On canceling parties

1. 1. Imagine that there is a party next Saturday night. Imagine too that many of your friends will be there. Imagine finally that you have two options: (1) attend the party or (2) skip the party. Contingent on there being a party, you might choose (1). In fact, your preference might be both clear and strong. You might be willing to pay a great deal to attend the party. At the same time, you might prefer another option: (3) the party does not take place. You might wish, on reflection, that the party had not been arranged in the first place. Or you might hope that the party will be canceled. Your preference ordering is (3), (1), and (2).

Why is that an imaginable preference ordering? A general answer is ‘fear of missing out,’Footnote ¹ but that phrase is ambiguous. What does it mean? It is often understood to mean that if you miss something, there is some probability that you will lose something of importance – which suggests that you should prefer (1) to either (2) or (3). The question remains: Why would you prefer (3) to (1), but (1) to (2)?

One possibility is that if you do not attend the party, you will give a signal that you would prefer not to give. The signal might be that you do not like parties, that you do not like your friends, or that you do not like the host. The cost of giving any of those signals might seem very high. You might jeopardize or lose friendships, or compromise relationships. But if the party is canceled, you can avoid an event that you prefer to avoid without giving the undesired signal. Note that a crucial feature of your thinking is the social meaning of failing to attend the party (Lessig, Reference Lessig1995). You might not intend the received meaning (you do like your friends; you do like your host), but you have little or no control over it. This, then, is a possible reason for the preference of (3) to (1) and (1) to (2): The unwanted signal that is given by refusing to go to the party.Footnote ²

There is another possibility. You might think that if the party happens, there will be opportunities for relationship-building. You might not much care about those opportunities; if you did, you would not want the party to be canceled. Still, you might think that if other people take advantage of the opportunities and if you do not, you will be at a comparative disadvantage (Mason, Reference Mason1995; Bursztyn et al., Reference Bursztyn, Ferman, Fiorin, Kanz and Rao2018). (This is one conception of ‘fear of missing out.’) If the party is canceled, you do not have to go, which is good (very good!), and you also will not be at a disadvantage, which is good (very good!) as well. You go to the party to avoid that disadvantage, but without the party, you would be better off.

1. 2. Now suppose that you go to the party, and they have alcohol. You have two choices: (1) drink or (2) do not drink. You might choose (1), even though you would prefer it if (3) no alcohol is served at the party. You might choose (1) over (2) for the same reason that you went to the party in the first place. A possible problem is the signal of (2). You might fear that people would wonder: Are you an alcoholic? Are you a puritan? Do you hate fun? Why can’t you loosen up?

2. 3. Now suppose that you are in the midst of a conversation. It is not a terrible conversation; you are not suffering; it is pleasant enough, even though you are bored and a bit restless. You have two choices: (1) continue the conversation and (2) stop the conversation and move on. You choose (1), even though you would prefer it if (3) the host interrupted you, such that you had no choice but to stop the conversation. You might choose (1) over (2) for the same reason that you went to the party in the first place. In fact, people sometimes engage in conversations for longer than they wish, perhaps above all because terminating a conversation gives a signal that people do not want to give (Mastroianni et al., Reference Mastroianni, Gilbert, Cooney and Wilson2021).

3. 4. What is true of purely social events might also be true of charitable or political activities. Suppose that there is a fundraising event on Tuesday night or a protest of some kind. You might show up for one or another, perhaps because you do not want to signal that you do not care about the relevant cause. In the extreme case, you actually do not care about that cause, or perhaps you even despise it (Kuran, Reference Kuran1995). But the personal costs of not showing up might loom very high. You might wish that the event was not occurring or that it would be canceled, but if it is on, you are there. You might well focus, in such cases, on self-signaling; you do not want to signal, above all to yourself, that you do not much care about some cause that, on reflection, you approve. What kind of self-image can you retain, if you give yourself that embarrassing or even shameful signal?

(Compare: Your friends are getting married, and you are invited; they are having a large wedding. You hate weddings. You wish that they were not having a large wedding and that you were not invited; but you go. Or: Someone you know has died, and the family is having a memorial service. You hate memorial services. You wish that they were not having a memorial service; but you go.)

Note that in some such cases, people’s desire not to give an unwelcome signal to others or to oneself might be solving a collective action problem: The group benefits from widespread participation, even though there is an individual incentive to defect (and stay home). A beneficial social norm might be in place: Go to events that involve relevant fundraising activities or protests. On reflection, you might celebrate the norm. I will return to this point.

