pin down exactly how to cash out this difference. The three alternatives explored so far—(2a), (2b), and (2c)—are unsatisfactory. In the next subsection, I will propose a way to overcome the identified problems while retaining Greenawalt’s core insights.
2.1. Conditional Threats: Primary and Announcement-Dependent Reasons
One way to preserve Greenawalt’s insight that threats are situation-altering and at the same time circumvent the problems faced by (2a), (2b), and (2c) is to identify the kind of intention threateners must display and to better qualify the idea that threats give rise to announcement-dependent reasons. For this we must bring in another (intuitive) feature of threats: the fact that they are action-inducing.
When I say that threats are action-inducing, I mean that issuing a conditional threat involves more than just presenting the threatenee with a choice between a set of alternatives. By issuing a threat, the threatener either intends to get the threatenee to act a certain way or intends the threatenee to see them as someone who is attempting to use the announcement as a means to get the threatenee to act a certain way. Hence, the mere fact that the threatener has announced that the non-satisfaction of their demand will be followed by unwelcome consequences is (or is supposed to be seen by the threatenee as) the primary reason for bringing about the announced consequence; a reason that is more salient or weightier than the threatener’s underlying reasons or motives for (or against) bringing about such consequence. This suggests that the threatener either has or intends to be perceived as having what I will call a ‘primary announcement-dependent reason’ to bring about the announced consequence.
To illustrate, compare a typical threats-case (e.g., ‘Give me your wallet, or I’ll kill you’ uttered by an armed street robber to a passer-by) with a warning-case (e.g., the earlier example featuring the Professor, or Snitch). In the threats-case, the robber’s underlying reasons for bringing about the announced consequence are not supposed to be seen as the main reason for bringing about the announced consequence. In fact, bringing these underlying reasons forward can even compromise the felicity of the threat and its ability to induce action (suppose, e.g., that the robber let it slip that they plan to kill the victim anyway).
Warnings work differently. While some warnings about one’s own prospective action may give rise to announcement-dependent reasons to bring about the announced consequence, such announcement-dependent reasons are neither primary nor intended to be seen as such. Evidence of this is the fact that by specifying or making the underlying reasons more salient, the warner may actually reinforce or clarify that a warning, and not a threat, is being made. For instance, in the case of the Professor, we’ve seen that the announcement—‘Unless you answer all questions in the exam, I’ll have to fail you’—gives rise to announcement-dependent reasons. But the phrase ‘I will have to fail you’ already hints that the underlying reasons are the salient and weightier ones: they are what would justify (or motivate) the Professor’s act of bringing about the announced consequence. In fact, the Professor could have made that clearer by, for example, adding something like ‘those are just the rules’ after the announcement without compromising the felicity of the warning. A similar reasoning applies to Tim’s announcement in Snitch and other clear cases of warnings.
Now, I have suggested that threats may involve the threatener’s intention to be perceived as having a primary announcement-dependent reason to bring about the announced consequence. And it is the intention to be perceived which allows us to explain why threats can still be issued in (perhaps less typical) cases where the threatener’s announcement-dependent reason is not actually primary.
Recall Kidnapper: in that case, the threatener—L—could not have taken the chance of being perceived as not issuing a threat, for that would have resulted in the death of his son. If L was serious about saving his son from captivity, L must have intended that his announcement, and not his underlying reason to save his son from captivity, was perceived as the primary reason for breaking M’s arm. Had L given away his underlying reasons—by, for example, saying ‘This is the only way to save my son: give me your wallet or I will have to break your arm’—we would question whether L was complying with the kidnapper’s demand, as the utterance could suggest that L did not really intend to be seen as threatening M, but rather as pleading or warning M. Hence, even though we may doubt whether L had a primary announcement-dependent reason to break M’s arm, to make sense of this case and of the fact that the father intended to comply with the kidnapper’s demand, we must assume that L intended to be perceived as having such reason.Footnote 29
With these qualifications in hand, here is a summary of my proposal for necessary conditions of conditional threats:
A conditionally threatens B only if:
-
(1) A communicates to B that unless B does y, x will (or may) be brought about;
-
(2) A has or intends that B believes that A has a primary announcement-dependent reason to bring about x (or to have x brought about).
