Proterrence & Rule Illegitimacy in an Age of Creeping Social Control: The Ban on Tobacco Smoke in Amsterdam’s Coeeshops

This paper examines the concept of proterrence: scaring people into doing something to stop others from doing something bad. This contrasts to deterrence, which involves threatening persons to not do something bad. The tobacco ban in Amsterdam coffeeshops and, more specifically, coffeeshop personnel’s reaction to it is used as the empirical vessel to examine proterrence. Proterrence permits examination of the interface between order maintenance and social control against a backdrop of perceived sanction illegitimacy. It also permits exploration of the process by which formal sanctions thread through informal mechanisms—where that threading is enforcement rather than consequence-based and where rule implementers face the brunt of the sanction that a third party violates. Data are based on in-depth fieldwork in Amsterdam coffeeshops. The wider applicability of proterrence is discussed.


Introduction
Without rules, there can be no social order, but rules must be legitimate if social order is to be respected (Tyler 1990). Not infrequently, rules emerge that lack face validity: They are illogical in light of other conduct that is similar or worse yet permitted.
Persons charged with administering such rules may recognize the hypocrisy, but failure to enforce them threatens consequences that they, as rule-implementers, must face.
In 2008, Dutch coffeeshops came under the purview of a new rule, widely regardedby personnel and outsiders alike-as the most befuddling of all: The government banned tobacco smoking in coffeeshops but continued to permit that of cannabis (marijuana and hashish). Violations would apply to the coffeeshop and carried penalties ranging from a warning to a few hundred euros fine to several times that amount. Like other paternalistic policies of its general kind (Eriksen and Cerak 2008), this one was intended to protect the health and welfare of hospitality workers. But coffeeshop personnel did not see it this way. The encroachment onto their "lifeworld" (Habermas 1984(Habermas , 1987 was unwelcome and gave rise to efforts to defy and deflect the policy's reach. The significance of these efforts, however, is less empirical than conceptual. By exposing the interface between order maintenance and social control against a backdrop of perceived sanction illegitimacy, coffeeshop personnel's actions reveal a novel expression of social control-which we call proterrence (Author XXXXa).

The Dutch Law
Dutch coffeeshops are retail establishments that specialize in cannabis distribution (Leuw and Marshall 1994;MacCoun 2011;MacCoun and Reuter 1997). They are the most (in)famous example of Dutch tolerance (Buruma 2007;Mascini and Houtman 2011;Pakes 2005;Reinarman 2010). Yet, that perception is misleading. In fact, Dutch coffeeshops are highly regulated. They are places of order, not anarchy, established and maintained through the joint functioning of formal and informal control (Author XXXXa).
To be clear, it is a crime to sell cannabis in the Netherlands. There are severe sanctions for breaking the law. At the time of the study (described shortly), the maximum penalties for possession, cultivation, sale, transport and production of cannabis for commercial purposes are six years' imprisonment or a fine of €67,000 (NMFA 2008). However, legal authorities tolerate the retail sale of cannabis by coffeeshops "to protect cannabis users … from exposure to hard drugs and the criminal elements who traffic in them" (NMFA 2008:18). This "deliberate ambivalence" (MacCoun 2011) is a case of harm-reduction, meant to separate the market in cannabis from that in hard drugs like cocaine, heroin and ecstasy (Leuw 1991;NMFA 2008;NMHWS 2003;Reinarman, Cohen, and Kaal 2004;Wouters and Korf 2009) The government grants coffeeshops the privilege to break the law, but this comes with a catch-a set of strictly enforced rules or regulations. Everyone inside a coffeeshop must be at least 18 years of age. No customer can be sold more than 5 grams of cannabis per day, and there can be no more than 500 grams on premises at any time.
Hard drugs are strictly prohibited. Nuisance is not allowed, and cannabis advertisements are forbidden. Violation of the rules may result in the closing of the business or criminal prosecution. These rules are to be actively enforced: Police are required to perform a random search of every coffeeshop twice yearly (Trimbos Institute 2010;van der Gouwe, Ehrlich, and van Laar,2009).
The mechanisms for enforcing the rules (i.e., police checks and punishments) are part deterrence, part proterrence. The rules scare personnel out of making violations (deterrence), such as intentionally selling to minors, peddling more than 5 grams, storing more than 500 grams, advertising cannabis, or tolerating hard drug use and nuisance at the establishment. Moreover, the rules scare personnel into preventing other persons from breaking rules (proterrence).
[2] They actively shoo away minors, hard drug dealers and users, as well as persons prone to cause fights due to being "drunk" on alcohol (Author XXXXa). Personnel do so because, otherwise, police may walk in the door and sanction the coffeeshop for breaking the rules. Thus, per the law, coffeeshops are not treated as victims, but rather culpable-and therefore publishable -for the bad behavior of non-personnel. In this way, formal control (that exercised by the government) is used to motivate informal control (that not exercised by the government).
