A Treatise on Online Dating
It was in the vanishing spotlight of a late Tuesday evening when, in the midst of date number four of that week, she tilted her auburn-clad head gently and with a wry smile said, “I knew there was something different about you,” but I knew there wasn’t.
Countless dates later, but no closer to the promised prize - the shimmering embrace of a soulmate — I realized I had a problem. And the problem was dating apps.
That year was a gluttonous parade of dates, a never-ending techno-waltz of swiping and matching played out on mobile screens, bars, and coffee shops in my small town. I fell for the saccharine promise of a better life and true love behind the glass, and it swept me up in its current. No different than any of the other slot-machine junkies, I realized it was all for the next fix.
So I quit online dating, and began an earnest attempt to dismantle it, looking for some shred of insight that would help me understand it’s effect on the people around me, and more importantly, myself.
Dating apps are predicted to make $1.4bn this year while only 1.5% of their user-base is converted to a paid account. Last year, Tinder, the king of dating apps, was the top grossing app in the app store, surpassing media stalwarts like Netflix and Pandora. This year it is expected to gross $800m, with a hefty 40% of that as profit margin. For comparison, that is around the same amount of gross income as Snap, Inc., yet Snap is an unprofitable public company while Tinder is a subsidiary of the public entity Match Group. Dating apps are taking the world by storm and it’s no surprise they are big money.
At their best, dating apps sell technology that creates connections otherwise not possible in the real world. They have more people and more choice, removing inhibitive social barriers and traditional stigmas to interactions between genders. Walking up to somebody at a bar is hard, you’re gambling valuable social capital; the streamlined software version is easier and is far less of a commitment. People who would otherwise never have met are now in happy partnerships and marriages because of dating apps.
Despite their positives, not enough is known about the impact dating apps are having on our brains and our relationships. As with any other technological innovation, dating apps are not free from unintended consequences. Similarly to how smartphones created a new risk factor for distracted driving, dating apps could have negative impacts that are difficult to foresee. And similar to smartphones, drastic technological shifts are more turbulent when they happen over a shorter timeline. The same is true with dating apps. Short development cycles have changed the tools people use to connect with each other at an ever-increasing velocity that makes it difficult for individuals and society to adapt.
Because dating apps are nearing almost ubiquitous adoption with millennials, the purveyors of these apps wield almost hegemonic control over how people date. Unfortunately the goals of dating companies are not always aligned to human needs. The purpose of a dating app will always be to make money. The industry benefits from people spending more time on the app, and most apps are built to keep users swiping, not swipe them a soulmate. Because of this fundamental misalignment, dating apps incorporate manipulative software design that is either directly or indirectly intended to change behavior, all while benefiting the best interests of the company. These designs encourage quick and habitual activity. Their habit forming design— whether intentionally or not — leads to overconsumption, increasing racial biases, and hurting their ability to connect.
Dating app design practices are now sowing the seeds for large scale consequences on a generation of young adults. Every minute the paths of thousands of people are changed as they log in, like, chat, and connect in this wild, unpredictable sociological experiment with one of humanities greatest drives.
Swipe apps like Tinder (where users can swipe a profile right for “like”, left for “pass”) are the most used type of mobile dating app, with Tinder and Bumble combining for 90 million registered users, far surpassing apps without swiping like Hinge. Their roots can be traced back to an early web 1.0 site called “Hot or Not”, founded in the fall of 2000, and one of the web’s first viral sensations. It was a simple website where you could rate the attractiveness of a photo of a person with binary “Hot” or “Not” buttons. Within 8 days of launch, it was drawing an incredible 2 million hits per day. This simple idea was the inspiration for Facemash, Mark Zuckerburg’s Facebook precursor, and YouTube, originally a video version of Hot or Not. As the inspiration for two of the world’s largest tech companies, Hot or Not clearly stumbled upon a popular design premise.