Nothing that has been said up until now affects the plausibility of condition (1). Its formulation encompasses both verbal and non-verbal threats, and it allows one to issue a threat that will be carried out by a third party. As can be inferred from the preceding discussion, the proposed account is capable of distinguishing conditional threats from conditional warnings. I have also hinted that the present account vindicates two common intuitions about conditional threats: the intuition that conditional threats are situation-altering and the intuition that conditional threats are action-inducing. These explanatory virtues strike me as giving us independent reasons to adopt my proposed account of conditional threats. Of course, much more could be said about each element of my account. But in view of the main goal of the paper, I’ll leave this to another occasion and move on to discuss whether typical legal systems in fact conditionally threaten citizens.
3. Threatened by Law?
I am now in a position to assess if legal authorities typically conditionally threaten citizens in the name of the system by performing the actions listed in the first section. For ease of exposition, I will not strictly follow the order in which the actions have been first presented. Instead, I will organise the discussion around four themes and fit the relevant actions into them.
3.1. Threats by Law-Makers
(t1) issuing laws that prescribe sanctions or enforcement mechanisms.
The thought is that by issuing a law that prescribes a sanction or an enforcement mechanism, law-makers communicate that citizens who perform a particular action will (or may) be subjected to a sanction or to an enforcement mechanism. This announcement, the argument continues, amounts to a conditional threat. In order to facilitate discussion, let’s assume that law-makers have issued (L), which is formulated as follows: ‘If citizens steal, then citizens shall be imprisoned for three to five years.’Footnote 30
To evaluate whether issuing (L), or any other sanctioning law, amounts to a conditional threat we must first understand what it is to issue a law. When we say that law-makers have issued a law we might mean three different things: (i) that law-makers have enacted a law; (ii) that law-makers have promulgated a law that has been previously enacted; or (iii) that law-makers have both enacted and promulgated a law.Footnote 31
Because threatening involves an announcement, we should not read ‘issuing a law’ as enacting a law. Enactment is the act by which laws become part of the legal system. It is an act that law-makers typically perform by voting. By merely enacting a law such as (L), law-makers would not necessarily announce to the public that if they steal, they would be imprisoned. This announcement (i.e., promulgation) is a separate act; an act which can even fail to take place.Footnote 32
Though we can conceptually separate enactment from promulgation, in practice these two acts go hand in hand. In typical legal systems, laws are fully enacted—and hence can change citizens’ and authorities’ legal positionsFootnote 33 —only if a series of procedural and substantive constraints are met. Among such constraints number rule-of-law considerations of publicity. So, as a matter of fact, in a typical legal system the enactment of a law is usually conditioned by its promulgation: a law is considered fully enacted only after its promulgation.
For the purposes of the current discussion, it then makes sense to see law-makers’ act of issuing a law as a complex act which involves promulgation and, therefore, an announcement directed towards citizens. Hence, when law-makers issue (L), they both make it the case that (L) is part of the system and announce to citizens that if they steal, they shall be imprisoned for three to five years.
The upshot is this: if issuing a law like (L) were to count as a conditional threat, law-makers must have intended that citizens believe that the fact that (L) was announced (i.e. promulgated) was the primary reason for the relevant authorities to imprison those who steal. But that condition doesn’t seem to be often satisfied in typical legal systems.
As a rule, law-makers of typical legal systems are required to justify both the need for a certain law and the prescription of particular sanctions and enforcement mechanisms. Perhaps in all contemporary democratic jurisdictions, law-makers face stringent constitutional constraints to enact sanctioning laws. For example, law-makers are not usually permitted to enact (and promulgate) laws the contents of which involve the imposition of unmotivated, disproportionate, or unfitting sanctions or enforcement mechanisms. Thus, to issue a law like (L), a typical legal system would require law-makers to give prior justification for subjecting those who steal to imprisonment.
