The Upside of Irrationality: The Unexpected Benefits of Defying Logic at Work and at Home

by Dan Ariely

Consider what makes supermarkets super. They save you the hassle of having to walk or drive to the baker, butcher, vegetable stand, pet store, and drugstore; you can efficiently buy all the things you need for the week in one convenient place.

In addition to markets for food, housing, jobs, and miscellaneous items (also known as eBay), there are also financial markets. A bank, for example, is a central place that facilitates finding, lending, and borrowing. Other market players, such as real estate brokers, for example, try in a yenta-like way to understand the needs of sellers and buyers and match them properly.

Once I learned what the real process of online dating involves, my enthusiasm for this potentially valuable market turned into disappointment. As much as the singles’ market needed mending, it seemed to me that the way online dating markets approached the problem did not promise a good solution to the singles problem. How could all the multiple-choice questions, checklists, and criteria accurately represent their human subjects? After all, we are more than the sum of our parts (with a few exceptions, of course). We are more than height, weight, religion, and income. Others judge us on the basis of general subjective and aesthetic attributes, such as our manner of speaking and our sense of humor. We are also a scent, a sparkle of the eye, a sweep of the hand, the sound of a laugh, and the knit of a brow—ineffable qualities that can’t easily be captured in a database.

But could it be that, in their desire to make the system compatible with what computers can do well, online dating sites force our often nebulous conception of an ideal partner to conform to a set of simple parameters—and in the process make the whole system less useful?

Mike Norton, and I set up our first online dating study. We placed a banner ad on an online dating site that said “Click Here to Participate in an MIT Study on Dating.” We soon had lots of participants telling us about their dating experiences.

We found that people spent an average of 5.2 hours per week searching profiles and 6.7 hours per week e-mailing potential partners, for a total of nearly 12 hours a week in the screening stage alone. What was the payoff for all this activity, you ask? Our survey participants spent a mere 1.8 hours a week actually meeting any prospective partners in the real world, and most of this led to nothing more than a single, semifrustrating meeting for coffee.

Talk about market failures. A ratio worse than 6:1 speaks for itself. Imagine driving six hours in order to spend one hour at the beach with a friend (or, even worse, with someone you don’t really know and are not sure you will like). Given these odds, it seems hard to explain why anyone in their right mind would intentionally spend time on online dating.

Of course, you might argue that the online portion of dating is in itself enjoyable—perhaps like window-shopping—so we decided to ask about that, too. We asked online daters to compare their experiences online dating, offline dating, and forgetting about the first two and watching a movie at home instead. Participants rated offline dating as more exciting than online dating. And guess where they ranked the movie? You guessed it—they were so disenchanted with the online dating experience that they said they’d rather curl up on the couch watching, say, You’ve Got Mail. So it appeared from our initial look that so-called online dating is not as fun as one might guess. In fact, on...

the results of our studies suggested that using searchable attributes for online dating is

unnatural, even for people who have lots of practice with this type of activity.

Online daters aren’t particularly excited about the activity; they find the search process difficult, time-consuming, unintuitive, and only slightly informative. Finally, they have little, if any, fun “dating” online. In the end, they expend an awful lot of effort working with a t...

I once met a student at MIT who adopted an extraordinary method for sorting potential dates into categories. Scott’s objective was to find the perfect woman, and he used a very complex, time-consuming system to accomplish his goal. Every day, he went online to search for at least ten women who met his criteria: among other attributes, he wanted someone who had a college degree, demonstrated athleticism, and was fluent in a language other than English. Once he found qualified candidates, he sent them one of three form letters containing a set of questions about what kind of music they liked, where they had gone to school, what their favorite books were, and so on. If they answered the questions to his satisfaction, he would advance them to the second step of a four-stage filtering process.

In stage two, Scott sent another form letter containing more questions.

he logged the results in a spreadsheet that listed each woman’s name, the stage of the relationship, and her cumulative score, which was based on her answers to the different questions and her overall potential as his romantic partner. The more women he logged into his spreadsheet, he thought, the better his prospects for finding the woman of his dreams.

After a few years of searching, Scott had coffee with Angela. After meeting her, he was sure that Angela was ideal in every way. She fulfilled his criteria, and, even more important, she seemed to like him. Scott was elated.

Two weeks later, I learned that Scott’s fastidiously chosen beloved had turned down his marriage proposal.

we decided that we would try to bring some elements from real-world dating into online dating.

we set up a simple virtual dating site using “Chat Circles,” a virtual environment created by Fernanda Viégas and Judith Donath at the MIT Media Lab. After logging on to this site, participants picked out a shape (a square, triangle, circle, etc.) and a color (red, green, yellow, blue, purple, etc.). Entering the virtual space as, say, a red circle, the participant would move a mouse to explore objects within the space. The objects included images of people, items such as shoes, movie clips, and some abstract art. Participants could also see other shapes that represented other daters. When two shapes moved close to each other, they could start an instant-message conversation. Obviously, this environment could not represent the full range of interactions one could experience on a real date, but we wanted to see how our version of virtual dating worked.

As we expected, the resulting discussions resembled, rather closely, what happens in regular dating.

we asked a group of eager daters to engage in one regular online date with another person (a process that entailed reading about another person’s typical vital statistics, answering questions about relationship goals, writing an open-ended personal essay, and writing to the other person). We also asked them to participate in one virtual date with a different person (which required the daters to explore the space together, look at different images, and text-chat with each other). After each of our participants met one person using a standard online dating process and another

person using the virtual dating experience, we were ready for the showdown.