1. 5. Now suppose that you are not happy with your job, and that you have another opportunity that you would prefer. Suppose that you like and admire your boss and your coworkers, which means that you are highly reluctant to quit your job; you would feel guilty and a bit ashamed. As between quitting your job and continuing, you prefer to continue, because of how you would feel if you quit. But if your position were magically terminated, for reasons independent of your performance – if you did not have the option of continuing with your job – you would be happier still. You might have this preference ordering: (1) Your position is terminated, (2) you stay in your job and (3) you leave your job. (For obvious reasons, you do not want the option (4): to be fired.) Here you are focused on the signal given by leaving and also the potential rupturing of relationships.

2. 6. Now imagine that there is a competition of some kind, and you might turn out to be the winner. Suppose that you very much hope that you will win. Suppose that you would pay a great deal to win, and also that you would pay something to have a 1/X chance of winning, even if X is a pretty big number. Even so, it is possible that you would prefer that the competition did not exist. The reason is that you are likely to lose, and if you lose, you will be very sad. It may be that everyone, or almost everyone, who is eligible for the prize feels the same way that you do. The basic problem has to do with competitiveness: Some competitors hate to lose, which means they also do not like to compete; and if a competition is on, they badly want to win.

3. 7. Now suppose that some aspect of your identity is socially salient: religion, race, ethnicity, age and gender. People expect you to claim some aspects of that identity: Catholic, white, Hispanic, middle-aged, male. You do so. You might even be expected to make a strong identification along some such lines. You might do as expected. But it is possible that you would prefer a situation in which you did not need to make that claim or that identification, or in which the relevant aspect of identity was not salient or took a radically different form, or did not exist at all.

4. 8. Now suppose that you are in love with someone and that you think that you want to spend your life with that person. But you are highly ambivalent about getting married. You might think that you would prefer (1) getting married to (2) not getting married, given the existence of the institution of marriage. But there might be a voice inside your head, asking: Might it not be better, for you, if the institution of marriage did not exist? If there is such a voice, it is because not getting married, given the institution of marriage, offers an unwanted signal to the person you love (as well as to others).

The various examples I have presented are not the same, but they are overlapping. The shared characteristic is that they involve some good that people consume (understanding the words ‘good’ and ‘consume’ very broadly), or some activity in which people engage, even though they wish that good or activity did not exist. The reasons for their preference ordering vary. The problem might involve signaling to others: It might involve networks and comparative disadvantage; it might involve self-signaling; it might involve competitiveness. (This is not meant as an exhaustive list.)

Paying for things you don’t like

A great deal of work attempts to elicit people’s valuation of social media platforms. If people are asked how much they would demand to stay off (say) Instagram or YouTube for a month, they state a significant amount, often in the vicinity of $50 (Allcott et al., Reference Allcott, Braghieri, Eichmeyer and Gentzkow2020; Sunstein, Reference Sunstein2020). Findings of this sort suggest that social media platforms produce a substantial consumer surplus (Allcott et al., Reference Allcott, Braghieri, Eichmeyer and Gentzkow2020). If people would have to be paid a great deal to stop consuming a good, it would seem clear that the good is delivering significant benefits.

But we might want to hesitate before accepting that conclusion. People might demand money to stay off a social media platform for a month, supposing that everyone else who is relevant will be on that social media platform for that month. But what exactly does that tell us? It might tell us that people benefit from being on the platform, in the sense that it provides them with a variety of goods. It might tell us that people enjoy or otherwise benefit from being part of a network. But does it tell us that they are glad that the platform exists? Not necessarily. We need to ask: If they could push a button and abolish the platform, would they do that, if their only concern was their own welfare?

Leonardo Bursztyn and collaborators tried to answer that question (Bursztyn et al., Reference Bursztyn, Handel, Jimenez and Roth2023). They found that on average, users would demand $59 to deactivate TikTok for a month and $47 to deactivate Instagram for the same period. From those findings, we might conclude that the two platforms make people much better off. But in sharp contrast, users would on average be willing to pay $28 to have all of the world, including themselves, deactivate from TikTok for a month and $10 to do the same for Instagram. Almost two-thirds of active TikTok users appear to lose welfare from the existence of the platform, and the same is true for almost one-half of active Instagram users. It appears that many people would demand significant money to stop using a product that they wish did not exist. Notably, there is heterogeneity on this count; some people truly gain from the existence of both platforms. But a very large number believe that they lose.