Whereas the above rules were long established, that prohibiting tobacco smoking in coffeeshops was new at the time of our study. The notion that customers should be prohibited from using a legal drug (tobacco), but not an illegal one (marijuana), was difficult for personnel to fathom and even more difficult (perceptually) for them to enforce. That dissonance produced two byproducts. The first was defiance-a net increase in the prevalence, incidence, or seriousness of offending caused, at least in part, when a norm or rule is perceived as "substantively arbitrary, discriminatory, excessive, undeserved, or otherwise objectively unjust" (Sherman 1993:459-61). The second was restrictive deterrence, where the rule is recognized but broken in ways that decrease the overall risk of formal detection (Jacobs 2010). When decisionmakers take specific measures to evade sanction threats with which they also have a "fundamental problem," restrictive deterrence and defiance become one and the same. In such cases, offending is both insurgent and adaptive.
In essence, coffeeshop personnel became rule-implementers with a rule-breaking mentality. This paradox occurs when informal and formal social control intersect against a backdrop of perceived rule illegitimacy. That intersection provides a novel opportunity to examine an increasingly important, but broadly unspecified, conceptual phenomenon.

The Formal/Informal Social Control Connection
The nexus between formal and informal social control has long been a fundamental concern of deterrence. As Bishop (1984:405) classically observed: [Despite focus] in the deterrence literature on the effects of formal sanctions, there is no reason to expect that obedience to the law is solely or even primarily a product of legal coercion. Threats of punishment from parents, peers, and other significant others are likely to have a similar or greater effect. Moreover, the deterrent effect of threats of legal punishment may be contingent upon the perceived risk of informal sanctioning. Fagan and Meares (2008:182) echoed this sentiment, proclaiming that "social norms are not simply imposed on individuals by societal structures or processes. Rather, they are effective only when internalized through formal and informal processes of socialization, or where formal sanctions are reinforced by informal sanctions" (Fagan and Meares 2008:182). It may even be the case that informal sanctions are the "real deterrent and that formal punishment is important only insofar as it triggers informal sanctions" (Bisho 1984:405; see also Gibbs 1975Gibbs , 1989Paternoster and Iovanni 1986:769;Tittle 1980:241;Williams 1985:148).
When informal social control inhibits rule-breaking, the mechanism typically is indirect. That is, it is based on a generalized fear of what would-be offenders believe that others will think about them if they are caught. This mechanism is consequencebased rather than enforcement-based: It flows from the prospect of social reaction rather than the actual delivery of the formal sanction by the informal party.
Proterrence departs from this tendency by leveraging a sanction source that is formal but a sanction delivery mechanism that is informal, where the informal party is charged with implementing a rule against a primary rule-breaker whose violation the rule-implementer ultimately is responsible for. In this way, the rule implementer is an agent of enforcement but a target of it as well.
This paradox is not limited to Dutch coffeeshops. Virtually any setting in which formal authorities try to scare intermediate parties into doing something to stop others from doing something bad, while punishing those intermediaries for not doing it, should-in theory-promote these processes. Examples abound in business (inspector/merchant/customer relations), industry (regulator/manufacturer/end user relations), medicine (government/doctor/patient relations), sports (commissioner/coach/player relations), and even home life (police/parent/child relations).
The notion that secondary parties can be held at least partly blameworthy for the rulebreaking of those over which they have putative control is not new. It is called vicarious liability-which has long been an established principle within American jurisprudence (Laski 1916). Thus, parents can be culpable for the intentional and willful acts of their un-emancipated children (Weinstein 1990/91). Employers can face potential responsibility for the negligent conduct of their employees (Sykes 1984).
Landowners can be liable for violent acts committed by third parties against invitees on their premises (Kennedy 2013 implementers (i.e., the secondary parties) face the brunt of the sanction that a third party violates. This intersection becomes conceptually relevant when the rule lacks legitimacy because it muddles the line between rule-breakers and rule enforcers.
Indeed, proterrence should not present any novel conceptual issues when rules are, in fact, legitimate. The intermediate party is simply implementing policy set by some higher authority, that policy is perceived to be fair and right, and there is no tension between formal and informal processes.
In an increasing number of circumstances, however, no such alignment will exist.
Misalignment results from two trends that pervade contemporary rule administration: Creeping social control (Innes 2001) and the delegation of rule enforcement to informal parties. On the first point, laws have exponentially increased in number, scope, and coverage (Harsanyi 2007). That increase flows from the perceived need for safety and control in a dangerous and uncertain world (Innes 2001). Three forces have given rise to it: paternalism (government as parent figure), infantilism (citizen as child who needs to be taken care of), and moralism (government as moral standard setter; see, for example, Harsanyi 2007). Together, these forces have triggered massive growth in the administrative state and a concomitant inability to enforce all of the rules. That inability forces the state to outsource at least some of the workload to informal parties (Estlund 2010). This outsourcing, in turn, creates scenarios in which informal parties face rules they do not necessarily agree with, but, if they do not implement, risk punishment. [4] When formal rules vector through informal mechanisms and cause the informal party to conspire with primary rule violators to defy and evade the threat, the structure and process of the threat is being changed endogenously. That change flows from game theoretic principles. Consistent with game theory, proterrence involves players, information, actions, decision points, and consequences (Rasmusen 1994). But proterrence involves simultaneous cooperation and non-cooperation between participating parties against a backdrop of asymmetric awareness. The cooperative aspect applies to the functional alignment between coffeeshop personnel and patrons and flows from the fact that personnel do not perceive the official law to be legitimate.
