The Color of the Pen

I · The Color of the Pen

The man arriving for the experiment was told it concerned workers’ performance and attitudes, which was a lie of the ordinary academic kind. A woman was already sitting in the room. The experimenter introduced her as Paula, explained the task—a large matrix of digits, in which pairs were to find and circle particular number sequences for a reward—and set them to work. Paula went first, in red ink. Then the pens were passed to him: one red, one black. Which one he picked up was the entire study.

The logic is worth admiring for its economy. If he wrote in red, the page would show a single undifferentiated scrawl and nobody afterwards could say who had done how much. If he wrote in black, the record would be legible: her work, his work, and the ratio between them. Neither choice is odd. Neither is remarked on. But across a group of people the proportion choosing black tells you, without asking, whether they are trying to keep track.

In the first of three studies, published by the psychologist Margaret Clark in the Journal of Personality and Social Psychology in 1984, thirty-three male undergraduates made that choice. Eighty-eight point two per cent of one group took the pen that made the accounting visible. Twelve point five per cent of the other group did. Both are a long way from the fifty per cent that would mean nobody cared, and a long way from each other; the difference had a z of 9.42. The two groups had done the same task with the same woman for the same reward. The only thing that differed was a sentence they had heard beforehand. Told that Paula was married, they kept score. Told that she was new at the university and anxious to meet people, they went out of their way not to.

Clark replicated it twice more with the manipulation changed to something less contrived—working with an existing friend against working with a stranger—and got ninety-four point four per cent against forty-two, and, in the version where each pair produced a single datum, ninety against thirty. The direction never wavered. People who expect a certain kind of relationship count. People who hope for another kind take active measures not to be able to.

This magazine has twice asked who is wanted and why. The question here is different and, in a way, prior to it. Nobody lives inside one relationship. A life is a set of them, running simultaneously on incompatible terms: relations in which you give and defer, relations in which you are given to, relations that keep a rough level. The terms are rarely negotiated and almost never stated. And the striking thing about Clark’s pens is that the terms were fixed before anyone had done anything at all—before the task, before the reward, before either party had been generous or stingy. What follows is an attempt to describe how wanting somebody is produced inside that arrangement, what communication and status and timing actually contribute to it, where the sexes differ and where the evidence says they do not, and whether the word people reach for—consumption—is doing honest work or standing in for an attitude.

II · Two Ledgers

Five years before the pens, Clark and her collaborator Judson Mills had run the experiment that made the pens interpretable, and it produced one of the more counterintuitive results in the attraction literature. Male undergraduates were introduced to an attractive female confederate under the same two descriptions, married or newly arrived. All of them were then maneuvered into doing her a favor. Afterwards she either repaid the favor or did not. The men who had been told she was married liked her more when she repaid them. The men who had been told she was hoping to meet people liked her less.

Being paid back is not a neutral act. It is a statement about what kind of relationship this is, and in one of the two available kinds it is an insult. Clark and Mills drew the distinction that has organized the field since: an exchange relationship is one in which benefits are given in response to benefits received, and a debt is incurred and discharged; a communal relationship is one in which benefits are given in response to need, and receiving one creates no specific obligation to return a comparable one soon. The pens follow directly. In the first kind you must keep track, because you cannot allocate in proportion to inputs without knowing the inputs. In the second, keeping track is not merely unnecessary but evidence against you, which is why Clark’s communal subjects picked the pen that would make the page illegible.

Neither of these is generosity and neither is meanness. They are accounting rules, each perfectly rational under its own. The trouble is that people run several at once. The colleague you will invoice, the friend you will not, the sister whose debts to you are unnameable, the person you have just met and cannot yet classify: each of these is a separate book, and there is no exchange rate between them. Later work by the same group found that the direction of attention differs as well—members of communal relationships monitor the other’s needs, members of exchange relationships monitor their inputs—so the two kinds are not even watching the same variable.

That the books stay separate is not an accident of modern life; it is the condition that makes them work. Georg Simmel, writing in 1908 about what he called the intersection of social circles—the phrase reached English, in Reinhold Bendix’s translation, as the web of group-affiliations—argued that a person’s individuality is constituted precisely by the particular set of circles they stand in, no two of which coincide. Half a century later Erving Goffman gave the operational version, audience segregation: the performer keeps his audiences apart so that no one of them ever sees two performances at once. Both were describing something the exchange experiments would later measure. If the sister and the colleague were in the same room and the same ledger, one of the two relationships would have to be rewritten.

