with Spencer Greenberg
the podcast about ideas that matter

Episode 034: Search Data and Self-Improvement (with Seth Stephens-Davidowitz)

Enjoying the episode? Want to listen later? Subscribe on any of these apps or stores to be notified when we release new episodes:

April 1, 2021

What can we learn about people from search data? What does search data reveal about human nature that surveys and polls fail to reveal? What patterns of searching do people exhibit when they're suffering from mental and physical illnesses? How can we use data to make better decisions, become better parents, date better, and be happier?

Seth Stephens-Davidowitz is a data scientist, author, consultant, and keynote speaker. His book Everybody Lies was a New York Times bestseller and an Economist and PBS NewsHour Book of the Year. He has worked as a contributing op-ed writer for the New York Times, a data scientist at Google, and a visiting lecturer at the Wharton School at the University of Pennsylvania. He received a BA, Phi Beta Kappa, from Stanford in philosophy and a PhD in economics from Harvard. Find more about him at, follow him on Twitter at @seths_d, or email him at

JOSH: Hello, and welcome to Clearer Thinking with Spencer Greenberg, the podcast about ideas that matter. I'm Josh Castle, the producer of the podcast, and I'm so glad you've joined us today. In this episode, Spencer speaks with Seth Stephens-Davidowitz about the power of search data to indicate health trends and social behavior, public perception of disease, the impact of environment on child development, and the relationship between attractiveness and happiness.

SPENCER: Seth, welcome! I'm really glad to have you here.

SETH: Glad to be here, Spencer.

SPENCER: The first topic I want to talk to you about is what we can learn about people from search data.

SETH: I wrote this book three years ago, "Everybody Lies," and I'd written a whole bunch of New York Times pieces and a couple academic papers based on the idea that people are really honest in their searches. They tell things — particularly to Google (I mostly study Google searches) — that they don't tell other people. A big problem with the main way to understand people — surveys — is that sometimes people lie to surveys. So there's something called social desirability bias. People don't say what they're really thinking or really going to do, or the reasons they do things. They don't have an incentive to tell the truth. But on Google searches, they tend to be really honest. They'll tell you their sexual desires or their racist thoughts or their mental health problems or all kinds of sensitive topics and struggles that people don't usually talk about in polite company. And it can give you, in my mind, a much more accurate view of the human psyche, an unprecedented window into the human psyche. That's the thesis of most of my adult life career.

SPENCER: It's a cool way of looking at human nature. And I think, especially when it comes to interviews that were done by (let's say) a psychologist who's talking to a person face to face, it seems it's going to be so full of the common social biases, like not wanting to look bad or not wanting to feel embarrassed. I've even heard of cases where this was used unscrupulously, where (let's say) you're testing a drug and you don't want it to look bad, instead of asking the patient, "Did you suffer from this symptom?" — let's say some kind of sexual problem — only if the patient actually brings it up do they count it as a symptom. How many people are just gonna bring up the sexual problem with their doctor? Probably most people don't do that. And so, you can craft the data to look the way you want. To what extent do you think this is a big problem with anonymous surveys though? Do you think people are lying on those?

SETH: I think there's some lying. There's also an issue that people lie to themselves. If you ask people before an election, "Are you going to vote?" Many more people say they're going to vote than actually go out to vote in the election. And they may not deliberately be lying in that situation. They may think they're going to vote, but then they don't get around to it. And actually, I've shown that you can predict, with much higher accuracy, where voting rates are gonna be high, by Google searches: how to vote, where to vote, polling places. That's much more predictive than asking people. Even though it's an anonymous survey, people think they're gonna vote. To some degree, Google searches know them better than they know themselves. And I think there are times where people do lie deliberately even on an anonymous survey, just because they don't feel any need to tell the truth. If a survey asks you, "Are you racist?" or "Did you not like Barack Obama because he's Black?" I think some people may, in their heart of hearts, know that they had an issue with Obama's race. But they don't feel comfortable telling Gallup or Pew or Quinnipiac that they had that attitude even though the survey is anonymous. Why admit that? There's no reason to admit that. So I think that can lead to full-on deception even in anonymous surveys.

SPENCER: It seems like there's two advantages that you're pointing at for stuff like search data. One is that there's a real incentive to be honest because you're actually trying to get an answer to a question. That's the reason you're doing the search. The second is that you're actually tracking real behavior. Someone might say, "Yes, I'm going to vote." But an even better indication is that they are actually searching for where the polling center is. So it's a behavior rather than just a reported belief. Is that how you think about it?