Non-cooperation refers more to the tension between the law itself and coffeeshop personnel's reluctance to abide by it. The principle of awareness (Glaser and Strauss 1965) is layered on top of this, given that personnel are fully cognizant of the law while primary rule-breakers are not aware until specifically told, which results in the alliance between personnel and rule-breakers and the cooperative efforts to circumvent the rule. Asymmetric awareness resulting from imperfect information means that the interaction between the secondary rule enforcers (i.e., coffeeshop personnel) and the primary rule-breakers is sequential and contingent (see, for example, Ryan 1994).
Cooperation between the secondary rule enforcers and primary rule-breakers inherently translates into non-cooperation between primary (i.e., tobacco police) and secondary rule enforcers (coffeeshop personnel) when the latter engage in measures to make sure patrons do not get caught. Patrons not getting caught means secondary rule enforcers will not be punished. In this way, both cooperative and non-cooperative gaming co-occur (Rasmusen 1994), which is made possible by the perceived illegitimacy of the rule the secondary parties are charged with enforcing. All of the parties are acting in the name of self-interest, while both primary and secondary ruleenforcers are also acting with moral concerns in mind (the former by enforcing the law, the latter by coming up with ways around it because they perceive it to be arbitrary and unfair)-which is not always the case in game situations with zero-sum outcomes.
These processes are significant because they cause sanction properties to behave in novel and unexpected ways. Thus, coffeeshop personnel bifurcate the certainty dimension of sanctions (i.e., certainty of detection vs. certainty of formal apprehension) and cause it to go in opposite directions. Personnel also manipulate the celerity of the sanction threat to influence the (formal) certainty with which it is applied. Proterrence permits identification of these and related nuances, which opens important new avenues for research in an era of creeping social control and the delegation of rule enforcement to informal parties.

Methods
The second-author, hereafter referred to as the fieldworker, obtained qualitative data one side was written in English and the other in Dutch. Next, he visited each business to request the participation of personnel. The goal was to interview the highestranking representative possible (e.g., a manager over a server). The minimum requirement was that the person must have worked there for at least 6 months. Upon meeting a potential participant, the fieldwork introduced himself, handed them his business card, and briefly stated the study's purpose and methods, including that respondents would receive a remuneration payment of €50.
The participants are 64% male; average age is 34; 70% identified as White, 6% as Black, and 24% other; 40% immigrated to the Netherlands; 10% were married; 26% graduated from secondary school; 56% and 30% reported daily use of cannabis and alcohol, respectively. Participants also provided information on characteristics of all personnel of their coffee shop. They reported, on average, 67% are male; 81% White; 47% immigrants; and, 15% married. Table 1 provides an overview of the interviewees' characteristics; table 2 provides that of all personnel at their respective coffeeshops.
- Table 1 about here--Table 2 about here- The primary purpose of the interviews was to obtain data about coffeeshops' prevention, experiences with, and handling of victimization (for findings, see Author XXXXb). However, the project's beginning coincided with that of the ban on tobacco smoking in coffeeshops. An intriguing topic, the fieldworker also questioned personnel about it. First, he would ask an informal, open-ended question along the lines of, "What do you think about the ban?" After the response, he would pose follow-up questions to elicit further details. All questions were asked in English because he only speaks English fluently; fortunately, this language is the "coin of the realm" in the study area due to tourism and the influx of working immigrants from European Union nations. As with any self-report study, some participants may have (un)intentionally said something not wholly reflecting the "truth." The fieldworker sought to reduce this problem by asking follow-up questions, promising confidentiality, and informing participants of their rights as a research subject through a consent form.