What none of this yet explains is why any of it should bear on wanting someone. The connection turns out to run through the one variable neither ledger records.

III · The Interval

The anthropologist and sociologist Pierre Bourdieu, working from Kabyle ethnography in the nineteen-seventies, noticed that a gift returned immediately is not a gift. It is a swap, and everyone present knows it. What converts a transfer into a gift is the interval—the stretch of time between giving and returning, during which the return is neither guaranteed nor scheduled. The delay is not incidental to the transaction; it is what allows both parties to experience a sequence of obligations as a sequence of free acts. Hand the money straight back and you have not settled the account, you have refused the relationship, and rather pointedly. This is theory rather than measurement, and it is offered here as theory. But it named the variable, and the variable turns out to be measurable.

The sociologist Linda Molm spent three decades measuring it. Her experiments hold constant everything an economist would care about—what is exchanged, how much it is worth, how often, how equally, the shape of the network, the history between the parties—and vary only the form of reciprocity. In negotiated exchange, the two sides bargain to an agreement and both are then bound by it; benefits flow bilaterally and simultaneously. In reciprocal exchange, one side simply gives—help, advice, a favor—without negotiation and without knowing whether or when anything will come back. In generalized exchange, you give to one person and receive from another, and no direct debt exists anywhere in the system.

The results are consistent enough to be slightly uncomfortable. Partners in reciprocal exchange trust each other more, feel more affective regard for each other, and are more committed to each other than partners in negotiated exchange getting exactly the same goods on exactly the same terms—a finding Molm, Nobuyuki Takahashi, and Gretchen Peterson published in the American Journal of Sociology in 2000. Three years later, in the American Sociological Review, the same three reported an inversion that ought to be better known: although negotiated exchange has every feature the procedural-justice literature says people want—joint decision-making, voice, advance knowledge of the terms, mutual assent—people judge their partner to have treated them more fairly under the reciprocal form. This holds when the exchange is objectively equal and when it is objectively unequal. In further work with Jessica Collett and David Schaefer, Molm found that people are also more willing to go on accepting a losing deal when it takes the reciprocal form; and that generalized exchange, the riskiest arrangement of all, produces the strongest bonds of the three.

The mechanism the authors settled on is risk. You cannot demonstrate that you are trustworthy inside a binding agreement, because the agreement did the work; and you cannot receive credit for a gift that was already owed. Only where the return might not come does the act carry information, and Molm and her colleagues argue that acts of voluntary, uncertain reciprocity carry an expressive value over and above whatever is transferred. When the same group made negotiated agreements non-binding, trust did rise—and the trust so produced was, in Molm’s own summary of the program, “less resilient” and “less affect-based”: quick to collapse at the first lapse, and never fused to warmth.

One result from that program deserves to be set down on its own, because it is the whole essay in a sentence. Under some conditions, people traded most often with the partner whose resources were worth more, and formed their strongest bonds with the partner whose reciprocity was most constant. They were not the same partner. Value and attachment came apart, and what separated them was not the size of the benefit but the reliability with which it arrived unasked.

IV · What Status Buys

Set the exchange literature aside for a moment and ask the ordinary question, the one every survey of mate preferences has asked since the psychologist David Buss’s thirty-seven cultures: how much does standing matter? Stated preferences give a clear and much-quoted answer, and men and women do reliably differ in how much a partner’s earning prospects and a partner’s looks move their ratings of a hypothetical person on a page. The difficulty is that the page is not a person.

In 2014 the psychologists Paul Eastwick, Laura Luchies, Eli Finkel, and Lucy Hunt pooled ninety-seven data sets in Psychological Bulletin and asked what those preferences predict once someone is actually being evaluated. Physical attractiveness predicted romantic evaluations at a correlation of about .40—a solid, unsurprising effect—and it did so for both sexes. Earning prospects predicted romantic evaluations at about .10, a small effect, and it did so for both sexes. The difference between men and women in these correlations came to .03, and was, in the authors’ phrase, uniformly nonsignificant. The finding is not that the stated difference is imaginary; the authors are explicit that it is real and reliable when people rate hypothetical targets. The finding is that it does not survive contact.