This is an exceptionally important set of findings. Bursztyn and collaborators speak of ‘product market traps,’ in which people are maneuvered into participating in a market that they wish did not exist (even if they are active participants in that market). As they show, product traps exist when the value of relevant products stems in whole or in large part from the fact that people would lose out from nonuse or nonconsumption, on the assumption that other people are using or consuming those products. Significantly, product traps may exist even if there is no issue of addiction (a question to which I will turn in due course).

It is independently important to note that because people make ‘hedonic forecasting errors,’ an ex-ante prediction of the welfare effects of deactivation may not measure the actual welfare effects of deactivation (Gilbert et al., Reference Gilbert, Pinel, Wilson, Blumberg and Wheatley1998; Gilbert and Wilson, Reference Gilbert, Wilson and Forgas2000). People might be willing to pay something for a good from which they lose welfare, even apart from the network effects. People might not be willing to pay something for a good from which they would gain welfare, even apart from the network effects. But if people are willing to demand payment to give up a good for whose abolition they would also be willing to pay, we have good reason to think that people simultaneously benefit from having access to it, contingent on its existence, and would benefit from eliminating it, if only they could. Bursztyn et al. find an assortment of evidence consistent with this conclusion (Bursztyn et al., Reference Bursztyn, Handel, Jimenez and Roth2023). One of the points I am making here is that product traps are everywhere. Sometimes they are deliberately engineered by savvy marketers. Sometimes they are a result of norms whose existence cannot be attributed to particular designers, and that might have emerged from invisible hand mechanisms (Ullmann-Margalit, Reference Ullmann-Margalit2017).

Note that these findings, and my major claims thus far, have not involved precommitment strategies at least not on the level of individual choice. This is not a situation of Ulysses and the Sirens (Elster, Reference Elster1984). People do not say that they would like to commit themselves in advance not to make certain choices. They want to make those choices. The problem is not their own recklessness or impulsiveness, or some form of akrasia. We are not dealing with a disparity between planners and doers. The problem is that not having access to the relevant good creates a loss, given that the good is available to and used or consumed by others.

The central findings here are in a sense the mirror image of a plausible hypothesis about willingness to pay to protect the environment (Sen, Reference Sen1995). Suppose that people are asked how much they are willing to pay to save an endangered species or a pristine area. Suppose too that the amount is trivially small and that if aggregated across a large population, it remains relatively small. We might think that people do not much care about the relevant good. But people may be willing to pay little to protect the environment unless they are assured that other people will pay as well (id.). People might think: ‘My own payment will do little or nothing. But if everyone is paying that admittedly small amount, we can achieve a great deal. I would pay a great deal, supposing that everyone else is doing so.’ This possibility raises serious cautionary notes about contingent valuation studies that attempt to measure people’s willingness to pay for public goods at least if those studies do not distinguish between (1) willingness to pay, without information on what other people will do and (2) willingness to pay, with information that if one gives, others will be giving too.

The Barbie Problem

Return to the example with which I began and suppose that someone has decided to have a party this Saturday night. Suppose that people are willing to go, given that the party is occurring, but that all or most people would prefer that the party not occur. The ambivalent partygoers face a prisoner’s dilemma. If they could coordinate and agree not to go to the party, they would be better off (unless the signal given by their collective action imposes costs; perhaps, the party-giver has power over them). But how can they do that? Such coordination might be very difficult.

Companies can maneuver consumers into analogous situations. Suppose that there is a product that consumers want to have, contingent on its existence, but whose existence consumers do not welcome and might deplore. If so, the product is akin to the Saturday night party. Companies might market a product by emphasizing (1) that people will feel ashamed or bad if they do not have it (because of the negative signal given by not having it) or (2) that people will lose some status by not being part of a network. We can readily imagine products that fall in this category.

We might call this the Barbie Problem. Notwithstanding the success of the 2023 movie, and with apologies to anyone who really loves Barbie, we might speculate that many children, and even more parents, wish that there was no such thing as Barbie, even if children play with Barbie, and even if parents purchase Barbie. Not having a Barbie might provide an unwanted signal. Not having a Barbie might produce exclusion from a relevant network (and a loss of relative position). In the examples with which I began, the dinner party is a Barbie, so is alcohol and so is a dull conversation.