Interviews were audio-recorded and transcribed verbatim. Using NVivo 10, a qualitative, software package, files were coded with identification tags corresponding to relevant research issues. Initial tags were generally broad. One of these was tobacco-related information. The first and second authors sifted through the data to create narrower distinctions. The major emergent categories were personnel's On July 1, 2008, coffeeshops became regulated by the new policy. Violations were punishable with administrative penalties, which ranged and escalated from a warning, to a few hundred euros fine, to several times that amount (NMHWS 2005). Questions about enforceability in coffeeshops were addressed in the government publication, "Frequently Asked Questions on Smoke Free Catering" (NMHWS 2008). As to whether coffeeshops must be tobacco smoke-free, the official position is stated as: "Employees of coffeeshops also deserve protection from exposure to tobacco smoke. Coffeeshops are catering establishments and therefore subject to the tobacco legislation. Like other catering establishments, they may set up enclosed smoking areas." Also, the publication addresses if "the smoking ban appl[ies] to cannabis not mixed with tobacco," which the Dutch refer to as smoking "pure." The answer: "The Tobacco Act and related legislation apply only to tobacco products, that is, products consisting wholly or partly of tobacco." In short, tobacco smoking, but not that of cannabis, became outlawed. [6] When that happened, the hypocrisy of the edict came into stark relief among those it was putatively designed to protect. The notion that one should not be exposed to tobacco while working at a coffeeshop was, to many personnel, bewildering. " [I] with tobacco]," Hassan explained. "You open a coffeeshop for smoking. Someone who doesn't smoke won't go in a coffeeshop. You open a coffeeshop for the kind of people who smoke." Claire declared the policy "rubbish," while Luuk labeled it "crazy." Or, as Jana expounded: "I think it is ridiculous for the government to say it is ok to smoke weed inside [coffeeshops], but not tobacco. It is so weird for a government to say that." Indeed, a number of personnel chose the coffeeshop profession precisely because it permitted them to smoke (tobacco) while working-an intrinsic benefit available to few other customer service occupations. "If you want to work in a coffeeshop," Anna surmised, "you know you will be working with [tobacco]. It [basically] is like smoking [tobacco]. If you don't want to work in a coffeeshop because of your health, then you don't go work in one." Readers must understand that tobacco is, and has been, embedded in Dutch culture for centuries, providing historical context for these sentiments (Roberts 2017). As early as the 1600s, the Netherlands was a central tobacco hub for all of Northern Europe (Postma and Enthoven 2003). Dutch ships were among the first to introduce the plant to the Far East (Gilman and Zhou 2004), and by the 18 th century, Dutch colonists created a thriving tobacco trade in the New World-exporting the plant back to Europe for widespread consumption (Postma and Enthoven 2003). To this day, Holland and tobacco remain culturally synonymous, with Dutch Masters® Tobacco depicted in rap/hip hop songs in which blunts are stylistically consumed by posturing artists (Foster 2016).
Against this backdrop, the tobacco ban was as illogical to personnel as the punitive threat was real. "[A]s a coffeeshop there is this huge spotlight on you," Linda explained, "They have a close eye on coffeeshops and if you make one wrong move. If they [the police] see you [or anyone in the coffeeshop] smoke a cigarette, you are doomed. You don't want to lose your permit or anything." The threat, of course, was mediated: Violations by the customer would be attached to the shop itself-requiring personnel to "police" customers to protect themselves.

Message Conveyance
For personnel, the first line of defense was message conveyance. Would-be rulebreakers cannot be threatened into compliance by a rule about which they do not know (Beccaria 1995), and customer ignorance was a problem that personnel needed to confront head-on. A venue filled with smoke, after all, would presumably not differentiate between good smoke and bad smoke, much less criminalize one and not the other. The distinction is about as artificial as the one between water and holy water (Szasz 1975).
Ignorance was especially common in coffeeshops in the study area, given the ubiquity of tourists therein. "Most of [the customers] don't know [about the policy]," Noah explained, adding, "It is ridiculous to people from other countries that you can smoke weed, but you can't smoke tobacco." Gijs fully understood how "strange" it would to "go to a city where you can smoke dope and you cannot smoke tobacco. It is very confusing. You can get weed but you can't smoke tobacco. They [customers] don't understand. It's very confusing for them." The confusion required coffeeshop personnel to teach and re-teach newly arrived entrants in an endless onslaught of verbal social control. "We are constantly on top of people to tell them to put their tobacco away," Lizzie complained, "I explain to them that in the coffeeshop you cannot smoke tobacco because we have a tobacco ban generally in Holland. Only marijuana [is allowed]." Gijs lamented that the tobacco ban is "a burden. You have to play police officer next to being a dealer, a manager, or bar lady. It's tourists who don't know." Alongside the verbal instruction were visual devices used to broadcast the message.
Placards proclaiming "No Tobacco," "Smoke Free Zone," "Rookvrije Zone," "Smoking Forbidden," and "Verboden te Roken" populated coffeeshop interiors. Other signage publicized the ban with a universal "no" symbol over the image of a burning cigarette.
The size, formatting, and placement of these signs was similar to those depicting other rules, such as "No entry under 18 years," "No hard drugs," and "No violence"-with one exception: Tobacco signs were uniquely detailed, offering information on the rule's background, where it applied, the punishment risk, and the legal consequences of consumption.
Given the frequency with which potential rule-breakers ignore warnings or fail to fully process them cognitively (Stewart and Martin 1994), message conveyance also hinged on the anticipatory removal of symbols that might otherwise signal tolerance. Of particular relevance was the ashtray. Restricting their availability provided an ideal mechanism for rule announcement at the moment the rule was going to be violated: Jana: If the ashtray is there, people think they can smoke cigarettes here. Since the smoking ban came in, I give it out, and I say they are not allowed to smoke cigarettes, only joints [i.e., marijuana in rolling paper]. So then they know. That works a lot better than in the beginning when we still had all the ashtrays out, you could see people also smoked cigarettes, but now you have the chance to tell them not to.
Olivia: People have to ask for one [an ashtray] now because of the smoking ban.
We tell people they have to ask for one, and they will get one. There used to be ashtrays on every table, because obviously people were coming here to smoke [tobacco]. Now we don't always have ashtrays on the table, and people have to ask for them, and we have to explain about the smoking ban and stuff. We take them off the tables now [when they leave].