Tinder took this model and made it more addictive with a powerful potential reward. Instead of just rating a profile as “Hot or Not”, users can “match” with the person and potentially get a chance to meet. All of a sudden, the stakes were heightened with an injection of realism and hormones. Now mimicked by almost every dating app in existence, the “match” is the dating app drug. Users receive matches on a variable ratio schedule, each match is celebrated as the reward for multiple “actions” of swiping, providing a powerful dopamine hit to the user. Variable ratio schedules have been shown to be the most addictive schedule for behavioral conditioning, whether the subject is a pigeon, chimpanzee, or human with a fully developed prefrontal cortex. According to one study users were tracked logging into dating apps more than 9 times a day, clearly demonstrating a concerning pattern of addiction.
These variable reward schedules, in addition to creating addictive usage patterns, are designed to move users quickly from profile to profile. The quicker the user swipes left or right the greater volume of profiles they consume, and the more possible rewards for the user. To swipe app companies this is crucial. Increasing the rate of action taken on a double-opt-in dating network (where both users have to take action to “match”) has direct advantages for a dating company. The more actions that a user takes, the more engagement triggers they create. A trigger as termed by the Hooked model is anything that gets the user to return to the product, e.g. push notifications, emails, etc. The more a user swipes and likes, the more notifications and emails they send out, and the more potential triggers they cause. These serve the dual purpose of reinforcing the same primed reward pathways for the users on the receiving end with new “like” notifications.
To magnify this effect, swipe apps use clever design tricks to encourage users to stay in the swiping flow and continue seeking matches. In the example below, after a match, users can “keep playing” or message their match. Both options receive the same design treatment, signifying the options are of equal importance. Continuing to “play” (Tinder), “bumble” (Bumble), or “scout” (The League) benefits the app and triggers more rewards to the user. Conversely, sending a message to a match triggers no reward pathway. In fact, the message could receive no reply at all, actually a negative punishment that deters usage. In our own testing we found this preference to hold true, members report significantly less value from sending a message than receiving a match or receiving an inbound like. While users may prefer the dopamine hit of the match, it is the app’s duty to promote healthy usage patterns, not play to user’s psychological weaknesses. If swipe apps truly wanted to encourage connections between their members, they would discourage or restrict the ability to swipe on other people after a match until the user engages with that match.
Another method towards increasing actions taken per user is to limit the amount of information available when making a choice. On apps like Tinder and Bumble, the personal profile descriptions are limited to a paltry 500 characters or less. Trying to accurately convey the complex subtleties of a person with that amount of characters is an impossible task. Instead of profile space, the majority of initial screen real estate is dedicated to the photo — around 70% for Bumble and close to 100% for Tinder. This plays to our unconscious preferences as humans. Research suggests that humans are hardwired to have incredible facial processing abilities from birth; we can process an individual face at around .17 of a second, and we’re born with both a specialized area of brain and neurons to preform this highly specific task. We’re so exemplary at this task that despite modern computers’ incredible advantage in computations per second, we’re still two or three times quicker than some of the fastest software at facial recognition. Increasing the speed a user can process a profile by reducing the information down to a photo reduces friction, increases number of actions per session, and encourages a user to make rapid-fire decisions.
Inevitably there’s a feeling of consumption that comes with the speed and superficiality of swipe apps. One user, who had never used online dating before she tried Bumble for the first time, told me in an interview: “I didn’t like the swiping, [it felt] objectifying, it felt like it rewired my brain. I never used to go out in the world looking for potential [dates], and I found myself at the gym looking at people differently.”
This feeling of consumption and of objectification is partially created by skeuomorphic animations that cause an ersatz-like dating experience. Swiping left on Tinder is one such suggestive interaction; the user pulls a potential match’s profile (card like representations of real people, stacked in something resembling a card deck) down and to the left, implying the motion of throwing an object away. A person is reduced to something tossed aside, discarded indefinitely. The ultimate outcome is a reductive representation of a three-dimensional human. It is hard to imagine this type of interaction not having some sort of psychological repercussion when combined with habitual behavior repeated hundreds of times.
