Why We Cannot Find a Partner: How Recommendation Engines Broke Human Selection

Have we become too good at choosing, or have we forgotten how to choose at all? When we open a dating application and face a sequence of faces, each accompanied by a brief biography and a photograph, are we selecting a human being or shopping for a product with the wrong specifications? The question sits uncomfortably at the intersection of technology, psychology and the most intimate human experience: the search for love and partnership.

The paradox is not new, though its manifestation is. Barry Schwartz, the psychologist who articulated the paradox of choice, observed that the expansion of options, which ought to liberate us, often paralyses us instead. When we face unlimited choices, we do not become happier. We become more anxious, more prone to regret, more inclined to wonder whether we have selected the right option from the infinite menu before us. Schwartz noted that we experience a kind of existential vertigo in the face of abundance. Yet dating applications have taken this paradox and industrialised it. They have transformed it into architecture. They have made it the fundamental operating system of modern romance.

Consider the elegant experiment conducted by Sheena Iyengar and Mark Lepper in the 1990s, before the algorithmic era of dating truly took hold. They placed jam samples in a supermarket: one day six varieties, another day twenty four. When faced with six options, customers were more likely to purchase. When faced with twenty four, they were overwhelmed and less likely to buy. The abundance itself became paralytica burden rather than a gift. Now imagine this experiment not in a supermarket but in the domain of human partnership. Imagine the shelf not containing six options but six hundred, or six thousand. Imagine the shopping not taking place once but continuously, with the shelves refreshing every day. Imagine believing, as an algorithm might suggest to you, that the perfect option exists somewhere in that infinite inventory if only you keep searching.

The distinction between maximisers and satisficers, a concept Schwartz developed, becomes crucial here. A satisficer sets a threshold of acceptable qualities and chooses the first option that meets that threshold. A maximiser continues searching for the best possible option, the one that optimises across every relevant dimension. In traditional contexts, without unlimited options, being a maximiser was merely ambitious. In a context of infinite choice, being a maximiser becomes a form of suffering. The maximiser on a dating application faces not just the current pool but the knowledge that another scroll, another session, another day might yield someone more beautiful, more intelligent, more aligned with some imagined ideal. The fence is never closed. The shopping never ends.

This is the trap into which recommendation engines have led us. They are not neutral tools. They are persuasive instruments designed to keep us engaged, to encourage us to swipe, to match, to message. They commodify human beings in a manner we would never tolerate in other spheres of life. The algorithm does not know love. It knows engagement metrics. It knows retention. It knows the profitable exploitation of human longing. And yet we have internalised its logic so thoroughly that we have begun to think of partnership itself in algorithmic terms: as a problem to be optimised, a selection to be maximised, a product category within which to find the best available item.

What has happened, I suggest, is that we have allowed the language and structure of retail commerce to colonise the vocabulary of intimacy. We speak of typing in criteria, of filtering for specific traits, of swiping through options. We have absorbed the metaphor so completely that it no longer feels like metaphor at all. It feels like reality. The recommendation engine does not merely reflect our preferences; it shapes them. It teaches us to think of other people in terms of attributes to be assessed, boxes to be ticked, standards to be met. It teaches us that somewhere in the system is the person who meets all our criteria, and that our task is simply to find them, to optimise the search, to apply the right filters.

Yet human partnership is not optimisable in this way. It is not a problem that yields to algorithmic solution. The qualities that matter most in a relationship are not easily quantifiable. They include tolerance for ambiguity, the capacity to grow alongside another person, the ability to find meaning in shared limitation rather than unlimited choice. They include humility before the other, a recognition that we do not know what we want, and that what we want might be transformed by encounter. They include, paradoxically, the willingness to be satisfied, to stop searching, to commit to a particular person not because they are the best available option but because commitment itself has value.

The research suggests something troubling. Those who approach dating with a maximising mentality report lower satisfaction even when they find partners. They are more prone to regret. They are more inclined to wonder whether they have missed someone better. The algorithmic mindset, even when it succeeds in producing a match, fails to produce the psychological conditions necessary for contentment. It has taught us to believe that satisfaction lies always just beyond the current option, that the perfect match is always one more swipe away. It has colonised not just the infrastructure of dating but our internal experience of desire itself.

There is a deeper issue at stake. Recommendation engines do not merely present options; they constrain what options exist in the first place. They filter the world through engagement metrics and personalised prediction. They show us people who are statistically similar to people we have previously liked. They narrow the aperture of possibility precisely at the moment when we might most benefit from expansion, from encounter with difference, from the unexpected. They create the illusion of limitless choice whilst actually delivering a narrowed, personalised, algorithmic approximation of choice. They make us lonelier not despite abundance but because of it.

The solution is not simple. We cannot uninvent these technologies. We cannot return to a world without options. But we might cultivate a different relationship to choice itself. We might recognise that satisficing is not a failure of ambition but a prerequisite for contentment. We might acknowledge that the person we are seeking cannot be optimised into existence. We might notice when we are thinking like maximisers and ask whether that mindset serves us. We might, occasionally, swipe left on the algorithm itself, stepping away from the application entirely, meeting people in contexts where they are not ranked, filtered or rated. We might remember that love has never been a shopping problem. It has always been a surrender.

So what can we conclude? The paradox of choice is real, but it is not inevitable. We have absorbed a particular way of thinking about human selection from the architectures that surround us. But we are not condemned to think this way forever. The question is not whether better algorithms exist. The question is whether we can reclaim the capacity to choose, not by optimising, but by learning again to be satisfied.