Or consider wearing ties: Many men have worn ties, even though they derive no pleasure from doing so and might wish that ties had never been invented. Ties are Barbies. Or consider high-heeled shoes: Women might wear such shoes, even though they find them uncomfortable and wish they did not exist. A great deal of fashion might be characterized as Barbies.

Or consider a new iPhone: You might not want to keep using an iPhone from a few years ago, even if you quite like it if you know that a newer model is available, even if you would be happier if no newer model existed. Bursztyn et al. find evidence consistent with this conclusion (Bursztyn et al., Reference Bursztyn, Handel, Jimenez and Roth2023). They find that people would prefer less frequent new product launches, even though they would buy the relevant product, given frequent product launches. The reason for this distribution of preferences is not self-evident. Perhaps people are satisfied with the current product and anticipate (and regret) their own impulsiveness (the Ulysses and the Sirens problem). Perhaps, people know that they will be embarrassed for others to see that they do not have the latest model, and so they wish that the latest model was not released so often. New iPhones might be Barbies. For some people, marriage is a Barbie, and for other people, so is religion, and for other people, so is gender.

In some of these cases, the structure of the problem is that some person P engages in some action (e.g., holds a party, serves alcohol, markets a product, develops or emphasizes a norm) that forces another person Q to make a choice (attend or not, purchase or not) that reveals information about Q that Q does not want to disclose, and that requires Q to decide whether to give a signal that Q would prefer not to give.Footnote ³ In many cases, Q’s dilemma does not arise by accident. It is a product of P’s clever, mischievous or self-interested action. If P is an employer, it might ask employees to reveal their political leanings, including, for example, whether they would want to unionize. If P is a government, it might give citizens the choice of whether to take a loyalty oath (or to decline, politely).

Norms might do the work of law; they might prevent people from defecting from some endeavor from which the group benefits (Ullmann-Margalit, Reference Ullmann-Margalit1977). People might contribute to some shared task – for example, moving office equipment, giving blood, giving to charity and contributing in one or another way to national defense – even though they would like to defect, were the norm not in place. When a norm solves a collective action problem, it makes people better off, even if each individual would prefer to defect and allow the problem to be solved by others. On reflection, people should welcome the norm.

These cases are not my concern here. I am focused on the Barbie Problem – on cases in which people do not, on reflection, welcome the relevant norm. They do not think that it solves a collective action problem. They think that they have been maneuvered into, or trapped in, a situation that they dislike or deplore (cf. the discussion of partiality norms in Ullmann-Margalit, Reference Ullmann-Margalit1977).

Positional goods

There has been a great deal of discussion of the difference between positional and nonpositional goods (Frank, Reference Frank1985; Reference Frank2005). Health is plausibly a nonpositional good; people want to be healthy, whether or not most people are healthy or not. Motor vehicles are plausibly positional goods; you might want what now counts as a very good car, or a fancy car, only because of the current vehicle mix. This might be true for three different reasons. The first involves signaling: You might want to signal something good about yourself, or at least you might not want to signal something bad about yourself. Motor vehicles are, among other things, signals. The second reason involves your self-image: Even if no one else would know what car you drive, you might not want to think of yourself as the kind of person who drives a less-than-very-good car.

The third reason, and the most interesting, is that the current vehicle mix provides the frame of reference against which you measure your own car (Frank, Reference Frank1999). If the mix were greatly inferior to what it now is, you might like a greatly inferior car; you would not experience it as greatly inferior. But if the mix is much better than it was (say) 20 years ago, your experience of a car from that period would not be very positive. You might find that car to be hopelessly primitive in multiple ways. This is so even if your experience of the car, at the time when it was released, was or would have been exceptionally positive. Positional goods give rise to positional externalities: If some people purchase a new car, they impose a cost on those who do not have that car (Frank, Reference Frank2005).

Positional goods are among the category of goods that people consume but wish did not exist. Certainly this is true if people consume such goods to avoid giving an unwanted signal, or to avoid a form of unpleasant self-signaling. The frame of reference issue must be analyzed differently. It is hard to unthink what one knows, and if you know about an amazing current motor vehicle, you will not easily wish that everyone had less-than-amazing vehicles from 20 years ago.