The emphasis on photos and rapid consumption in swipe apps also raises the stakes for minorities. The data routinely shows that asian men and black women are less desirable on dating apps, likely because their design inadvertently encourages racial biases. These biases are stored in parts of the brain that activate unconsciously. When dating apps prioritize consumption of people at rapid speeds and force a “yes” or “no” decision, they tap into the parts of the brain containing these unconscious biases. Implicit association tests have shown that people take longer to understand positive words paired with pictures of racial minorities than when those positive words were paired with white people. Therefore, shortening the time between decisions does not allow people’s conscious brain enough time to combat their unconscious biases. In person, people may give somebody they have an unconscious bias towards a chance, however when a user is trying to make an artificial yes or no decision on a dating profile, one largely composed of a photo and barren of other information, biases will come into play.
Ultimately, swipe apps are designed this way because they have a fundamental incentive misalignment. Consumption is the top priority, not matchmaking. Consumption drives engagement triggers and creates a sticky app. Conversely, perfect matchmaking would cause the number of users to quickly drop to zero as every person found their perfect match. If users quickly found a match and deleted their account it would cause a catastrophic supply-side network failure where the number of users available for matching was insufficient to attract new users. Compounding this incentive misalignment, the freemium revenue model of most swipe apps is usually designed to adjust the rules of the game in the paying user’s favor. This shows up as features that get premium users more visibility, more matches, and more engagement. Unfortunately, the more difficulty a person has finding a partner, the more likely they will pay into this set of tools designed to improve their success.
Thus, the goals of the ecosystem and the user become irreconcilable; the app wants to keep the user searching, swiping, and hopeful, but ultimately convert them to a premium account when they are unsuccessful in their search. The result is that almost half of all first messages are never reciprocated, and only 19% of mutual conversations lead to a phone number exchange (indicating the two parties possibly met offline). In a survey by Lendedu, a massive 70% of millennial Tinder users claim to have never met somebody from the app at all. Dating apps where the users don’t talk and don’t meet each other might as well be realistic-seeming video games instead of actual connection platforms, no better than their “Hot or Not” predecessor. Being good at matchmaking is bad for business.
was perfect it would cause the number of users on the marketplace to quickly drop to zero as every person found their match. This would cause a catastrophic failure to their network on the supply side. Compounding this incentive misalignment, The result is that almost half of all first messages are never reciprocated, and only 19% of mutual conversations lead to a phone number exchange (indicating the two parties possibly met offline). In a survey by Lendedu, a massive 70% of millennial Tinder users claim to have never met somebody from the app at all. Dating apps where the users don’t talk and don’t meet each other might as well be realistic-seeming video games instead of actual connection platforms, no better than their “Hot or Not” predecessor. Being good at matchmaking is bad for business.
Inevitably, there’s a feeling of “consumption” that comes with the speed and superficiality of swipe apps. One user, who had never used online dating before she tried Bumble for the first time, told me in an interview: “I didn’t like the swiping, [it felt] objectifying, it felt like it rewired my brain. I never used to go out in the world looking for potential [dates], and I found myself at the gym looking at people differently”. This feeling of consumption and of objectification is compounded by skeuomorphic animations that create an ersatz-like dating experience. Tinder’s “swipe-left” animation is one such suggestive animation; the user pulls a potential match’s profile card (card like representations of real people, stacked in something resembling a card deck) down and to the left, strongly suggesting the motion of discarding an object or throwing an object away. A person is represented as a cartoon card in a stack, is reduced to something tossed aside, discarded indefinitely. It is irrelevant whether this was the intended purpose or emotion behind the interaction or not, the ultimate outcome is a reductive representation of a three-dimensional human. It’s hard to imagine this type of interaction not having some sort of psychological repercussion when combined with habitual behavior and repeated hundreds and thousands of times.