SETH: Exactly. And also you keep the data over time. People change over time. One of my first studies was analyzing racism in the United States. I looked at searches for N-word jokes and that includes data from some people who may have made those searches six, seven months ago. So they might be correct in saying, "I don't make those searches now," but they still probably have those same attitudes. So keeping that behavior over time, people forget who they were, what they did. That's another advantage of the search data.

SPENCER: Do you see there being downsides to using the search data? For example, is there a challenge around like, well, not everyone who wants to do a thing or believes a thing is gonna bother searching, and so maybe the people who search are not always representative of the people who have that belief.

SETH: There are definitely disadvantages to search data. Just about everybody I tell the idea to or read my book asks me, "I made a search for something horrible, and I don't have that attitude. I was just curious or I was working on a research paper." Certainly, when I was writing my book, I was doing a section on racism so I searched for N-word jokes because I needed to know what came up.

SPENCER: Some social scientist is now studying you and being like, "This is the most racist person we've ever encountered. We've never seen someone search so many racist search terms."

SETH: I always say that people are like, "Would you be embarrassed if someone released your Google searches?" And I think I'd be the only one who would be okay because I could legitimately claim every embarrassing search was for research purposes. So 80 or 90% of the embarrassing searches really were for research purposes. And the 10 to 20% that weren't, I could just claim they were.

SPENCER: Excellent smokescreen to hide behind! There actually was, apparently, a plug-in at one point. They would do hundreds and hundreds of random Google searches with the idea that it would mask any real searches you were doing. So if a company was monitoring your search traffic, they would never know what you're really searching. But then of course the problem with that is that this random thing that's just searching random terms might search like 'donkey sex' or something, then suddenly you're like, "No, no! That wasn't me searching, really!"

SETH: That's interesting. Another issue is that people sometimes don't search exactly. They search for general things that don't tell you exactly how they feel about a topic. Everyone's saying, "Can you use searches to predict who someone's going to vote for?" And the first initial approach is using Google searches. Let's see how frequently each candidate is searched. Is Trump searched more than Biden? And a big problem with this is that, searching for Trump or Biden doesn't tell you how you feel about that candidate. You might search for Trump because you love Trump, or you might search for Trump because you hate Trump. And same with Biden. I actually found that there's this subtle indicator that's pretty fun, although I don't think it's going to work at this election for various reasons. But you could actually predict, with decent accuracy, which way someone's gonna vote based on the order in which they put the candidate. For example, a lot of people search, 'Trump-Biden polls' or 'Biden-Trump polls.' It's very clear that the people who go Biden are much more likely to put Biden first. People who support Biden are much more likely to say 'Biden-Trump polls' or 'Biden-Trump debate.' People who go Trump are much more likely to say, 'Trump-Biden debate' or 'Trump-Biden polls.' It's interesting, the ways in which we give away how we feel about a topic without necessarily being aware that we're giving away how we feel about that topic.

SPENCER: That's super interesting. I also want to know, how do you respond when people say, "I searched for this bad thing but it doesn't mean I have that bad attitude. I was just curious about it." What's your thought on that?

SETH: In general, I'm looking at anonymous aggregate geographic data, so I'm not necessarily saying that one particular individual is racist. When I did my first study of measuring racism in the United States, I found that the N-word was searched the most in West Virginia, Mississippi, Louisiana, Western Pennsylvania, Eastern Ohio, and upstate New York, and lowest in Hawaii, parts of California, and parts of Idaho. And it is true that in that data, probably that included people... Certainly, I was in that data. I was searching N-word from Brooklyn to do my research. There were probably some African-Americans searching the actual N-word for various reasons related to experiences they'd had. But that gets swamped in the data by all the people who are searching for N-word jokes because they really find N-word jokes funny and want to read them. So it tends to be a smaller factor and probably makes it difficult to draw individual level conclusions just based on at least one search. But usually, at the area level or time series level, those curious searches get swamped in the data by people with a more direct reason for making the search.

