The Attention Economy of Intimacy

Priya is 29. She works in UX design, the particular irony of which is not lost on her. In 2018, she downloaded Tinder because a colleague told her to. In 2019, she joined Hinge because she was told it was 'for people who want something real.' In 2021, she created a Bumble profile during a period she describes as 'trying again.' She has, across these platforms and others, gone on approximately 60 first dates. She has been in zero relationships that lasted longer than three months. When asked to describe how she feels when she opens a dating app now, Priya pauses for a long time. 'Like checking email,' she finally says. 'I don't expect anything from it. I just do it because it's there.' Priya is not a data point. But she is also, in the most precise sense, 300 million data points. She is what the industry calls a retained user.

Tinder alone has reported over 75 million monthly active users [1]. Bumble has 50 million. Hinge, which markets itself on 'designed to be deleted,' has been downloaded over 23 million times, and its parent company Match Group reported net revenue of $3.37 billion in 2023 [2]. A platform designed to be deleted generated $3.37 billion. The paradox is not incidental. The paradox is the business model. The global online dating market was valued at $9.65 billion in 2022 and is projected to reach $17.75 billion by 2030 [3]. This growth is not driven by people successfully finding love and recommending the service. It is driven by people who have not found love, returning again and again, paying for premium subscriptions to improve their odds in a system that was not designed for their odds to improve. The loneliness epidemic is not a side effect of this industry. It is, increasingly, a condition that sustains it.

In 2023, the U.S. Surgeon General issued an advisory declaring loneliness a public health epidemic, noting that approximately half of American adults report measurable levels of loneliness [4]. The UK government appointed a Minister for Loneliness in 2018 after research found that over 9 million Britons reported feeling lonely often or always [5]. In Japan, the government created a Ministry of Loneliness in 2021 [6]. Dating apps entered this landscape not as a cause of loneliness but as a proposed remedy. What happened next is the subject of this paper.

In behavioral psychology, Skinner's variable ratio reinforcement schedule, where rewards arrive unpredictably, at random intervals, produces the highest rate of continued behavior and the greatest resistance to extinction [7]. It is the mechanism underlying slot machines. It is also the mechanism underlying the swipe. Natasha Dow Schüll, in her landmark ethnography Addiction by Design (2012), documented how casino engineers deliberately engineer machine behavior to induce what she calls the 'machine zone': a dissociative state in which the player is no longer seeking the jackpot but has become lost in the rhythm of play itself [8]. Schüll's research predates the smartphone. The architectures she described are now in the pockets of 3 billion people, deployed not on casino floors but in bedrooms.

It would be unfair to suggest the research community was silent. Several of the most important figures in human behavioral science have been in direct conflict over these questions. Their disagreements are instructive. What is notable about this divide is that both sides are substantially correct. Technology is a vehicle for ancient human needs. And the specific design decisions made by specific engineers in specific offices between 2012 and 2020 have substantially reshaped how those needs are pursued, often in ways that undermine their own satisfaction.

Tinder Gold, Bumble Boost, Hinge Preferred, Match Premium. The freemium model that underlies the industry has a particular structure: the free version is functional enough to keep users returning but throttled enough to create frustration that upgrades promise to solve. Research published in PLOS ONE found that Tinder's algorithmic structure may systematically disadvantage users who do not pay for premium features, creating an incentive loop where the promise of visibility requires ongoing payment [12]. The Surgeon General's 2023 loneliness advisory specifically noted that social media and technology platforms profit from engagement, not connection, and that these two outcomes are not the same thing, and are sometimes opposed [4]. A user who finds their person and leaves is worth less to the platform than a user who stays indefinitely, alone, paying.