Given this, it would be odd, to say the least, if by issuing a law like (L), law-makers intended that citizens believed that the primary reason for imprisonment was the fact that (L) was announced. If we are to believe that law-makers intend to be seen as doing their job and respecting their constitution, we must assume that the fact that (L) was announced is meant to work as a mere enabling condition. The announcement (i.e. promulgation) merely removes what would otherwise be an obstacle to doing what the relevant authorities have prior reasons to do; it makes it permissible to imprison citizens who steal but it is not a primary reason to do so (it might not even be seen as a reason).Footnote 34
It is therefore more fitting to construe law-makers who issue sanctioning laws like (L) as intending that citizens believe that the primary reasons for bringing about sanctions and enforcement mechanisms are those underlying reasons which justify the relevant sanctions and enforcement mechanisms in the first place; reasons law-makers ought to have attended to at the time of drafting and voting. This suggests that by promulgating a law like (L), law-makers are simply changing citizens’ epistemic position; they are communicating to citizens that they will imprison those who steal primarily because of the existence of (L) (and its underlying justification) and not because (L) has been promulgated. Therefore, if I’m correct, law-makers shouldn’t be seen as threatening citizens by issuing laws that prescribe sanctions and enforcement mechanisms. In most circumstances, law-makers merely warn citizens by performing actions that fit the description of t1.
Some, however, may dispute this claim. Some criminal laws that impose sanctions have a strong deterring component attached to them. The language in which criminal provisions are formulated in a variety of jurisdictions sometimes corroborates that.Footnote 35 They are issued, one may argue, primarily to deter criminal behaviour. And because deterrence heavily relies on the announcement (i.e. promulgation), one might think that law-makers intend to be perceived as having primary announcement-dependent reasons to bring about those sanctions that are prescribed mainly to deter criminal behaviour. And if that is the case, then it seems that we are back to threats.
Two things can be said in reply. First, as some criminal law theorists have pointed out, the widespread presence of ‘deterrence speak’ is not sufficient to conclude that deterrence is actually being used as a reason to issue particular criminal laws and to bring about the sanctions or enforcement mechanisms attached to them. On many occasions, as Paul Robinson puts it, deterrence speak is just used as “fabricated arguments to give a deterrence justification for a formulation [of a law] based, in truth, on other considerations.”Footnote 36 And we know that because if deterrence considerations were actually relied upon, those criminal laws that are allegedly issued primarily on the basis of deterrence would receive a completely different formulation.Footnote 37 Second, and more importantly, one need not assume that law-makers threaten citizens in order to account for the fact that deterrence is often an important part of the reasons why sanctions and enforcement mechanisms are performed and applied. Warnings can perform a deterrence function just as well as threats. Warnings, after all, convey information that may influence the addressee’s expectations about the likelihood of the occurrence of the warned event and can serve to guide the addressee’s behaviour.
Notice, however, that the existence of cases where law-makers threaten citizens by issuing laws that prescribe sanctions and enforcement mechanisms is compatible with the main claim I’m advancing here: that law-makers of typical legal systems typically do not threaten citizens by issuing sanctioning laws. These cases are not unfamiliar. Sometimes, from ulterior motives or gross negligence, law-makers fail to respect the constitutional provisions that are contrived to limit their law-making power and ensure that they pass legislations that they have reasons to pass. We all know a number of cases in contemporary jurisdictions where legal authorities apply disproportionate, cruel, pointless, or arbitrary sanctions and enforcement mechanisms primarily because laws with that content have been publicised (regardless of whether the publication was lawful or not).
Be this as it may, it is worth asking why these cases are familiar to us. They are familiar, I submit, not in virtue of reflecting what usually happens when law-makers issue laws that prescribe sanctions or enforcement mechanisms. Instead, they are familiar precisely because they are a glitch in the way law-makers are supposed and expected to operate. They are the cases worthy of academic debate and press scrutiny; cases where we can question whether law-makers are really acting in the name of the legal systems and in accordance to the constitutional framework adopted in their polity. In fact, those who lean towards strong natural law conceptions would perhaps even fail to recognise the laws promulgated under these conditions as laws.