Limits and Adaptations to Message Conveyance
Readers understand that message conveyance had limited utility. The sheer number of people who patronized coffeeshops, the rate at which they turned over, the inability to alert every conceivable patron of the ban, and the recalcitrant few patrons who would defy the ban despite being told of it meant that message conveyance was never wholly effective.
Personnel could still conceivably prevent patrons from violating the ban, however, yet permit them to smoke. This is because the new policy allowed tobacco smoking in a designated/enclosed area inside businesses, or outside if there was at least one nonpartitioned, unblocked side (NMHWS 2008). Smoking outside was the superior option -because the smoke would dissipate into the sky-but that option was often neither feasible nor practical. Coffeeshops with a terrace could provide it, but terraces were not altogether common, and constructing one could be cost-prohibitive. Even with a terrace, Amsterdam's climate was not necessarily conducive to outside smoking.
Boasting nearly 30 inches of annual rainfall and average minimum temperatures under 40°F almost half the year (Royal Netherlands Metereological Institute 2013), smokers might-literally and figuratively-be "snuffed out." Interior solutions had their own set of problems. In coffeeshops with multiple levels, one floor could be dedicated to tobacco smoking. The stipulation was that the tobacco floor could not be one in which personnel primarily worked. Because the entry-level floor tended to be the level with the sales counter(s), either the basement or upstairs would become the default tobacco area. The problem was that personnel would continue to walk into tobacco smoke when cleaning up after customers on these other floors or enforcing other rules, triggering a de facto rule violation.
Another solution required constructing a glass wall and door to separate areas within a floor, so that the primary work area became the non-tobacco space. Many coffeeshops in Amsterdam, however, were small, which made floor-splitting impractical (both spatially and economically). Floor-splitting also had a chilling effect on sociability-the reason so many Dutch coffeeshops thrive in the first place. "If you want to kill any form of social intercourse in a public place," Finn intoned, "put up a glass wall. That is so stupid."

Restrictive Proterrence
These challenges and the impracticability of enforcing a blanket ban caused personnel to rely on a coterie of tactics that obfuscated tobacco use by patrons or provided personnel plausible deniability of the fact that patrons were indeed smoking tobacco.
Such tactics involved a combination of conspiracy and ignorance, both real and feigned, in which personnel leveraged the functional limits of enforcement to minimize their own culpable awareness of violations or the probability that violations would result in sanctions. The proterrent effect was not absolute but restrictive, resulting in continued rule violation at a lower perceived risk.
To see how this dynamic operated, it first is necessary to contextualize how patrons smoked cannabis and the mechanisms it provided for subtle ban violation. One modality was to smoke out of a reusable device, such as a pipe, bubbler, bong, hookah, or vaporizer. Other than the hookah, these devices were only packed with cannabis.
The more popular method was to sprinkle cannabis into a small, thin piece of paper and roll it into a joint. The cannabis could be mixed with tobacco, an herbal substitute, or alone (again, referred to as "pure").
Joints posed the obvious enforcement problem for personnel and police alike, as there was no practical way to tell if tobacco was inside unless one witnessed the joint being rolled. When burned, tobacco-infused joints could not reasonably be differentiated by smell from joints containing only cannabis. This impediment allowed patrons to openly smoke tobacco-mixed joints without being exposed as rule breakers, something personnel knew and exploited. "Sometimes it happens with people," Wouter admitted, "because you cannot smell it." Or as Gijs quipped, "I don't have a little machine that detects nicotine, like an alcohol test where you blow in alcohol fumes and it can be states that you are not supposed to smoke tobacco, period. If we don't see it then we don't know about it. You are not allowed to smoke a cigarette, but you can mix your tobacco with a joint [as long as we don't see it]." Clearly, the issue was plausible deniability and whether personnel knew that tobacco was in a joint. That knowledge putatively required intrusive measures. As Jack thusly proclaimed, he was "not a tobacco Nazi." Theoretically speaking, detection might come retrospectively after seeing, for example, dismantled cigarettes or other tobacco residue in or around patrons' consumption spaces. But residue did not necessarily equate to rule breaking. "[Finding a broken cigarette] doesn't mean it's been smoked in there," Stefan rationalized. Nonetheless, personnel minimized the risk further by requiring customers to hide source clues of tobacco, such as cigarette packs or broken cigarettes: In the event personnel came across tobacco remnants, they would quickly whisk them away. "We take [the ashtrays] off the tables quicker now and clean them out to make sure there is no tobacco left in there," Olivia explained. Note her concern was raw tobacco, since the ashes of cannabis and tobacco look and smell the same.