Even if an interaction like this is mostly playful and harmless, most people would likely agree that we’ve lost the plot a bit when we’re deciding who to date based solely on a photo. Research does suggest that we can determine a lot from a single photo of a face, unfortunately these rapid judgements of attractiveness are not only extremely quick, they’re based on incomplete data — physical features in photos can be easily manipulated, and attractiveness is often misjudged in a fallible 2–D medium. Plus, research on assortive mating (i.e. people who are similar in attractiveness, resources, etc. tend to up together) suggests that in a 1-to-many setting like online dating, people change their individual assessments of attractiveness. Dating sites, by virtue of being a 1 to many platform, likely have the same effect. Furthermore we know that the longer people know each other the more their perceptions of attractiveness changes and evolves, so it’s unlikely a few seconds on each profile is enough to come to an accurate assessment of a person.
And for minorities the story gets even more disheartening. The data routinely shows that dating app users view Asian men and Black women as less desirable, and it’s likely that this is partially explained by the emphasis on speed and superficiality in dating apps. Our racial biases are stored in parts of our brain that activate unconsciously. When apps prioritize consumption of people at rapid speeds and force a “yes” or “no” decision, they likely tap into the elements of our brain that contain these unconscious biases. Implicit association tests have shown that people take longer to understand positive words paired with pictures of racial minorities than when those positive words were paired with white people. Shortening the time between decisions does not allow our conscious brain to combat our unconscious biases. In person, people may give somebody they have an unconscious bias towards a chance if they meet them in person (this is called TK individualization) and it is incredibly important to combatting racial and other stereotypes. However when a user is trying to make an artificial yes or no decision on a dating profile, one largely composed of a photo and barren of other information, biases will come into play.
And when user do send a message dating apps use subtle design tricks to encourage brevity for the same reasons: shorter messages mean more time for messaging other people, which creates more engagement triggers. Most swipe apps would rather have you send 10 short messages to 10 people than 1 long message to 1 person. The shape of the chat box on Tinder and Bumble is rounded (corners give more space for typing) and their chat boxes are only the height of a single line. Thus their message length usually sits around 50 characters per message. In comparison, other apps that encourage more in depth communication, like OKCupid, have chat boxes that are the height of a few lines with square corners, visibly indicating one is supposed to fill the box with text.
Taking holistically, swipe apps employ multiple tools to make user more efficient at consuming profiles, encourage them make to make quick judgements and engagement, and addict them to the process with variable rewards. It’s in direct conflict with the way sexual and romantic judgements would have been made in the past, a slower process with far less choice, and this rapid transition is jarring and disorienting for their users. Inevitably, there’s a feeling of “consumption” that comes with the speed and superficiality of swipe apps. One user, who had never used online dating before she tried Bumble for the first time, told me in an interview: “I didn’t like the swiping, [it felt] objectifying, It felt like it rewired my brain, I never used to go out in the world looking for potential [dates], and I found myself at the gym looking at people differently”.
The feeling of consumption and of objectification is compounded by skeuomorphic animations that create an ersatz-like dating experience. Tinder’s “swipe-left” animation is one such suggestive animation; the user pulls a potential match’s profile card (card like representations of real people, stacked in something resembling a card deck) down and to the left, strongly suggesting the motion of discarding an object or throwing an object away. A person is represented as a cartoon card in a stack, is reduced to something tossed aside, discarded indefinitely. It is irrelevant whether this was the intended purpose or emotion behind the interaction or not, the ultimate outcome is a reductive representation of a three-dimensional human. It’s hard to imagine this type of interaction not having some sort of psychological repercussion when combined with habitual behavior and repeated hundreds and thousands of times.
Even if an interaction like this is mostly playful and harmless, most people would likely agree that we’ve lost the plot a bit when we’re deciding who to date based solely on a photo. Research does suggest that we can determine a lot from a single photo of a face, unfortunately these rapid judgements of attractiveness are not only extremely quick, they’re based on incomplete data — physical features in photos can be easily manipulated, and attractiveness is often misjudged in a fallible 2–D medium. Plus, research on assortive mating (i.e. people who are similar in attractiveness, resources, etc. tend to up together) suggests that in a 1-to-many setting like online dating, people change their individual assessments of attractiveness. Dating sites, by virtue of being a 1 to many platform, likely have the same effect. Furthermore we know that the longer people know each other the more their perceptions of attractiveness changes and evolves, so it’s unlikely a few seconds on each profile is enough to come to an accurate assessment of a person.