SPENCER: Right. I guess the way I think about it, there are many different tools for studying humans. No tool is going to be perfect. But I really like the tool that you rely on because I think it has an interesting set of properties for answering, especially questions that are very hard to ask directly. Maybe it's not perfect; it's gonna have flaws. But every single one of these tools has flaws. And so it's just a question of, "What's the best tool to answer what you're trying to answer?" And I really appreciate the way that you pursue answers to these kinds of questions. Do you want to tell us about some of the cool questions that you were able to study on that?

SETH: The racism one is how I started. I basically found that parts of the country that had the highest racism support Obama a lot less than the previous Democratic candidate, John Kerry. It's a very, very strong relationship and very robust to all kinds of controls. I think it was overwhelming in the data that Obama lost (I say) about 4% of the vote just from the continued racism in the United States which, when I published the paper, was considered pretty shocking. Actually, some people initially didn't publish the paper because they just couldn't believe there was this level of racism in the United States, and they're like, "There's no way millions of people are searching the N-word in the United States every year." Remember after Obama was elected, there's this idea that we lived in a post-racial society, that racism was a thing of the past. But I think it was abundantly clear in the search data that, despite what people might be telling pollsters, explicit racial animus was still a major factor in the United States, was really activated by Obama. And then that paper just languished for a while. But then when Trump was running for the Republican primary in 2016, Nate Cohn at the New York Times, he's like, "Trump's saying all these racist things. He started his political career by questioning whether Barack Obama, the first African-American president, was born in the country. He didn't repudiate support from members of the KKK. He'd retweet false statistics about how frequently African-Americans committed crime. And every time he did all these things, people said, 'Trump's just committed this huge gaffe. His support is about to fall apart.' But he does better and better. Does your search data point to a reason for this? Is Trump's support stronger among places with higher racism?" And so I sent him the data, and Nate Silver confirmed it later, that the single highest predictor of anything they tested in that model for support for Trump in the 2016 Republican primary, was that Google Search indicator I had come up with a few years earlier: the searches for the N-word. Trump's map of support matched Google searches for racism almost perfectly. It's pretty clear that racism at that time was hidden. Real racial animus that was really activated by Obama's election was driving Trump's early support. That was one paper. I've done studies on child abuse. (I only choose cheery topics.) During the Great Recession, there was a drop in reported cases of child abuse, which is surprising because you would think when economic conditions get worse, that child abuse would go up. Many of the risk factors of child abuse are (we think): unemployment, anxiety, anger, alcoholism. All those went up during the Great Recession, but it turns out that child abuse actually went down. And I actually look at data a bunch of different ways. There's explicit searches, and they're really sad searches — like 'my dad hit me' or 'my mom beat me' — by older children, and they went up, straight up, during the recession. I looked at the data a bunch of different ways. I think it's very clear that during the recession, child abuse actually went up, although reporting of child abuse cases went down because basically, Child Protective Services were overworked. They couldn't take as many cases and the people who are most likely to report cases — teachers, firefighters, police officers — were losing their jobs or unemployed, so they were less likely to report cases. Basically, child abuse went up; reporting of it went down.

SPENCER: That's a great example of how this kind of approach can supplement other traditional approaches because you're able to say, "We want to be able to separate what's being reported from what's actually happening."

SETH: Exactly. I've been looking a little bit at this; other people have been looking a little bit at this. It seems pretty clear that the same thing happened during the early days of COVID. Initially, when schools were closed and people were in lockdown, there was a drop in reported cases of child abuse and domestic violence, but probably a rise in actual cases, in part detected through Google searches. That is a good use case, I would say, particularly for those direct searches where you know what it means when someone searches (again these sad searches) 'my dad hit me' or 'my husband beats me.' It's a pretty direct statement of what's going on.

SPENCER: Going back to the racism topic for a moment, you were saying that about 4% of the votes you could attribute to racism. Is that right?

SETH: Yeah.

SPENCER: It's interesting, because back then, people were like, "Four percent?! There's no way that many people are racist," was the reaction. But I feel like, today, the reaction might be the opposite, where people might say, "Only 4%? People are so racist, I would expect it to be a lot higher." The question I have for you: do you feel like you have any way of estimating how many people are racist according to your definition?

SETH: That's tough. I don't even know what my definition is. I think racism is such a complicated topic. My friend was recently complaining that we have one word for racism. Among your thousands of fascinating Facebook or Twitter posts, I think you may have had a post on the different types of racism.

SPENCER: Yeah, I did.