(t2) issuing laws that criminalise conduct.
By issuing laws that criminalise conduct law-makers seem to be solely conferring criminal status on an action and announcing this fact. This doesn’t amount to conditional-threatening: an action-token can have criminal status without there being any consequence attached to those who engage in it.
In typical legal systems, however, laws that criminalise conduct are usually issued in a context where authorities are both empowered and permitted to apply sanctions or enforcement mechanisms. Besides, laws that criminalise conduct are usually followed by laws that prescribe sanctions and enforcement mechanisms. Perhaps because criminalisation in paradigmatic legal systems occurs in this particular context, some philosophers of criminal law tend to say that, “to criminalize ϕing is, inter alia, to threaten to punish those who ϕ.”Footnote 38 These philosophers might use ‘criminalise’ and related terms to refer to something broader than ‘conferring criminal status on an action.’ They might think that criminalisation is a process that finishes only when law-makers give publicity to (i.e. promulgate) a law that prescribes a criminal sanction to those who engage in the criminalised conduct. And by promulgating such a law—they may add—law-makers conditionally threaten citizens. Their claim, in other words, could be reconstructed as the claim that the process of criminalisation always involves a law-maker issuing a conditional threat.
This conception of criminalisation strikes me as incorrect.Footnote 39 Nevertheless, if this conception turns out to be correct, then the discussion about t2 would be absorbed by the previous discussion about t1. And as I have argued in relation to t1, law-makers conditionally threaten citizens by issuing laws that prescribe sanctions or enforcement mechanisms in a much more exceptional fashion than it is generally assumed.
(t3) authorising the use of coercive enforcement mechanisms.
Recall Ken Himma’s claim that “[t]he authorization of coercive enforcement mechanisms is … analogous to the gunman’s threat of violence.”Footnote 40 To begin, we need to see what the authorisation of the use of coercive enforcement mechanisms amounts to. When law-makers authorise the use of coercive enforcement mechanisms, they grant permission for legal authorities to use coercive enforcement mechanisms. This could be a general permission: ‘coercive mechanisms are permitted in this jurisdiction from now on.’ This doesn’t seem to be the sort of authorisation Ken Himma is concerned with. He seems, I think, to be concerned with situations where law-makers issue a law that grants a certain group of authorities the liberty to e in the event citizens ϕ, where ‘e’ stands for a particular coercive enforcement mechanism.
Contrary to what Himma claims, there seems to be a dis-analogy between the traditional scenario in which the gunman conditionally threatens the passer-by with violence and what law-makers do when they authorise the use of coercive enforcement. The gunman’s scenario doesn’t involve the gunman announcing that someone will have the permission to use violence against the passer-by. Rather, it involves the gunman announcing that violence will be used.
Himma, however, could escape this objection; at least if he is claiming that the law-makers’ authorisation is analogous to a scenario in which the gunman conditionally threatens someone with violence. We could imagine a scenario where the gunman has a subordinate waiting for the gunman’s permission to beat someone. In this new scenario, the gunman announces to the passer-by that unless she hands over her wallet, he will allow his subordinate to beat her. This announcement is certainly a conditional threat. Can we, therefore, conclude the same about law-makers’ permission-granting announcement? I don’t think so.
The very picture of the law-maker as someone who, as it were, unleashes authorities on citizens if they don’t comply with the authorities’ orders strikes me as an odd one. There is a temporal element worth paying attention to. In the gunman’s scenario, the gunman announces that he will allow the subordinate to use violence unless the passer-by complies with his orders. That is not what typically happens in typical legal systems. In typical legal systems, it is generally the case that law-makers first establish a framework that allows authorities to apply and enforce legal norms; law-makers first grant powers and permissions to authorities to apply and enforce laws. Once it is established who has the powers and permissions to apply and enforce particular kinds of laws (criminal laws, private laws, and so forth), only then do law-makers issue the laws and the corresponding sanctions and enforcement mechanisms (which corresponds to the action in t1). Thus, this second version of the gunman’s scenario doesn’t yet provide a good analogy with legal systems.