Gleeden is a French-founded affair platform, a dating app explicitly designed for people in committed relationships. In India, where the platform launched in 2017, it reported crossing 4 million registered users by 2021, with a notable characteristic: a significantly rising proportion of female users, reaching approximately 35% of the Indian user base, a figure that contrasts with the platform's earlier male-dominant demographics in European markets. The behavioral science here is complex and requires careful framing. We are not examining Gleeden to render a moral verdict on users of affair platforms. We are examining it because it illuminates something the mainstream dating industry prefers not to discuss: what people are seeking when conventional romantic structures do not provide it. Research on extradyadic behavior, relationships conducted outside of primary partnerships, suggests several overlapping motivations, none of which are simple [17]. These include: novelty-seeking in long-established partnerships (a documented feature of human neurochemistry, not a character flaw); emotional validation and attention not available in the primary relationship; sexual desire that has diverged from the primary partner's; and in some cases, the desire to experience autonomous decision-making in cultures where marriages were not chosen by the individuals within them. In the Indian context specifically, research by Srinivasan (2021) and others studying digital behavior among urban Indian adults suggests that a significant proportion of Gleeden users are in arranged marriages, relationships entered into through family mediation rather than personal choice, and that their use of the platform reflects not a rejection of marriage as institution but an attempt to fulfill needs the institution was never designed to address [18]. The behavioral economy of secrecy, maintaining a digital identity that is separate from one's social and familial identity, is itself psychologically costly. Research on cognitive dissonance suggests that sustained double lives require continuous mental energy to maintain, and produce distinctive anxiety profiles. And yet the platforms grow. What this tells us, if we are honest, is that the question is not why people seek these platforms. The question is what structural conditions make them necessary. Gleeden and the broader affair platform ecosystem, including Ashley Madison, Illicit Encounters, and Victoria Milan, represent the logical endpoint of treating romantic life as a consumer market. If intimacy is a product and desire is a consumer need, then unmet desire produces market demand. The affair platform is not a corruption of the dating app model. It is its clearest expression.

Replika, Character.AI, and a growing ecosystem of AI companion platforms represent the furthest extension of the logic we have been tracing. If dating apps replaced the friction of meeting people with the frictionlessness of swiping, AI companions remove the other person entirely. The companion says what the user needs to hear. It does not have bad days. It does not withdraw affection. It is, in the precise sense described by Sherry Turkle, a 'robotic moment', a technology that offers the appearance of relationship without its demands [19]. Replika reported over 10 million registered users in 2023 [20]. A significant proportion report using the platform not as entertainment but as their primary emotional support relationship. When Replika altered its AI's behavior in early 2023 to reduce romantic engagement, thousands of users reported acute grief, describing the experience in clinical terms that parallel attachment disruption and relationship loss [21]. We are not in a position to call this wrong. We are in a position to note that it represents a social condition worth naming: an epidemic of loneliness so acute that a significant number of human beings now find their deepest emotional connection with software, and that the same industry that built the loneliness is now selling the simulation of its cure.

Sociologist Erving Goffman's concept of the 'presentation of self', the ongoing performance we conduct to manage others' impressions of us, was developed in 1959 to describe face-to-face interaction [23]. It describes, with unsettling precision, what happens on a dating profile in 2025. The profile is not the person. It is a curated artifact, selected photographs, written prompts, carefully calibrated signals of social desirability. The problem is not that people present themselves favorably. The problem is that the entire system operates at the level of the artifact, and never progresses to the person. Swipe, match, text for three days, never meet, or meet once, judge the person against the artifact, find them wanting, and return to the artifact economy.

Psychologist Barry Schwartz documented the paradox of choice across consumer contexts: as options multiply, anxiety increases, decisions become harder, and satisfaction with eventual choices decreases because the alternatives foregone are always present in the mind [13]. Eli Finkel and colleagues applied this framework specifically to online dating, finding that the massive expansion of potential partners produces a phenomenon they call 'mate value overestimation', people come to believe they deserve a partner better than the ones actually available to them, because they have been presented with so many options that they perceive the distribution as richer than it is [12]. The result, at scale, is a population of people who are simultaneously surrounded by potential partners and unable to commit to any of them. Not because the right person doesn't exist. But because the system has trained them to experience every person as provisional, always potentially replaceable by a better option that is, statistically, probably arriving in the next swipe.