SETH: I totally agree with you there. There are so many different types of racism and it's hard to know. There is searching for N-word jokes. There is secretly judging African-Americans. There is implicit bias towards African-Americans. There is having violent thoughts towards African-Americans. There are all kinds of different ways that racism can play out. But I think it's hard using any methodology to even settle on a definition, let alone to say 1% are that way. And it's also probably going to be a lower bound for various reasons. I said that Obama lost about 4% to racism. That actually means, if you do the math, that about 10% of White voters who (I argue) would have voted for a White Democrat, did not vote for Barack Obama. Because the 4% total lost votes — basically, there's a certain part of the country that was going to vote for the Republican whether he was Black or White — the 4% comes only among the people who would have voted for Obama to begin with and that also includes some African-Americans. So it is (I think) a high percentage of White people, who (I argue) would have voted for a Democrat, did not vote for Obama just because he was Black, which I think, even in this day and age, that still strikes me as maybe surprisingly high.

SPENCER: Right. Certainly, I think there's been shifting definitions of racism. We've seen different books come out that will propose, this is how to think about racism, or that's how to think about racism. And that makes it all the more complicated because we're actually in a situation where there's debates happening about, when you say 'racism,' what do you actually mean by that? 'I mean this and everyone should adopt my definition.' Now it's a very confusing question to ask, 'how many people are racist?' And I think the answers range from everyone to a few percentage of people depending on exactly where you draw that line.

SETH: After I'd done that research, I was thinking (and some other sociologists have said this) that there probably hasn't been enough emphasis on explicit racial animus. Maybe there is now more since Trump has been president and racism has come out of the closet to some degree. But there was this idea that the focus of academic research for the beginning of the 21st century, I would say, was largely on implicit bias and subconscious bias. And that's the idea that everybody's racist. So you might not be searching for N-word jokes and you might not think consciously that African-Americans should get worse treatment or that an African-American shouldn't be president, but that subconsciously you make these associations between African-Americans and criminal behavior, laziness or lack of intelligence. And I think that's definitely an issue. There still is a lot of explicit racial animus in the United States and people literally search for N-word jokes and you look at when these searches are rising. One of the periods they rose a lot was during Hurricane Katrina. I don't know if you remember back then, but there were all these videos of poor African-Americans borderline drowning in New Orleans.

SPENCER: Oh, my gosh.

SETH: At the same time, for whatever reason, this activated racists to make searches like N-word jokes. I guess they were getting off on humiliating Black people when they were down. So that still is a big factor in American life and has big impacts on things like who people vote for president. Probably also on how police treat people and many other areas of life and it probably hasn't gotten the right amount of attention, in part because it's so hard to measure in polls. Polls told us incorrectly that that explicit animus had gone away when it was really always there.

SPENCER: It's interesting because, if you look at some of these long-standing polls that will ask about things that just seem very explicitly racist like, if someone of a different race moved next door, would you be upset, or these kinds of things. The numbers just went down and down over the years to the point where some polling places stopped running them because you're getting such low numbers of people agreeing. And yet people are still searching for these horrible N-word jokes when people are dying in tragedies. I think there is a cost to some degree when we make words change so that they apply to a broader and broader group. On the plus side, that can make us more aware of the biases we all have. But on the minus side, it can obscure the fact that there are some people that actually have really atrociously bad views and lump them in with people who are much less racist. It actually may be a bit of a smokescreen that they can hide behind. What do you think about that?

SETH: Although, to be fair, racist searches have also gone down, at least since Google Trends has measured it since 2004. So even on this kind of explicit racism, there seems to be continued progress. I'm gonna name drop Steven Pinker; we met and had a bunch of discussions and I think we have similar attitudes towards social science and various areas. But he's like a total optimist and I'm a total pessimist. He wrote this book, "Enlightenment Now," about how the world's getting better. And he wrote a book, "The Better Angels of Our Nature," about how violence has declined over time. So of course when I saw the Google searches, my bias has always defined probably the bad things happening, which considerably were fair, the racism that still exists in the United States that hurt Obama and helped Trump. I think Steven Pinker, interestingly, looked at the data and was like, "Wow! This has really gone down a lot over time," which is also true and also has to be pointed out. Actually it has continued to go down during Trump's presidency, partly because (I think) some racists are dying. Google searches aren't broken down by demographics but you can compare the area level search data to area level demographics and you definitely see that one of the biggest predictors is age that places a lot of older people making more searches for things like N-word jokes, so places with a lot of people 65 and older, 80 and older. And sometimes people literally just die over time. So racism, to some degree, is dying out in the United States. It's not dying out as fast as many of us would like and it hasn't gone down to the levels that polls might suggest it has gone down to, but it is continually going down in the United States.