It is possible, however, to offer a gunman’s scenario that is closer to how typical legal systems usually work. Here it is: Before going out on the street, the gunman tells his subordinate: ‘if passers-by disobey my orders, you are allowed to beat them.’ The gunman goes out on the street and announces to a passer-by: ‘if you don’t give me your wallet, my friend here may beat you.’ I take it that most would agree that it is the second announcement, rather than the first, that looks like a conditional threat. The first announcement is not even addressed to passers-by. The same is true of law-makers’ permission-granting announcements. They are not addressed to citizens; they are announcements made to guide authorities’ actions, not citizens’. Himma, therefore, is mistaken: law-makers don’t conditionally threaten citizens by authorising coercive enforcement mechanisms.
3.2. Threats and Duties
(t4) announcing that unless citizens ϕ, law-enforcers will perform an enforcement action that they have the duty to perform.
(t7) announcing that unless a citizen ϕs, the law-applier will apply a law that prescribes sanctions or enforcement mechanisms that they have the duty to apply
I’ll consider the actions in t4 and t7 simultaneously, given that the considerations I’ll raise apply equally to them.
Take the law-enforcer who, because, and only because, of the duty to use force to stop rioters if they refuse to stop, announces: ‘Stop, or I’ll use force!’ There is a simple reason why this announcement isn’t a conditional threat: the fact that the enforcer has made the announcement does not work as a reason for them to use force. Had they failed to make the announcement, the enforcer would still have the same reasons to use force. The enforcer’s announcement just changes the epistemic position of rioters. It doesn’t set in place the conditions for enforcers to use force. And if the enforcer wanted to portray themself as doing their job and acting lawfully, they must not have intended that rioters believed that the primary reason for using force was the fact that they have made the announcement. Rather, the primary reason must be the fact that rioters were in violation of a law that the enforcer has the duty to enforce.
This suggests something more general. When law-enforcers and law-appliers are committed to their legal duties and want to be perceived as such, their conditional announcements to carry out these duties don’t qualify as conditional threats; they are best construed as warnings about authorities’ prospective actions.
But law-enforcers and law-appliers may not be committed to act according to their legal duties. When that happens, it seems possible for them to conditionally threaten to perform an action that they have the legal duty to perform (think of corrupt cops who don’t generally perform their duties but threaten to do so to elicit a bribe). But this raises a worry. When law-appliers or law-enforcers are not committed to performing their legal duties (or not committed to being perceived as being committed) and, for some reason, threaten to perform them, we may wonder whether they are acting in the name of the system, given that we may question whether the authorities are fulfilling their roles.
Here is then what we can extract from what I’ve said. When law-enforcers and law-appliers are committed to the duty to perform a given action if certain conditions obtain, announcing that they will perform it won’t amount to a conditional threat: in those cases, for the authorities the primary reasons to perform the relevant action are announcement-independent. On the other hand, when law-enforcers and law-appliers are not committed to the duty to perform a given action if certain conditions obtain, announcing that they will perform it may amount to a conditional threat. Nevertheless, the conditional threat cannot be a conditional threat made in the name of the system.
3.3. Threats and Discretion
(t5) announcing that unless citizens ϕ, law-enforcers will perform an enforcement action that they have discretion to perform.
(t8) announcing that unless a citizen ϕs, the law-applier will apply a law that prescribes sanctions or enforcement mechanisms that they have discretion to apply.
As before, I’ll discuss the actions in t5 and t8 together given that similar remarks apply to them.
It strikes me that the actions in t5 and t8 are the best candidates for being classified as conditional threats. Having discretion to perform an action solely implies that there is no duty not to perform the action and no duty to perform the action. Thus, if one has discretion to ϕ, one doesn’t thereby have a reason that favours ϕ-ing. If that is right, then the fact that, say, a law-enforcer had discretion to perform an enforcement action wouldn’t work as an announcement-independent reason to perform it. This opens space for thinking that when law-enforcers and law-appliers announce that they will perform discretionary actions unless some further facts take place, they want the addressee to believe that they are going to perform the announced consequence primarily because they have announced that they would.