Addiction

Some people struggle with addictions. They might be addicted to cigarettes, alcohol, heroin or social media (Sharot and Sunstein, Reference Sharot and Sunstein2024). The characteristic feature of an addiction is that people experience (1) lower intrapersonal benefits from consumption over time and (2) higher intrapersonal costs from nonconsumption over time (Sunstein, Reference Sunstein1986). The welfare cost comes because people increasingly suffer from nonconsumption, which means that they wish, on reflection, that they had not started consuming the good in the first place.Footnote ⁴ People might even wish that the relevant good did not exist in the first place, even as they consume it. They might adopt a precommitment strategy to deprive them of access to goods that they would freely choose. Or they might wish the relevant goods did not exist in the first instance, but given the existence of the goods, and given their addiction, they may or may not wish the goods to be rendered nonexistent.

All this is true and important, and it has strong implications for policy and law. For example, it supports a ban on smoking (Goodin, Reference Goodin1989), perhaps limited to people born after a certain year (Gretener and Magramo, Reference Gretener and Magramo2022), to recognize the current existence of addiction and to prevent further cases of addiction. People consume addictive goods even when they wish those goods did not exist in the first place. Addiction is a cousin to my primary interest here, but it is only that. It is not a Barbie Problem.Footnote ⁵

Policy and law: a new frontier

What are the implications for policy and law? For addictive goods, the answer is relatively well-understood. People might be informed or warned; people might be nudged; the relevant goods might be taxed (Friedson et al., Reference Friedson, Li, Meckel, Rees and Sacks2021); the relevant goods might be regulated (Cutler et al., Reference Cutler, Jessup, Kenkel and Starr2016) or even banned. For positional goods, the answer is also well-understood (even if policy, in most nations, remains in a primitive state). Such goods might be taxed (Frank, Reference Frank2005; Antinyan et al., Reference Antinyan, Horváth and Jia2020), or people might be nudged not to buy them. A progressive income tax might be thought to be the best response to positional externalities (Frank, Reference Frank2005).

What about the Barbie problem? What about goods that people consume so as not to give an unwanted signal, or so as not to be disadvantaged in some network or some status competition? Should people really be warned about Barbies, ties or high heels? Why warn people not to go to parties that they (rationally) want to attend, given the existence of those parties? Should Barbies, ties or high heels be taxed? Should Barbies, or Instagram, be regulated or banned, to the extent that they are Barbie problems?

In general, the best route for solving the problem lies in voluntary action on the part of those who are stymied or trapped, perhaps through consumer pressures or the creation of a new norm of some kind (Bursztyn et al., Reference Bursztyn, Handel, Jimenez and Roth2023).Footnote ⁶ However, norms cannot easily be created without solving serious collective action problems (Lessig, Reference Lessig1995). In that light, we could imagine cases in which legal action might be considered. Recall the possibility that some agent has deliberately engineered a situation in which people must either (a) consume a product they do not like or engage in an activity they do not enjoy or (b) offer a signal that they would greatly prefer not to offer, or disclose a fact that they would like to keep private. Or consider a similar situation, caused not by some agent but by custom or by invisible-hand mechanisms. With respect to law and policy, there is a great deal of room for future work.

Suppose that the result of the existing situation, or clever marketing, is to cause serious welfare losses (Bursztyn et al., Reference Bursztyn, Handel, Jimenez and Roth2023) or to intrude on protected rights. An example might be a case in which a public employer gives employees an opportunity to support an incumbent politician, or to come to a holiday celebration associated with a specific religious tradition, or to pray together on Monday mornings. If so, legal interventionFootnote ⁷ ought not to be out of the question – perhaps in the form of nudges, perhaps in the form of taxes, perhaps in the form of subsidies and perhaps in the form of mandates and bans. In the United States, the Hatch Act can be understood as a specific response to a Barbie problem: It forbids certain public employees from engaging in political activities, in part because of an understanding that if asked to do so by high-level officials, such employees could not easily refuse. They would be trapped.

The problem is large and pervasive. Barbie Problems are common. In important domains, we would seem to have a basis for rethinking the law of worker and consumer protection and considering potential action. Indeed we can go further. In the case of social media, which I have emphasized here, many people face a Barbie Problem, which argues for policy reforms, perhaps above all to help young people against product traps. Private and public interventions, designed to help people to escape those traps, might increase social welfare; consider efforts to nudge teenagers not to spend a great deal of time on social media platforms late at night. And if people are asked to waive their rights, it might be best to ban the waiver request. Waivers might be Barbies.