SPENCER: That's a really good point. It's amazing watching movies even just from the 80s and seeing elements of sexism or racism that today, we'd be shocked that that was in a film. And then if you go back 60 years ago or 80 years ago, it's unbelievable by our standard today, how racist people were. I think that it is good to remember that at least there has been a ton of progress even if we're far from being all the way towards a completely equal and just society.

SETH: For sure.

SPENCER: Alright, so let's talk about a different topic, which is that of mental health. What have you done around that?

SETH: I've actually been working a while on this and I still haven't published it, but I think it's a super interesting topic: mental health. Actually, other scholars have correlated searches for suicide with suicide rates. When a lot of people are making searches — again sad searches: 'how to commit suicide,' 'suicidal,' 'suicide help' — suicide rates in the area tend to go up and the searches are more predictive than surveys asking people if they had suicidal ideation. I think that work is incredibly important and really can potentially be used to help save lives in interesting ways. I got really interested in what people search for before they search for 'suicide.' I thought that's a really interesting topic and we could learn a lot about the mind of the suicidal. There's actually an old dataset from 2006 that AOL released to researchers which has the same individuals over time anonymous. So you can see someone's three-month history of searches. And some of these people do make really sad searches suggesting that they basically want to take their lives. We don't know if they actually took their lives; that could be another project. And some of the searchers are just so sad, the unprecedented window you have into someone's mind from their search behavior. You see senior citizens searching 'can't pay rent', 'how am I going to get housing?' and 'not enough from Social Security,' all these searches suggesting a senior citizen in deep financial trouble. And then 'how to commit suicide,' 'I want to end my life,' such a sad series of searches.

SPENCER: Oh, my God!

SETH: Then you see some search series that are just from one person in that dataset, again anonymous. It's just going back and forth, almost all day, between searches for help with severe back pain and a desire to end their life. It's so sad. It also makes you (I think) more compassionate because we never know what's going on in someone else's life. Someone may seem like they have everything together. They have a great career; they have a great spouse. But that person may have some health challenge like severe back pain that totally dominates all their waking thoughts and makes them actually suicidal. And there was one search that got me down. I think it is an interesting trail where this young woman basically searched that she just got diagnosed with herpes, the sexually-transmitted disease, and then she searched that she wants to commit suicide. There's basically a couple-day window after she had gotten her herpes diagnosis that she was suicidal. Again, we don't know what actually happened to her. We hope that she did not go through with this suicidal desire. But I actually looked in the data and herpes (and some other STDs, but mainly herpes) was a surprisingly common thing to search before someone searched for suicide.

SPENCER: It's so surprising, given the incredibly high prevalence of herpes.

SETH: Yeah, high prevalence herpes and that it's not a severe disease that causes great continual discomfort. Something like back pain, you see a search series like that and you understand that person's in agonizing pain day in, day out. And we hope that we could get a solution to the problem and that you could see the right doctor where they could find a way to learn to live with that. But we have some understanding for why that severe condition is causing that man to search for ways to end his life. With something like herpes, as you said, it's very common and the long-term prognosis is good. The symptoms are not severe. It's not life- threatening. It shouldn't have a big impact on someone's life. I actually looked at this also on Google Trends. I said this is weird. It's surprising that herpes was causing some fraction of people to have suicidal ideation. I think the reason for that is the stigma, particularly for young people. Young people are so confused with how the world works and they greet many challenges with something like paranoia and feeling like there's something deeply flawed and deeply wrong with them and deeply unlovable about them. That stigma can cause some cracks for young people to think of ending their lives. Anyway, I went to Google Trends where you can actually see what people make in the same search window that they make other searches. So in a search window, where someone searches herpes and thoughts of suicide, what else do they search in that search window? And I found the number one other search that people searched in that search window was celebrities with herpes, which actually is a common search for many illnesses. It makes sense when you have a health condition... I've told people that I've struggled a lot with depression as an adult, and I've probably searched 20 times for celebrities with depression. Woody Allen suffered from depression, or Leonard Cohen suffered from depression; you could have guessed that from their art. But whoever you learn about — Brad Pitt had a depressive episode — whoever you learn has depression, it does make you feel better about your condition. It makes you feel less alone and makes you feel less stigmatized. So it's a common search for many illnesses, looking for people who could be role models who also suffer from that condition. I looked through what happens when you search celebrities with herpes, now knowing that some fraction of teenagers are getting diagnosed with herpes, searching for suicide and then searching for celebrities with herpes to make themselves feel better. And I found that, for every other disease I tested — for depression, 'celebrities with depression,' lots of celebrities have come out as having depression, 'celebrities with migraines,' lots of celebrities have come out with having migraines, 'celebrities with cancer,' lots of celebrities have come out with having dealt with cancer — just about any illness, there are many celebrities who have openly shared that they had this illness. But 'celebrities with herpes,' there's just about nothing, certainly not A-list stars, saying that they had herpes despite, as you said, herpes being very common.