An enforcer who has discretion to use lethal force if someone resists an arrest does seem to be threatening a citizen by announcing ‘Stop, or I’ll shoot you!’ The announcement doesn’t just inform citizens of something that was already going to happen; something the enforcer had primary reasons to do (as a matter of law) regardless of the announcement. The announcement itself sets up the condition under which the enforcer will bring about the unwelcome consequence (i.e., the condition that citizens do not stop resisting). Now that they have made the announcement, they have a new reason, a reason to shoot if the condition laid out by the announcement doesn’t take place. And it is not implausible to think that the enforcer may have intended that the citizen believes that the announcement is the primary reason for shooting. Notice that, contrary to those cases where law-enforcers and law-appliers threaten to perform actions that they have the duty to perform, no concerns about legal authorities acting beyond their roles emerge here.
Be this as it may, it would be too hasty to conclude that such announcements are threats most of the time. Even though law-appliers and enforcers have discretion to perform a certain law-applying or enforcement action, on many occasions it would be mistaken to construe them as intending that citizens believe that their primary reason for performing the relevant action is the fact that the announcement was made. In many, if not all, typical legal systems there are strict limits and guidelines to exercise discretion. Authorities who are committed to follow these guidelines may perform an enforcement action primarily because they believe that the enforcement action is an appropriate response and is in conformity with the existing guidelines. Besides, they may intend to not be perceived as a whimsical authority that performs enforcement actions mainly because they said so; instead, they may intend to be perceived as fair authorities that do what the law, other authorities, or even the public deem appropriate to the situation. When that is their intention, then we cannot say that these authorities are threatening citizens.
Here it would be surprising if the standard intentional or motivational profile of the typical law-enforcement and law-applier agent departs from what has just been described. But we also know of the existence of cases where law-appliers and enforcers seem to either have or intend that citizens believe that they have a primary announcement-dependent reason to perform the announced discretionary action (think of the re-emergence of the police motto “when the looting starts, the shooting starts” in response to the George Floyd protests).Footnote 41 How can we determine what is the standard, or more frequent, way in which law-appliers and enforcers operate?
This is not something we can do entirely a priori. What would determine whether law-appliers and law-enforcers are issuing threats when performing action tokens of t5 and t8 is a collection of empirical facts about the intentions of authorities who perform these actions. Until a proper empirical investigation on this issue is carried out, the best option is to suspend judgment as to the frequency of the occurrence of threats in these scenarios.
There is, however, a silver lining. Even if it turns out that law-appliers and law-enforcers do conditionally threaten citizens most or even all the time when they perform the actions in t5 and t8, it can still be the case that most citizens most of the time are not threatened by typical legal systems. The actions in t5 and t8 involve particularised demands, that is, demands that a particular authority issues to a particular citizen or group of citizens in the exercise of their discretionary powers. Hence, the number of addressees of these demands is much smaller than, for example, the number of addressees of announcements made by law-makers. Most people, after all, go about their lives without ever sitting in a court room or without ever receiving a direct order from a law-enforcer.
3.4. Contextual Threats
(t6) issuing orders in a context where it is a punishable offense to disobey the orders of legal authorities and where there are norms permitting and empowering authorities to sanction those who disobey their orders.
The core idea here is that when, for example, a law-enforcer declares that someone is under arrest or when a judge passes a sentence, the relevant announcements are made in a context where it is a punishable offense to disobey the authority’s orders. Hence it might be thought that legal authorities make implicit threats when they perform actions that fit t6’s description.
First, it is not the case that simply by making an order, e.g., ‘Pay Martin a thousand dollars,’ an authority is implying that ‘If you don’t pay Martin, I’ll apply a sanction.’ The authority may not have intended to mean that; perhaps they were not willing to apply a sanction in that particular situation and were just issuing the order to make the addressee aware of the duty to pay Martin a thousand dollars. We can only claim that an authority implicitly made the announcement to apply an unwelcome consequence if they intended to convey such an announcement. We should, therefore, reject the claim that whenever an authority issues an order (in a particular context), they thereby (implicitly) threaten citizens with the application of a sanction or enforcement act.