Why it does not survive is the most useful thing in the paper. In an experiment Eastwick ran with Finkel and Alice Eagly, single participants read a three-trait profile of a person they were about to meet, assembled either from the traits they themselves had rated most important in a partner weeks earlier or from the traits they had rated least important. Reading the profile, they wanted the ideal match more. Then came four or five minutes of deliberately constrained conversation—taking turns describing pictures, the other side’s replies scripted—after which the effect was gone. A second study found the reason, and it is almost comic in its ordinariness. On paper, a participant who wanted an outspoken partner read “outspoken” as frank, and one who did not read it as tactless. After five minutes of actual conversation, both read it the same way. The trait had not changed. Its meaning had, and it had changed toward whatever the live person seemed to be.

This is not a demotion of status so much as a relocation of it. The Toronto sociologist Adam Isaiah Green, extending Bourdieu, argues that desirability is not a property people carry around but a property of a person in a field: any given sexual field has its own structure of desire, its own currency, its own hierarchy, and the currencies do not convert at the boundary. A great deal of what looks like status effect is really the effect of being fluent in one particular room. Which returns us to Simmel’s circles from the other direction: if a life is a set of separate relationships, it is also a set of separate fields, and the same person is differently valuable in each of them. Status is real. It is simply local, and it is priced in a currency that stops working two rooms away.

V · The Longer Letter

Deference is difficult to observe in the wild, because people are shy about it and because it usually takes the form of things not said. Online dating solves the observation problem by writing it down. In 2018 the sociologist Elizabeth Bruch and the physicist Mark Newman published in Science Advances an analysis of messaging on a large free dating site across four American cities—New York, Boston, Chicago, and Seattle—using a ranking algorithm in which a person’s desirability is a function of how many messages they receive and how desirable the senders are. It is a circular definition on purpose, and it produces a stable hierarchy that turns out to look much the same in all four cities.

Two findings from that paper belong in this essay. The first is that almost nobody writes across; almost everybody writes up. Men, on average, messaged people twenty-six per cent further up the ranking than themselves; women, twenty-three per cent. The second is what they did when they wrote up. Both sexes wrote substantially longer messages to more desirable partners—in some cases, the authors say, up to twice as long. This is deference in its purest measurable form: additional unrequited effort, calibrated to the size of the gap.

It does not work. Of the eight groups the study distinguishes—men and women in each of four cities—only Seattle men showed any payoff to writing at greater length. And the ceiling is severe. Men, the authors report, are more than twice as likely to get a reply from a woman less desirable than themselves than from a more desirable one, and for messages sent to more desirable women the reply rate never rises above twenty-one per cent. At least four-fifths of what a man writes up the hierarchy goes unanswered, and the share does not improve with effort.

It is tempting to read this as a tragedy of the market, and it is at least a considerable waste of prose. Through Molm it is something more specific. The long message to a more desirable stranger is a unilateral, unnegotiated, wholly unguaranteed transfer—which is precisely the form of exchange her experiments find most productive of trust and regard. The difference is that in her laboratory the gift arrives inside a relation that will continue. Here it arrives at a stranger who owes nothing, cannot be observed to withhold, and will not be seen again. Risk without a relation is not a gift. It is postage.

VI · Nine Months

A market requires agreement about what things are worth. Take away the agreement and there is no market, only a set of private opinions. This is the reason the sorting of couples by physical attractiveness—among the most replicated observations in the field, at correlations that Hunt and her colleagues put at around .30 to .40 in earlier meta-analyses—has always been read as evidence of competition: people can only be outbid for a partner if enough bidders concur about who the desirable partners are.

In 2015 Lucy Hunt, with Eastwick and Finkel, published a study in Psychological Science that took this seriously enough to look for the condition under which it fails. They collected a hundred and sixty-seven couples, had coders rate each partner’s attractiveness, and asked how long the two had known each other before they began dating. The interaction was significant—the coefficient was −0.21, t(163) = −3.15, p = .002—and its sign is the interesting part. The longer two people had known each other before anything started, the less they looked like a matched pair. Among couples who began dating within a month of meeting, the predicted correlation between their attractiveness ratings was .72, which is very high indeed. Past roughly nine months of prior acquaintance it was no longer statistically distinguishable from zero. Couples who had been friends first showed the same collapse.