All of this social control occurred against a backdrop of acquiescence, if not downright encouragement, of rule flouting. Personnel understood that they could not permit tobacco smoking should they come across it, yet they also were not reticent to ." "We let them smoke joint with tobacco," Mara continued, "but no cigarettes. We only say no to cigarette smoking." James' justification for rule flouting was moralistic, suggesting that his real concern was the wellbeing of patrons: We were very strict about it to start with when the law first came in: only smoking pure weed. But when I was on my own I had three people pass out at the same time and there is no way I can drag unconscious people outside so it is just better to let people smoke tobacco in their joints. If you are used to smoking a mixed joint and then you put a gram of pure weed in and you smoke it because that's the law, you are not enjoying the experience. You are supposed to enjoy the experience, relaxing.
James' comment underscores the role of discretion in enforcement-which has long been the critical intervening variable that links rule creation with rule imposition. Dean explained lenience as a function of spatial factors and, specifically, the natural obstacles to detection that came with the large, multi-floor shop in which he labored.
Although he publicized the tobacco rule, he also recognized that global enforcement was infeasible. "Sometimes [patrons] are coming in and when people see it we tell them they are not allowed to smoke [tobacco]. It is not allowed, but, yeah, sometimes it

Discussion
In Dutch coffeeshops, it is de jure and de facto forbidden to smoke a legal drug, tobacco, but de facto permitted to smoke cannabis, despite the latter being illegal-a striking paradox that has gained media attention around the world (Connolly 2008;Kucharz 2008). Personnel reacted to the rule in decidedly defiant ways. Witnessing deviance and not reacting to it in any way demonstrates defiance, and our respondents were certainly not defiant in this sense. Yet, working with customers so that they could subvert and evade the formal threat was for all practical purposes still defiance.
The defiance was sneaky, wherein the informal agent of social control made it appear as if the sanction threat was being fully enforced when that enforcement was in fact partial.
The process we examined here is proterrence. A law was created; it was applied to a primary actor; a secondary party was given responsibility for preventing violation; and, the secondary party was held to account if that did not happen. Formal sanctions vectored through informal mechanisms; that vectoring was enforcement-based rather than consequence-based; and, rule implementers (i.e., the secondary parties) faced the brunt of the sanction that a third party violated. 1966:961;Bishop 1984:405;Williams and Hawkins 1986;Zimring and Hawkins 1973:174), nearly all of that research is consequence-based and affect-drivenfocusing on the shame and embarrassment that afflicts rule-breakers as a result of experiencing official penalties like arrest and prosecution (Grasmick and Bursik 1990). Researchers have paid far less attention to the conditional relationship between formal and informal social control when that relationship is process-(i.e., enforcement) rather than outcome-(i.e., consequence) related and when that The key, however, is a perceptibly illegitimate sanction threat. When that happens, as it did here, motives align between rule-implementers and primary violators. Perceived illegitimacy is that rare form of visceral affect that enhances consequence relevance and renders decision-making more careful, not less. This is because its moralistic underpinnings give rise to calculation designed to get around the threat. Indeed, we can think of no other example in the literature where affect simultaneously promotes consequence relevance and defiance/evasion. These dimensions are almost always expected to vary directly, not inversely.

Although criminologists have long documented how and under what circumstances formal and informal social control intersect in coercive environments (see Andenaes
[9] The sensitizing role of affect in decisionmaking ) is enhanced when responsibility for rule-breaking is put as much on the intermediary (e.g., the coffeeshop) as on the actual rule-breaker (i.e., the patron) because it incites the former to conspire with the latter to mitigate the threat's functional reach.
Perceived illegitimacy and formal/informal vectoring reveals the process by which decision-makers act on their risk environment, as opposed to simply being passive objects of risk. Scholarship in the neoclassical tradition typically presumes that risk is exogenously determined and that actors do not manipulate it real-time through their own conduct (but see Anwar and Loughran 2011;Thomas et al. 2013). Although it is true that offenders routinely take actions to avoid detection and are not automatons in the risk-responding process, the prevailing assumption within neoclassical theory is that sanctions are created, they are implemented, and they work or don't. Research typically does not assume that risks are agentically modified by the decision-makers those risks target (cf. Cherbonneau and Jacobs 2019). Our research is part of a growing body of work that reveals the endogeneity of risk manipulation, which the game theoretic quality of proterrence permits to happen. That is, sanction risk is not merely an inhibitor of offending but a resource that can be modified in real-time to allow crime to be committed more effectively. This view essentially turns the sanctionbased prevention process on its head, causing sanction properties to react in novel and unexpected ways, which we explore below.
In traditional settings, social control agents manipulate sanction threats in an irregular fashion to make the threat look more real or imminent than it really it is. This is because offenders are ambiguity averse (Loughran et al. 2011), and when the threat of sanctions is irregular, would-be offenders overestimate the probability of punishment (Pogarsky et al. 2018). Ambiguity aversion is precisely the logic behind "crackdown/back-off" strategies in policing (Sherman 1990), which are premised on the notion that constant risk promotes habituation and causes offending to creep back up to pre-enforcement levels (Koper 1995). When risk is varied and irregular, by contrast, its effectiveness increases because offenders perceive the threat to be more imminent than it really is.