To set the stage for the competition between these two approaches, we organized a speed-dating event like the one described in chapter 7, “Hot or Not?” In our experimental speed-dating event, participants had an opportunity to meet face-to-face with a number of people, including the person they’d met in the virtual world and the person they’d met in our standard online dating scenario. Our speed-dating event differed slightly from the standard experience in another way, too. After each four-minute interaction at the tables, participants answered the following questions about the person they had just met: How much do you like this person? How similar do you think you are to this person? How exciting do you find this person? How comfortable do you feel with this person? Our participants scored each question o...

TO RECAP, THE experiment had three parts. First, each of the participants went on one regular online date and one virtual date. Next, they went speed-dating with multiple people, including the person they met online and the person with whom they’d gone on the virtual date. (We didn’t point out people they’d met before, and we left it for them to recognize—or not—their past encounters.) Finally, at the end of each speed date, they told us what they thought about th...

initial experience—either virtual or regular online dating—would make a real-life date more likely. We found that both men and women liked their speed-dating partner more if they’d first met during the virtual date. In fact, they were about twice as likely to be interested...

WHY WAS THE virtual dating approach so much more successful? I suspect the answer is that the basic structure used in our virtual dating world was much more compatible with another, much older structure: the human brain. In our virtual world, people made the same types of judgments about experiences and people that we are used to making in our daily lives. Because these judgments were more compatible with the w...

You may sense that you like Janet more than Julia and that you want to spend more time with her, but it’s not easy to say why or to isolate the few variables that make you prefer her. Is it her body shape? The way she smiles? Is it her sense of humor? You can’t put your finger on what it is about Janet, but you have a strong gut feeling about it.*

for our next event, each person bring a personally important object (for example, a souvenir or a photograph) to use as a discussion starter. This time we could not stop people from talking. Their discussions were deeper and more interesting. The events resulted in many friendships. In this case, too, the presence of an external object helped catalyze the discussions and improve the outcome.

In the same way that the chemical composition of broccoli or pecan pie is not going to help us better understand what the real thing tastes like, breaking people up into their individual attributes is not very helpful in figuring out what it might be like to spend time or live with them.

In the end, our research findings suggest that the online market for single people should be structured with an understanding of what people can and can’t naturally do. It should use technology in ways that are congruent with what we are naturally good at and help us with the tasks that don’t fit with our innate abilities.

But what if—as behavioral economics has shown in general and as we have shown for dating in particular—we are limited in the way we use and understand information?

Building an online dating site for perfectly rational beings can be a fun intellectual exercise. But if the designers of such a Web site really want to create something that is useful for normal—albeit somewhat limited—people who are looking for a mate, they should first try to understand human limitations and use them as a starting point for their design.

But in the end, I do think that there are a few personal lessons to be learned from this research.

First, given the relative success of our virtual dating experience, Sarah should try to make her online dating interactions a bit more like regular dating. She can try to engage her romantic prospects in conversations about things she likes to see and do. Second, she might go a step farther and create her own version of virtual dating by pointing the person she is chatting with to an interesting Web site and, much as in real dating, experience something together.

What matters most is that she make an effort to do things she enjoys with other single people and this way learn more about her compatibility with them.

Fundamentally, the online dating market is a failure of product design.

Just as online dating sites that try to reduce humans to a set of descriptive words too often fail to make real matches, companies disappoint when they don’t translate what they’re offering into something compatible with the way we think.

though we may possess incredible sensitivity to the suffering of one individual, we are generally (and disturbingly) apathetic to the suffering of many.

gave participants $5 for completing some questionnaires. Once the participants had the money in hand, they were given information about a problem related to food shortage and asked how much of their $5 they wanted to donate to fight this crisis.

Food shortages in Malawi are affecting more than 3 million children. In Zambia, severe rainfall deficits have resulted in a 42% drop in the maize production from 2000. As a result, an estimated 3 million Zambians face hunger. 4 million Angolans—one third of the population—have been forced to flee their homes. More than 11 million people in Ethiopia need immediate food assistance.

The second group of participants,

was presented with information about Rokia, a desperately poor seven-year-old girl from Mali who faced starvation. These participants looked at her picture and read the following statement (which sounds as if it came straight from a direct-mail appeal):

Her life would be changed for the better as a result of your financial gift. With your support, and the support of other caring sponsors, Save the Children will work with Rokia’s family and other members of the community to help feed her, provide her wit...

Again, ask yourself how much you might donate in response to the story of Rokia.

If you were anything like the participants in the experiment, you would have given twice as much to Rokia as you would to fight hunger in general

we are willing to spend money, time, and effort to help identifiable victims yet fail to act when confronted with statistical victims (say, hundreds of thousands of Rwandans). But what are the reasons for this pattern of behavior?

First, there’s your proximity to the victim—a factor psychologists refer to as closeness. Closeness doesn’t just refer to physical nearness, however; it also refers to a feeling of kinship—you are close to your relatives, your social group, and to people with whom you share similarities.

The second factor is what we call vividness. If I tell you that I’ve cut myself, you don’t get the full picture and you don’t feel much of my pain. But if I describe the cut in detail with tears in my voice and tell you how deep the wound is, how much the torn skin hurts me, and how much blood I’m losing, you get a more vivid picture and will empathize with me much more.

The third factor is what psychologists call the drop-in-the-bucket effect, and it has to do with your faith in your ability to single-handedly and completely help the victims of a tragedy.

that intense level of personal involvement will save only a few people, leaving millions of others still in desperate need.

The similarity of the results when participants were primed with emotions and when they were not primed at all suggests that even without emotional priming, participants relied on their feelings of compassion when making their donation decisions (that is why adding an emotional prime did not change anything—it was already part of the decision process).