This is a cross-sectional study of couples who already existed, resting on their own recollection of how long they had known each other, and it cannot show that acquaintance causes the difference. What it can show is that the market’s grip is not a constant. Consensus about who is desirable is strong among strangers and weak among people who have watched each other for a year, and the reason is presumably the same mechanism Eastwick found in his five minutes of scripted conversation: given enough contact, the meaning of a trait migrates toward the particular person carrying it, and a shared scale quietly turns into a private one.

So timing operates at two quite different scales, and they point the same way. Inside a single exchange, Bourdieu’s interval is what lets a transfer read as a gift instead of a settlement. Across a life, the interval before anything begins is what dissolves the market’s valuation and replaces it with an idiosyncratic one. Both are cases of the same thing: value that has been left alone long enough to stop being comparable. The corollary is unflattering to almost every institution built for meeting people, all of which are designed to compress exactly that interval.

VII · Who Listens

If the form of a transfer matters more than its size, conversation ought to be the cleanest case available, since talk is a benefit that costs almost nothing and can be given in any pattern at all. In 2013 the sociologist Susan Sprecher, with Stanislav Treger and Joshua Wondra, put fifty-nine pairs of strangers through two twelve-minute structured disclosure sessions over a video link. In the first session one member of each pair did the disclosing and the other listened; in the second they swapped.

The listeners liked the disclosers more than the disclosers liked the listeners. After the first session, listeners reported greater liking (5.73 against 5.43), greater closeness (3.95 against 3.25), more enjoyment (4.98 against 4.56), and—the largest gap—more perceived similarity (4.29 against 3.75, at p = .003). These are modest effects, in the neighborhood of a third of a standard deviation. What makes them worth reporting is what happened next: after the roles reversed, the differences disappeared. The advantage did not belong to a kind of person. It belonged to a position, and it moved when the position moved.

The companion paper, published the same year in the Journal of Experimental Social Psychology, found the practical corollary. Pairs who alternated disclosure within a conversation ended up liking each other more than pairs who did it in blocks—one talking through the first session, the other through the second—even though the same amount was said by the same people. The sample was overwhelmingly female, which limits what can be concluded about men. Still, the shape is Molm’s shape: same value, different form, different feeling.

Whether this reaches desire is a separate question, and the answer is a qualified yes with a complication. The psychologists Gurit Birnbaum and Harry Reis, in a series of studies on first meetings, found that perceiving a stranger as responsive raised sexual interest in them—but chiefly among people low in attachment avoidance, so the effect is conditional on who is receiving it. Then, in 2014, Birnbaum, Tsachi Ein-Dor, Reis, and Noam Segal reported the sharpest sex difference in this essay. Men who met a responsive woman rated her as more feminine and, through that, as more sexually attractive. Women who met a responsive man did not show the corresponding pattern.

Two things should be said about that result before anyone builds on it. It is bounded to initial acquaintanceships—a first meeting with a stranger, not a relationship. And the variable carrying the effect is perceived gender typicality: responsiveness raised attraction in men because it made the woman seem more conventionally feminine. That is a fact about a prevailing script, not about a mechanism in the body, and a script is the kind of thing that differs by decade and by room. The finding is real. It is also, quite precisely, a measurement of what warmth is currently taken to mean when a woman does it and when a man does it.

VIII · The Invisible Half

The assignment behind this essay assumed that giving and receiving are unevenly distributed, and on the question of who does the giving the sociology is not equivocal. In 2019 the sociologist Allison Daminger published in the American Sociological Review an interview study of thirty-five couples—seventy conversations—about a dimension of domestic labor that time-use surveys miss because it leaves no trace: the work of anticipating what will be needed, identifying the options, deciding among them, and monitoring whether it happened. She calls it cognitive labor. Women in different-sex couples did more of it, and, more pointedly, more of its least visible and least powerful forms—the anticipating and the monitoring rather than the deciding. It is a qualitative study of a well-educated American sample and cannot support a population estimate. It describes a structure, and the structure is recognizable.