The mediated nature of threat delivery in proterrence shows how social control agents (in this case, coffeeshop personnel) exploit unpredictability to undermine rather than enhance the rule's preventive effect. This can only happen because the formal sanction threads through informal mechanisms and because personnel have a vested interest in reducing the probability of suffering a sanction due to something their customers might do. In essence, personnel exploit ambiguity because it makes their own potential guilt more difficult to establish. This occurs, for example, by ignoring the fact that joints could very well have tobacco in them and by otherwise concealing infractions that would be actionable (e.g., quickly removing discarded cigarette filters and/or tobacco debris from tables). These actions make the official sanction threat less probable, not more.
The proterrence process also bifurcates the certainty dimension and propels it in opposite directions (see, for example, Cherbonneau and Jacobs 2019). Bifurcation occurred when personnel communicated an elevated risk of punishment (i.e., "you're violating the law by doing this and here's how to not do that") to prevent patrons from getting caught as opposed to making formal punishment more certain. In so doing, they increased the certainty of detection to decrease the certainty of sanction. [10] This is precisely the opposite of how deterrence is supposed to operate (see, for example, Nagin and Pogarsky 2003). Indeed, our data show that the formal threat produced various forms of noncompliance and concealment to evade the sanction.
Although personnel who enforced the threat only did so to prevent the patron from being caught in the act, proterrence would predict this given that formal and informal agents were working at cross-purposes: Violations implicated the latter as readily as they did customers, requiring personnel to minimize the probability that sanctions would be suffered as opposed to simply minimizing rule-breaking itself. Given that many patrons did not know or understand the tobacco rule, fresh offenses were always being committed anew. Thus, the issue was as much about not getting caught for the violation as it was about attempting to reduce the frequency of violations. The enactment of message conveyance and sneaky evasion reveals the synergy between these two goals in practice. In the vein of endogenous risk altering, our data also demonstrate how the celerity of sanction enforcement was manipulated to mitigate its certainty. Celerity, or the speed with which sanctions follow rule-breaking, is the most understudied component of sanction-based prevention (Nagin and Pogarsky 2001). It is presumed to vary independently from certainty. Thus, sanction threats can be fast in coming but unlikely (in an overall sense) to be suffered, or slow in coming but certain in ultimate application. In mediating the threat's delivery, personnel essentially introduced measures that delayed the time gap between potential formal awareness of infractions and actual enforcement of those infractions. This occurred, for example, by instructing patrons to change their location within shops and by removing evidence from tables of tobacco. By introducing these delays, personnel manipulated the celerity of the sanction threat to undermine its certainty. The celerity/certainty relationship has not, to our knowledge, been specified previously in this way (but see Jacobs 2012 and, partly because celerity has historically focused on the gap between infraction and punishment, not infraction and detection. Detection, however, is a form of punishment (Pogarsky, Piquero, and Paternoster 2004;Pratt and Turanovic 2018). For some offenders, getting caught is as, or more, consequential as the official penalty that comes after getting caught (Mamayek, Paternoster, and Loughran 2017).
When the enforcement of rules or laws is mediated through informal mechanisms and where secondary parties are held responsible for the sins of primary others, the celerity dimension of sanctions can be leveraged to decrease the perceived certainty that a violation will be formally detected. Proterrence thereby shows that certainty need not precede celerity and severity-a traditional assumption within neoclassical research. When punishment is conceptualized as detection and not just the penalty that results from detection, tactics can be taken informally (by personnel, for example) to lengthen the gap between infraction and detection and thereby alter the certainty (i.e., probability) of formal punishment.
The manipulation of celerity to influence certainty has broader implications for the concept of punishment avoidance. Conceptually speaking, punishment avoidance has almost always centered on its post-incident implications for offending trajectories (Paternoster and Piquero 1995): Offenders commit an infraction, they get away with it, and their future perceptions of the likelihood of sanction threats are altered, usually downward, due to emboldening effects (Pogarsky and Piquero 2003 Piquero and Paternoster 1998). Our data, by contrast, reveal the prospective role of punishment avoidance in risk calculation. That is, punishment avoidance is a process, persons engage in it on-the-fly, and these practices influence subjective perceptions of risk as perceptions unfold real-time. Our data also demonstrate the manner in which punishment avoidance affects perceptions of sanction risk before sanctions are suffered, and how specific practices undertaken by individuals alter perceptions of risk in an anticipatory rather than reactive fashion (cf. Jacobs 1996). To our knowledge, this phenomenon has only been identified one other time in the literature (Cherbonneau and Jacobs 2019).

Conclusion
The practice of holding secondary parties responsible for the rule-breaking of primary others appears to be growing in size and scope (Tutin 2016). The trend is an artifact of creeping social control (Innes 2001) and also changing social norms in which blame attribution is more important-or easier-than addressing the root causes of rule breaking (for an overview on blame attribution, see Rosenthal and Schlesinger 2002).
Externalizing fault is convenient. Blaming others makes us feel better about our own shortcomings. It is consistent with what psychologists call self-serving bias, where people take credit for good things that happen in their life but externalize bad outcomes to outside factors (Miller and Ross 1975). The bias is not only ego-protective and cathartic, it reinforces a personal locus of control even as it removes responsibility for outcomes that do not go our way. Blaming secondary parties also is functional for society because it permits consequence dispersion over a wider swath of the population, which dilutes the burden to individual violators. Consequence spreading also encourages policies and procedures that enhance public safety: When responsible secondary actors know they are accountable for the behavior of their underlings, institutions run more efficiently-reducing the probability of rule-breaking and the consequences therefrom (Sykes 1981).