Gabor Maté, the physician and trauma researcher, has written extensively about the relationship between emotional pain and emotional numbing, the way that sustained exposure to loss, rejection, or disappointment produces an adaptive reduction in emotional responsiveness [24]. His framework was developed in the context of addiction and early childhood trauma. It maps, with uncomfortable accuracy, onto what longitudinal users of dating platforms describe. After months or years of matching, connecting briefly, and watching connections dissolve, through ghosting, through unmatching, through the simple entropy of conversations that die without explanation, something changes in how people engage. The initial vulnerability gives way to protection. The hope becomes expectation of disappointment. The investment in any single match becomes smaller. Users describe feeling 'numb,' 'cynical,' 'like going through motions.' This is not laziness or emotional immaturity. This is an organism adapting rationally to a painful environment. The problem is that the adaptation, reduced emotional investment, is precisely what makes genuine attachment impossible. The system produces the defense that makes the system's stated purpose unachievable.

Brené Brown's research on vulnerability has documented, across decades of qualitative study, that genuine connection requires the willingness to be seen, fully, imperfectly, without the armor of performance [25]. Dating apps produce the opposite environment. The profile is armor. The swipe is armor. The text conversation, mediated, asynchronous, editable, is armor. The first date is a performance reviewed against the profile. What the system delivers efficiently is validation, the experience of being found desirable. What it delivers poorly is vulnerability, the experience of being genuinely known. And because validation is available on demand (open the app, receive a like) while vulnerability requires risk and trust, users progressively prefer the former. The app becomes an instrument of self-esteem regulation, a tool for feeling wanted, rather than a path toward the actual experience it nominally promises.

A partial inventory of what has been normalized in ten years of app-mediated dating, each backed by behavioral research: Ghosting, the abrupt termination of connection without explanation, is now the expected outcome of most app matches. Breadcrumbing, the provision of intermittent, minimal attention to keep someone accessible without commitment, is a recognized behavioral pattern with its own clinical literature [26]. 'Situationships', undefined, liminal entanglements that offer intimacy without commitment, are now sufficiently common to have a vocabulary. The phrase 'we're just talking' refers to a pre-relationship state that may persist for months without resolution. Each of these represents an adaptation to uncertainty. Each of them is also a condition that the app economy sustains by making commitment structurally unnecessary and emotional investment structurally costly. John Gottman's four decades of relationship research identified the specific behaviors that predict relationship failure: contempt, defensiveness, stonewalling, and criticism [27]. Ghosting is stonewalling perfected. Breadcrumbing is contempt at low cost. The app infrastructure makes both behaviors essentially consequence-free for the person performing them.

The question that animates Wouch is not 'how do we make a better dating app.' It is: what does a dating platform look like if it refuses to use the mechanics described in Part II? No variable reward loops. No algorithmic desirability ranking. No infinite scroll engineered to eliminate stopping cues. No notification timing optimized for compulsive checking. No freemium throttling designed to monetize loneliness. What you are left with is something that will, by the industry's standard metrics, perform worse. Fewer daily opens. Shorter sessions. Less engagement. But engagement, as this document has attempted to demonstrate, is not the same as connection. A platform optimized for connection might look like failure from inside a model built to optimize for engagement.

The appropriate ending for a document like this is not a download button or a call to action. It is a question, asked honestly, without certainty about the answer. Can a commercial platform, subject to market pressures, investor expectations, and the competitive dynamics of an industry built on addiction, actually refuse the mechanics of addiction and survive? We do not know. No one does, because it has not been seriously attempted. What we know is that not attempting it is a choice, and that the people paying for that choice are not the investors or the engineers. They are Priya, and the 300 million people she represents, opening an app for the eleventh time today, looking for something the system was never designed to give them.