What that asymmetry does to desire is where the literature stops agreeing with itself, and the disagreement is instructive enough to be worth leaving open. In 2013 the sociologists Sabino Kornrich, Julie Brines, and Katrina Leupp reported in the same journal that married couples with more traditional divisions of core housework had sex more often—a result that traveled around the world’s newspapers under headlines about doing the dishes. Three years later the sociologist Daniel Carlson and colleagues, working with data from 2006 rather than the late nineteen-eighties and 1992, reported essentially the opposite: that couples sharing the work, including the care of children, did better on sexual frequency and on satisfaction. The two teams are looking at different decades through different instruments. Neither has been shown to be wrong. Anyone who tells you the question is settled has read one of them.

It is worth holding that uncertainty next to what happened to the most famous sex difference in the whole field. Reporting in 1989 on campus field studies run some years earlier, the psychologists Russell Clark—no relation to the Clark of the pens—and Elaine Hatfield had confederates proposition strangers outright. About three-quarters of the men accepted, and none of the women did, and the result became a fixture of every textbook. In 2011 the psychologist Terri Conley took it apart in the Journal of Personality and Social Psychology. Both sexes, it turned out, rated male proposers as less sexually skilled, less warm, and more dangerous than female ones. Once the proposer was someone attractive, or a close friend believed to be sexually competent, the gender difference in acceptance disappeared. Perceived sexual capability predicted acceptance strongly for women and for men. What had been measured as a difference in desire was substantially a difference in what the offer looked like.

That is the pattern across this literature, and the psychologists Jennifer Petersen and Janet Hyde, pooling eight hundred and thirty-four samples, found much the same: men report somewhat more sexual experience and more permissive attitudes on most measures, with the clearest gaps in masturbation, pornography use, and attitudes toward casual sex, and a great many other differences small enough to be uninteresting. The durable asymmetry in this essay is not in wanting. It is in who does the relational work—who anticipates, who monitors, who keeps the calendar of other people’s needs. And on the question of what that costs the person doing it, in the currency this essay is about, the evidence is genuinely divided.

IX · Consumption

The word in the assignment was consumption, and it carries a theory. The sociologist Eva Illouz gave the best version of it in 1997, arguing that romance in the twentieth century was built by two movements at once—the commodification of romance, in which courtship acquired a price list of dinners and outings and holidays, and the romanticization of commodities, in which the price list acquired an aura. The strong form arrived seven years later, when the psychologists Roy Baumeister and Kathleen Vohs proposed sexual economics theory in Personality and Social Psychology Review: a heterosexual community modelled as a market in which men offer resources to obtain sex from women, with women cast as sellers and men as buyers.

It is a genuinely predictive model in places, and this essay has already used the places. Bruch and Newman’s hierarchy is a market. So is assortative sorting on attractiveness, and so is the fact that its grip weakens exactly when consensus does. But the theory has taken serious fire, and the most interesting round is empirical rather than philosophical. The psychologist Laurie Rudman has argued at length that its central tenets do not survive the data. And in 2017 Janell Fetterolf and Rudman ran an experiment in which students were randomly assigned to watch a video explaining sexual economics theory; afterwards they endorsed sexual exchange more, and held more adversarial views of relations between men and women, than students who had not. Whatever else it is, the market model is a thing that changes people who learn it. Theories of gravitation do not do that.

The economic sociologist Viviana Zelizer offered, in The Purchase of Intimacy, the exit from a debate that had been stuck between two bad positions—what she calls hostile worlds, in which money contaminates intimacy and must be kept out, and nothing-but, in which intimacy is nothing but a market wearing a hat. Her alternative is that people are perfectly capable of using money and goods to make, sustain, and mark intimate ties, and that most of the work in an intimate relationship consists of matching the medium to the relation: this transfer is a gift, that one is compensation, this one must never be mentioned. Read against Clark’s pens, Zelizer is describing the same act from the outside. Choosing the red pen is matching the medium to the relation. It is not an absence of economics; it is a specific and demanding piece of it.

Which leaves the word roughly half deserved. The market metaphor is strong on search, sorting, and hierarchy—on how people find each other and how brutally they are ranked while doing it. It is weak on the good, because the good is not sex or status or attention but the relation itself, and a relation is destroyed by exactly the operations that define a market: immediate settlement, guaranteed return, comparability across vendors. This is also why the essay declines the sociologist Catherine Hakim’s “erotic capital,” her proposal of desirability as a fourth personal asset alongside the economic, cultural, and social. Green’s objection is the right one: the concept is overstretched, internally inconsistent, and asociological, and it treats as a portable holding something that a decade of field research says is only ever valuable somewhere.