There are, of course, cases where authorities do imply that a sanction or enforcement mechanism will be brought about if citizens disobey their orders. In relation to these cases, it is important to distinguish two kinds of circumstances: circumstances where law-appliers and enforcers have the duty to apply the unwelcome consequence on citizens who disobey their orders; and circumstances where law-appliers and enforcers have discretion to do so. As it may be clear, the discussions about these two circumstances are absorbed respectively by the discussions of t4, t7 and t5, t8 (sections 3.2 and 3.3.).
4. What Then?
We’ve seen that threatening doesn’t portray ‘a large part of how law operates.’ Legal threats are much more uncommon than most legal and political philosophers would have us believe. This conclusion does not only shed light on the way in which authorities of typical legal systems typically address citizens; it also has some bearing on important discussions in legal and political philosophy.
The broader lesson is that typical legal systems can and should be construed as typically addressing citizens in a distinct, and perhaps more respectful, way than the threats-centred picture portrays. Though typical legal systems may wrong and do wrong citizens in numerous ways, we can—most of the time—exempt typical legal systems from the wrongs associated with threatening. Legal and political philosophers who take up the burdensome task of justifying the wrongful actions of legal authorities and legal institutions or argue against anarchist objections can then be relieved from at least one justificatory burden.
But the main conclusion defended in this paper also has narrower implications. In discussions about global justice, it has been claimed that states’ obligation to promote distributive (egalitarian) justice is grounded on the fact that existing municipal legal systems violate citizens’ autonomy by threatening them.Footnote 42 If the view defended here is correct, then we can see that it is probably not a good idea to ground states’ obligations to foment equality on the fact that typical legal systems threaten citizens.
The main claim defended in the paper also allows us to reassess widely held conceptions of legal governance. If legal authorities typically address citizens via warnings, and not via threats, then we must explore how this claim impacts the pervasive idea that typical legal systems are coercive. According to a popular account, threatening is a necessary condition of coercion.Footnote 43 If this account is correct, it would follow that typical legal systems are much less coercive than usually thought. Now, it is far from clear that such accounts of coercion are correct;Footnote 44 in fact, some might be tempted to take the implication that typical legal systems are not coercive (or that they have a very low degree of coerciveness) as a reductio against such accounts. But this needs further discussion. After all, if we heed the fact that legal systems are purported to guide citizens, not goad them like a coercer or a threatener would,Footnote 45 it might not be that surprising if coercion turns out to have a much less prominent role in typical legal systems than commonly thought.
But if coercion does have a prominent place and if it turns out that coercion by threats is only one among the many possible ways in which legal systems can coerce citizens, the low frequency of threats in typical legal systems might at least prompt us to re-evaluate the plausibility of accounts that consider the threat of sanctions or other enforcement mechanisms as the paradigm of legal coercion.Footnote 46 It all suggests that more work needs to be done to explain what typically makes our legal systems coercive (if they are typically coercive).
Exploring these implications in detail is, of course, not a task for this paper. I hope, however, that the paper has provided a starting point for further reflection on legal governance devoid of threats.
Acknowledgments
This paper couldn’t have taken its current form without the help of many colleagues and friends. For written comments and discussion, I am grateful to Antony Duff, Ben Sachs, Euan MacDonald, Grant Lamond, Himani Bhakuni, James Edwards, James Manwaring, Luís Duarte d’Almeida, Martin David Kelly, Neil Walker, and Paul Burgess. Some early ideas that gave origin to this paper were presented at the Oxford-Edinburgh Legal Theory Colloquium. I thank the event’s participants for the fruitful discussion and for nudging me to further develop the ideas therein presented. I am also grateful to the anonymous reviewers for their detailed engagement with the paper and to the Canadian Journal of Law & Jurisprudence’s editorial team.
Open access