SPENCER: I looked at the prevalence. I think HSV-1 is about 48% of people who are under the age of 50. It's an extremely high prevalence.

SETH: If you actually look at what comes up with 'celebrities with herpes,' at least when I checked...

SPENCER: It seems like we can reasonably assume that something close to half of all celebrities have herpes. [laughs]

SETH: Yeah. When you go 'celebrity with herpes,' I just clicked on the first thing and it says you have famous stars with herpes and STDs but in all of them, it's not the celebrity saying they have herpes. It's that they were accused of having herpes. Anne Heche is the first one on the site I went to and she was rumored to be a patient of herpes. Derek Jeter. It's one of those controversial topics nowadays that Derek Jeter may have herpes, that he's rumored to be a big source of herpes transmission to his celeb girlfriends but he certainly never said he's had herpes.

SPENCER: So it just stigmatizes it further, basically, making it seem like some horrible accusation.

SETH: Exactly. Paris Hilton's herpes is the biggest rumor in the list of stars living with herpes. The rumor became public due to prescriptions she had that contained a medicine to recover from genital herpes. But it's not like Paris Hilton, after that rumor came out, said "Yeah, I have herpes. It's something that 48% of other people have and it's nothing to be embarrassed about." So I would think that a teenager making that Google search, 'celebrities with herpes,' would not feel better necessarily about what she'd found. But she could feel a lot better if some celebrities came out and said they had herpes and tried to fight the stigma. I think, just from the data analysis that I talked about, it literally would save lives if some celebrities came out and said that they had herpes and there was nothing to be ashamed of, nothing to be stigmatized. I think that's the power of the dataset, to uncover thoughts like the teenagers becoming suicidal over a diagnosis of herpes, and looking and failing to find celebrity heroes that may make them feel better about their situation.

SPENCER: That's such an interesting example. Herpes seems to me like a case where society has gone insane, where it's something that about half of people have and yes, it feels extremely stigmatizing and nobody talks about it. So it's like what the heck is going on? There's also this interesting euphemism, which is having cold sores. People say, "Oh yeah, I just have a cold sore," or whatever. People almost don't even put it together that it is often the same thing. And yet, one of them seems to be stigmatized and the other is like, "Oh, just a cold sore, no big deal." So there's a state of collective madness we have around that.

SETH: I was talking about this with a friend recently. In general, how much unhappiness in life is caused by having problems that are actually really common, that you think are not common or you think other people aren't dealing with, in part, because other people are usually putting on a front in their public presentation. In social media and in many other areas, people show their best selves so we see the best selves of other people and we see our complete self. We have a distorted view of how life is supposed to go. It's very easy to feel like we're freaks or we're messed up or we're unusually unpopular. I think people would be happier if they had an honest view of other people's lives, not a distorted view. I actually became happier after analyzing all this Google search data because I think a lot of my unhappiness is also due to feeling messed up in some horrible way or thinking other people have it so easy and I have it so bad. When you look through search data and you see that, okay, life is a struggle for most people some of the time, it can relieve some of the pressure that we can feel.

SPENCER: It reminds me of two quotes, both of which I like a lot. The first is, "Don't compare your insides to other people's outsides.'' Other people are constantly trying to project something to the world. And your own life, you experience from the inside without all the marketing on top of it. The second one is, "Be kind because everyone is fighting a great battle." And I think that's really true that almost everyone has some problem in their life, and unless you know them really well, you probably don't know what it is. It could be minor, like they have a little bit of trouble sleeping, or it could be major, like they're considering killing themselves or a family member of theirs is dying or...