One admission belongs here, and it is the largest limitation in the piece. The assignment’s premise is a person standing in several separate relationships at once, on different terms in each. No study located for this essay does that. The exchange experiments give a subject one partner, or a network of interchangeable ones. The attraction experiments give a subject one target. Nobody has taken a person who defers in one relation and is deferred to in another and measured what either arrangement does to their wanting of a third party. The web is the premise of the question and the blind spot of the field, and that is worth saying plainly rather than filling in.

X · What New York Gives

New York makes negotiated exchange unusually easy to see. The restaurant tip is a transfer agreed by custom and settled at the end of an encounter. The check split at a table and the favor called in next week have the same immediate accounting. Molm’s laboratory work warns against mistaking that clarity for the whole of social life: a transaction can settle accounts without creating the kind of trust that grows when nobody knows exactly who will repay whom.

That second form of exchange exists inside the city too. A friend introduces a friend without a commission; a neighbor helps before knowing whether help will return; a group keeps a place open for someone who has little to offer today. These are ordinary possibilities, not claims that New Yorkers give more or less than residents elsewhere. Their point in this essay is structural. Generalized exchange needs a network and an interval. If people disappear after every transaction, the next turn never arrives.

New York contains both conditions at once: encounters among strangers and circles that meet repeatedly. The former make rank and price legible; the latter give a person time to become known for something other than rank. The argument from Molm is that wanting somebody may depend less on how much can be extracted from a single encounter than on whether both people expect to remain in one another’s world. No experiment cited here measures the city as a whole. The question is what kind of ledger two people choose to keep.

XI · The Color of the Pen

Return to the man with two pens. He has not decided anything. He has been told one sentence about a stranger, and his hand has gone to the red or the black, and the choice has reported a classification he did not know he had made. This is the useful thing about Clark’s design and the reason it opens this essay: it catches the ledger being opened, which normally happens below the level at which anyone could describe it.

Set out plainly, the argument runs like this. Relationships run on two incompatible accounting rules, and the same act—repayment, a favor, a long message—means opposite things under each. Which rule applies is settled early and mostly without discussion. The rules govern separate circles that are kept separate on purpose, because the currencies do not convert. Within any of them, the form in which a benefit arrives matters more than its size: unilateral, unguaranteed giving produces trust and warmth that an identical bargain does not, and the mechanism is that only an unsecured act carries information. Status is real and local and much weaker in live evaluation than on paper. Timing is not a tactic but a structural condition, at the scale of the interval and at the scale of the year. And the difference between the sexes that the evidence supports most firmly is not a difference in wanting but a difference in who does the anticipating.

The limits are severe and worth stating as bluntly as the findings. The exchange experiments are laboratory studies with abstract resources and assigned partners; that they bear on sexual attraction is this essay’s argument, not their result. Clark’s pens are from 1984, with small samples of male undergraduates and one confederate. The mating literature is overwhelmingly American, student or online-panel, cross-sectional, and heterosexual, because that is who was sampled. Two of the load-bearing sociological sources—Bourdieu on the interval, Zelizer on relational work—are arguments rather than measurements and are used as such. And the arrangement the whole question is about, one person standing in several unlike relationships at once, has not been studied by anybody.

There is one further thing the evidence says, and it has the awkward property of being useless. If what converts a transfer into a bond is that it might not have been returned, then no version of it can be performed to obtain the bond, because a performance is a hedge and a hedge is exactly the thing the finding excludes. Molm’s subjects did not build trust by choosing to expose themselves; they built it by being exposed. The man in the experiment cannot pick the red pen in order to seem communal. He can only already be the sort of person who, told a stranger is new in town and hoping to meet people, does not think to count.

Notes on Sources

Clark, M. S. (1984). Record keeping in two types of relationships. Journal of Personality and Social Psychology, 47(3), 549–557. Full text read. Source of every percentage in Part I and of Fig. 1: Study 1, thirty-three male undergraduates, 88.2 per cent against 12.5 per cent, z = 9.42; Study 2, 94.4 against 42; Study 3, ninety against thirty for pairs.