Determining cause/effect of the aforementioned swipe-app designs and features above can be difficult. Dating apps themselves hold troves of data but it’s largely inaccessible as proprietary secrets, and they’re unlikely to share their findings with the greater scientific communities. Independent research is generally population level; partially because dating apps are so new and ever changing, well controlled studies on the effects of dating apps are hard to come by. We can look at tangental markers to suggest that these type of depersonalizing experiences change our behavior.
The data is starting to show concerning population level trends. Do these numbers reflect the success rate of in person interactions, or are they caused by the emphasis on speed and choice in online dating? A staggering 70% of students who have used Tinder have never met somebody from the app, furthermore, 50% of users have never sent or received a message. For an app the proports to be a matchmaking app, helping you find people, message, and ultimately meet people offline, that’s a concerning statistic.
34% of college aged Tinder users admitted they use Tinder primarily for “entertainment” over options like dating and hookups. And a poll by WhatsGoodly, a brand polling company, found that close to half of of millennial’s use dating apps to “procrastinate” instead of actually finding a date. While I’m sure there is more than one positive usage of a swipe app, procrastination never did anyone a favor. It’s likely the emphasis on variable rewards has lead to learned habit forming behavior and promote compulsive usage patterns of these apps in certain vulnerable groups of the population. In fact, millennials, are over twice as likely to admit to feeling addicted to the process of looking for a partner than other generations who have far less exposure to swipe apps.
For instance, TK assortive mating research strongly suggests people date and marry those who are around the same level of attractiveness as themselves. However on dating sites, people reach for those who generally 25 percent more desirable than they are. One could make the argument that a design that relies on a variable reward schedule simply relies on our already existing psychology and gives users what they want at the basest level. And that’s true, to an extent.
Ultimately, swipe apps have a fundamental incentive misalignment. Users want to meet somebody, have sex, fall in love, and maybe even get married. swipe apps, however, want you to use often, and eventually pay them. If matchmaking was perfect it would surely cause the number of users on a platform to quickly drop to zero as every person found their perfect match. This would cause a catastrophic failure to their network effect on the supply side. Furthermore swipe apps generally have a freemium model, meaning they only charge users for a “premium” feature or account, granting them special privileges. These privileges are usually designed to adjust the rules of the game in the paying user’s favor, in order to get them more visibility, more matches, more engagement. The more difficulty a person has finding a partner, the more likely they will pay into this set of tools designed to improve their success. Thus, the goals of the ecosystem and the user become irreconcilable; the app wants to keep the user searching, swiping, and hopeful, but ultimately convert them to a premium account when they are unsuccessful in their search.
It’s been 5 years since the dawn of Tinder and there’s now an entire generation of humans who’s dating experience and understanding is heavily steeped in the culture of swipe-apps. One in four college students have used Tinder, and Tinder or other dating apps’ effect on our culture and psychology will only continue to grow as they become more mainstream.
Unfortunately Tinder and it’s clone apps are designed to change behavior, for the benefit of the company and likely at the expensive of the people they serve. They are designed to de-emphasize messages and communication, and promote habitual behaviors to increase outbound triggers. This likely encourages incomplete communication, ghosting, and flaking. It leads men to see online dating as a numbers game, and many women to be overwhelmed with the experience when they are drowned in attention
It’s not simply that they affect their users on an individual level, there are greater systemic changes to behaviors, norms, and psychology that’s are likely the most damaging to our ability to connect with each-other. Software is proven to shift massive entrenched traditions, upheave culture and communications, overthrow dictators. Take into account the breakneck speed of app development and software cycles and we’ve got an overflowing pot of roiling cultural upheaval. The next five years will determine if this massive experiment in online dating was a catastrophic failure, or the first step towards a software solution to the worlds oldest analogue problem; love.