Whether holding secondary parties responsible for the rule-breaking of others is legitimate is a separate matter. Perceived legitimacy is more often than not procedurally defined because fairness of process determines the level and intensity of buy-in from social actors (Feather and Boeckmann, 2013;Tyler and Jackson, 2014;Van der Toorn et al. 2011). The substantive validity of rules and laws is, more often than not, presupposed. As our research shows, that assumption may be mistaken.
Perceived illegitimacy (from a substantive rather than procedural perspective) cuts to the core of feelings about what is right, just, and proper. Rules that lack face validity and which levy sanction threats in the name of protecting persons engaged in voluntary activity undermine human agency itself (Sherman 1993;Tyler 1990).
Although affect may traditionally be thought either to excite or inhibit conduct (Frijda 2004;Topalli and Wright 2014), perceived illegitimacy does both by energizing defiance/evasion while controlling its expression. The synergistic coexistence of these two countervailing forces remains seriously underspecified and is brought into focus by the proterrence processes explored.

Notes
[1] The first-author would like to thank Professor Christian Joppke for bringing this point to his attention during a lecture several years ago.
[2] Due to the threat of punishment, for example, personnel are deterred from distributing large quantities of cannabis; and, they are proterred into preventing hard drug dealers from setting up shop at their establishment, and stopping customers from getting into fisticuffs (an example of nuisance).
[3] Lest we believe that the indirect nature of these and related rule violations somehow makes them less consequential, the penalties for negligence can be quite severe: Parents have been jailed for the sins of their children (Molina 2018); corporations have forfeited millions in legal settlements for the negligent acts of their agents (Dowie 1977); landowners stare down massive punitive verdicts for inadequate security on their premises (Nguyen 2004).
[4] Evidence of the relevance of proterrent processes involving misalignment between informal agents and formal rules is not difficult to see. Medical practitioners are barred from prescribing or recommending certain treatments-because they are not officially approved, they are risky, or because they are not part of the standard of care -yet practitioners know for some patients in some circumstances that those treatments will work. Some practitioners defy those rules and work with patients to tell them how to obtain the illegal treatment with reduced probability of formal detection (see, for example, Levinson 2018 assistance of indigent players with workarounds that facilitate athletes' taking of contributions while minimizing the risk of incurring sanctions (Branch 2011). Social media platforms charged with administering arbitrary rules about speech and decency provide notice to would-be content providers as to the number and pattern of government removal requests and thereby educate those providers as to which venues are most likely to be flagged by authorities. Essentially, companies like Twitter are giving users a roadmap as to what type of material they can post and where (Hsia 2017). Automobile inspection services work with car owners to pass vehicles on emissions and safety tests that technically aren't pass-worthy-recognizing that the standards are too high or unfair and that the car is just fine (Landen and Dougherty 2018). The list could go on. In all of these cases, the misalignment between the substance of the formal rule and the need to enforce it against others in their charge so they, as informal agents, are not punished for it, gives rise to unique permutations in the sanction process.
[5] Part of the rationale for granting a long transitional period was the hope that Nederlanders would become less fearful of the ban's economic consequences by learning from Ireland's experience with respect to bars. Indeed, after Ireland began its tobacco smoking ban in 2004, its bars had less air pollution with little to no economic downturn, while employees enjoyed better health (Howell 2005). Similar effects have been found in other countries (Eriksen and Cerak 2008).
[6] The document also specifies, "If smoking club members smoke in private premises, the [Dutch] tobacco legislation does not apply" (NMHWS 2008). No coffeeshop in the 1012 (i.e., study area) was a smoking club, so this did not apply to our sample.
[7] For example, using hard drugs in open view.
[9] Even respondents' message conveyance activities, which putatively illustrate a neutral effort to publicize rules, was really more about reducing the likelihood that patrons would get caught than anything else.
[10] The law does not require coffeeshop personnel to report smoking offenses to the authorities, which is why detection certainty and apprehension certainty can go in opposite directions.
[11] Given that most patrons, especially in tourist areas, would be unaware of the tobacco ban, such patrons were relatively unconcerned about getting caught for smoking tobacco. This concern was significant for coffeeshop personnel, however, which is precisely why they engaged in the measures outlined here. Violations by patrons attached to coffeeshop personnel, meaning that detecting these violations and then managing them away reduced the likelihood that personnel would be sanctioned.
[12] Although detection avoidance is also a form of punishment avoidance, existing scholarship puts that conduct within the domain of restrictive deterrence. Restrictive deterrence captures the process by which offenders elude sanctions rather than its implications for future offending likelihoods. We focus on process rather than outcome and in so doing, seek to sensitize criminology to a temporal nuance of punishment avoidance (i.e., both prospective and real-time) that is less often studied.