Clark, M. S., & Mills, J. (1979). Interpersonal attraction in exchange and communal relationships. Journal of Personality and Social Psychology, 37(1), 12–24.† (The original is a scanned document; its design and direction of result are taken from the published abstract and from Clark’s own 1984 restatement of it. No cell means are quoted here, because none could be read.) Clark, M. S., Mills, J., & Powell, M. C. (1986). Keeping track of needs in communal and exchange relationships. JPSP. Clark, M. S., Mills, J., & Corcoran, D. (1989). Keeping track of needs and inputs of friends and strangers. Personality and Social Psychology Bulletin, 15, 533–542. Both cited for direction only.

Molm, L. D. (2010). The structure of reciprocity. Social Psychology Quarterly, 73(2), 119–131. Cooley-Mead Award address, delivered 9 August 2009; full text read. The summary source for Part III, including the definitions of the three forms, the independence claim, and the quoted characterization of trust under non-binding agreements as “less resilient” and “less affect-based”. The individual findings are: Molm, L. D., Takahashi, N., & Peterson, G. (2000). Risk and trust in social exchange. American Journal of Sociology, 105(5), 1396–1427. Molm, L. D., Takahashi, N., & Peterson, G. (2003). In the eye of the beholder: procedural justice in social exchange. American Sociological Review, 68(1), 128–152. Molm, L. D., Collett, J. L., & Schaefer, D. R. (2006). Conflict and fairness in social exchange. Social Forces, 84, 2325–2346. Molm, L. D., Collett, J. L., & Schaefer, D. R. (2007). Building solidarity through generalized exchange. American Journal of Sociology, 113(1), 205–242. Molm, L. D., Schaefer, D. R., & Collett, J. L. (2007). The value of reciprocity. Social Psychology Quarterly, 70(2), 199–217 — source of the trade-with-one, bond-with-another result.

Bourdieu, P. (1977). Outline of a Theory of Practice, trans. R. Nice. Cambridge University Press. Used as theory and named as theory; the argument about the interval is paraphrased rather than quoted, the 1977 English text not having been checked line by line.

Eastwick, P. W., Luchies, L. B., Finkel, E. J., & Hunt, L. L. (2014). The predictive validity of ideal partner preferences: a review and meta-analysis. Psychological Bulletin, 140(3), 623–665. Full text read. k = 97; physical attractiveness r ≈ .40 and earning prospects r ≈ .10, for both sexes; sex difference in the correlations .03 and, in the authors’ words, “uniformly nonsignificant”. The Eastwick, Finkel, and Eagly (2011) confederate studies, including the outspoken/frank/tactless result and the change-of-meaning mediation, are reported in full in that paper’s Section 1.

Green, A. I. (2008). The social organization of desire: the sexual fields approach. Sociological Theory, 26(1), 25–50; and Sexual Fields: Toward a Sociology of Collective Sexual Life (University of Chicago Press, 2014). Theory. Green, A. I. (2013). “Erotic capital” and the power of desirability. Sexualities, 16 — source of the objection to Hakim, paraphrased here from its abstract rather than quoted, C. (2010). Erotic capital. European Sociological Review, 26(5), 499–518.† (Hakim’s concept is contested at the level of construction, not merely of evidence.)

Bruch, E. E., & Newman, M. E. J. (2018). Aspirational pursuit of mates in online dating markets. Science Advances, 4(8). New York, Boston, Chicago, and Seattle. The twenty-six and twenty-three per cent averages, the “up to twice as long” messages, the Seattle-men exception, and the statement that the reply rate to more desirable women “never rises above 21%” are all the authors’. Their age–desirability curves are not reused here; this magazine spent them in The Measure of a Man.

Hunt, L. L., Eastwick, P. W., & Finkel, E. J. (2015). Leveling the playing field: longer acquaintance predicts reduced assortative mating on attractiveness. Psychological Science, 26(7), 1046–1053. Full text read. N = 167 couples; interaction E = −0.21, t(163) = −3.15, p = .002; predicted r = .72 at zero acquaintance; Johnson-Neyman boundary at 9.9 months. Cross-sectional and retrospective, as the text says.

Sprecher, S., Treger, S., & Wondra, J. D. (2013). Effects of self-disclosure role on liking, closeness, and other impressions in get-acquainted interactions. Journal of Social and Personal Relationships, 30(4), 497–514. N = 118 in fifty-nine dyads, 81.4 per cent female; source of the four means and of the reversal after roles swapped. Sprecher, S., Treger, S., & Wondra, J. D. (2013). Taking turns: reciprocal self-disclosure promotes liking in initial interactions. Journal of Experimental Social Psychology, 49, 860–866.

Birnbaum, G. E., & Reis, H. T. (2012). When does responsiveness pique sexual interest? Personality and Social Psychology Bulletin, 38(7), 946–958 — attachment avoidance as moderator. Birnbaum, G. E., Ein-Dor, T., Reis, H. T., & Segal, N. (2014). Why do men prefer nice women? Gender typicality mediates the effect of responsiveness on perceived attractiveness in initial acquaintanceships. Personality and Social Psychology Bulletin, 40(10), 1341–1353.† (Bounded to first meetings; the mediator is perceived gender typicality, a cultural variable, which the text states.)

Daminger, A. (2019). The cognitive dimension of household labor. American Sociological Review, 84(4), 609–633. Seventy interviews, thirty-five couples; qualitative, and cited for the structure rather than for any prevalence estimate.

Kornrich, S., Brines, J., & Leupp, K. (2013). Egalitarianism, housework, and sexual frequency in marriage. American Sociological Review, 78(1), 26–50, on National Survey of Families and Households data from the late 1980s and 1992.† Against it: Carlson, D. L., Hanson, S., & Fitzroy, A. (2016). The division of child care, sexual intimacy, and relationship quality in couples. Gender & Society, 30(3), 442–466, and the companion reexamination in the Journal of Marriage and Family, on 2006 data, N = 974.† (Two literatures in open contradiction; the essay reports both and settles neither.)

Conley, T. D. (2011). Perceived proposer personality characteristics and gender differences in acceptance of casual sex offers. Journal of Personality and Social Psychology, 100(2), 309–329, addressed to Clark, R. D., & Hatfield, E. (1989), Journal of Psychology & Human Sexuality, 2(1). Petersen, J. L., & Hyde, J. S. (2010). A meta-analytic review of research on gender differences in sexuality, 1993–2007. Psychological Bulletin, 136(1), 21–38. Eight hundred and thirty-four samples and seven national data sets; individual effect sizes are not quoted here because no accessible full text could be checked, so only the authors’ own characterization is used.

Baumeister, R. F., & Vohs, K. D. (2004). Sexual economics: sex as female resource for social exchange in heterosexual interactions. Personality and Social Psychology Review, 8(4).† Against it: Rudman, L. A. (2017). Myths of sexual economics theory: implications for gender equality. Psychology of Women Quarterly, 41; and Fetterolf, J. C., & Rudman, L. A. (2017). Exposure to sexual economics theory promotes a hostile view of heterosexual relationships. Psychology of Women Quarterly, 41 — the randomized study reported in Part IX.

Zelizer, V. A. (2005). The Purchase of Intimacy. Princeton University Press — hostile worlds, nothing-but, connected lives. Illouz, E. (1997). Consuming the Romantic Utopia. University of California Press. Simmel, G. (1908), “Die Kreuzung sozialer Kreise”, in Soziologie; trans. R. Bendix as “The Web of Group-Affiliations” (Free Press, 1955) — the German is closer to “the intersection of social circles”, as the text notes. Goffman, E. (1959). The Presentation of Self in Everyday Life. Doubleday. All four are used as framing, never as evidence for a quantity.

† marks findings that are contested, in open disagreement with a rival literature, bounded more narrowly than they are usually quoted, or resting on a source weaker than a primary publication. Figures 2 and 3 are qualitative: the orderings, ceilings, and named values are the cited authors’, the shapes are drawn to the argument. Figures 1 and 4 carry published values and name the study in the caption. Part X is an argument from institutional form, not a measurement; no citywide measure of these exchanges was found. And no located study places one person in several simultaneous relationships on different terms and measures desire in one of them, which is the arrangement the whole essay is about.

Drafted with AI (Claude, Anthropic) as a private research synthesis, August 22, 2026; the sources above are real publications, read in full where they could be reached and marked where they could not, contested science is flagged rather than smoothed, and the study the argument most wanted turned out never to have been run. Set, with affection and without affiliation, in the manner of a certain Manhattan weekly.New York revision drafted with AI on October 12, 2026; editorial review pending. ◆ ◆ ◆