SETH: [interrupts] They have horrible back pain, constant horrible, horrific back pain or something.

SPENCER: Exactly. I think these are both good reminders. And a lot of what we experience that is bad, is very common, yet people don't talk about it. We'd be a lot better off if people were more comfortable publicly sharing these things that are stigmatized but are really common so that others realize, "Hey, you know what, almost everyone has something like this, if not exactly the same thing. So there isn't so much reason to be embarrassed."

SETH: That is true. Although to be honest, I thought that I was gonna look through Google search data and find out that I was actually really normal. The things I thought were weird about me aren't that weird. And I would say the search data just made me feel a little weirder.

SPENCER: I would never describe you as normal, Seth.

SETH: Yeah. You didn't need to see search data. [laughs] I agree. I was trying to make that point. You agree that there are things that are really common, like herpes, or another one I talk about in the book is homosexuality. I estimate maybe 5% of men are gay and I think we both agree that definitely someone should not feel bad that they were attracted to members of the same sex. But I don't think the reason for that is because 5% of men are gay. If 1% of men are gay, or one in a thousand men are gay, it still (I think) shouldn't be wrong to be gay. And if someone had some health condition that wasn't shared by 48% of the population, if it was only shared by 1% or less of the population, still, we would hope, it wouldn't be a reason to kill themselves. So I think it is important to keep in mind that other people tend to share many of the struggles we share but it's also the particular struggles. Some of the struggles we have maybe are rare and that still doesn't make them bad.

SPENCER: Absolutely. It's fine to have your unique struggles, for sure. I think what I was referring to is more around embarrassment. If you're dealing with something that actually is super common, I think that helps us feel less embarrassed about it. Because it's like, "Oh well, other people are also dealing with the same thing." But I totally agree, obviously...

SETH: But I'd say that it would be nice if we also could be not embarrassed of it. We wouldn't feel ashamed about having some condition that's only one in a thousand. Although I agree, it does help. I think it is harder; we are probably more prone to embarrassment when we think something is rare.

SPENCER: Absolutely.


SPENCER: Okay, let's talk about your next book. How does your upcoming book differ from your last one?

SETH: My next book is about how you can use data to make better life decisions which, I know, is actually frighteningly similar to some of the work that you do, which is more evidence-based decision-making. I actually came up with the idea after I wrote "Everybody Lies." The theme of "Everybody Lies" is that you can't trust what people tell you. You have to look at their actual data, behaviors. After I wrote "Everybody Lies," I asked people, "What did you like about the book?" and everyone's like, "I like the sections on racism or child abuse. It's really important to uncover that." But Kindle actually now tells you what lines are most underlined in your book which, I would argue, is a different, more honest source of data. And I went through that data on Kindle and I found that people underline things about how they can improve their life, anything that relates to: How can I date better? How can I be happier? How can I lose weight? Or, how can I become richer? Anything in that domain was incredibly popular with people despite what they said. I think there's some tendency to downplay how narcissistic everybody is and how much people are looking for ways to improve themselves, rather than (let's say) learning about the world or improving the world. So I'd say my second book is just playing completely into this somewhat-secret narcissism in people, which is how they can use data to make better life decisions.

SPENCER: I love that you applied your own methodology to your own book. And all you narcissists out there, this is the book for you. Go buy it. When is it coming out?

SETH: Scheduled probably at the beginning of 2022.

SPENCER: Okay, awesome.

SETH: Hopefully, it'll be timed around New Year's. I think it fits some people's New Year's resolutions, I hope. But the broader thesis that I have is that, if you think of the book "Moneyball," which was famously about how the Oakland A's transformed baseball by basically using data instead of traditional methods or traditional scouts. Your intuition is flawed and data can find out who's really a good player, where people should really stand on the infield. And sometimes you do things that feel ridiculous but actually are right. So the shift has become really popular in baseball, I know Spencer... This is my favorite fact about Spencer, but I think he doesn't enjoy it as much as I do. Your grandfather...

SPENCER: Oh, yeah, My grandfather was Hank Greenberg, who's...

SETH: [interrupts] The greatest Jewish baseball player in history. But you'll probably cut this out of the talk right now.

SPENCER: [laughs]

SETH: Every time we're at a party, I bring this up and Spencer gets uncomfortable.

SPENCER: Well, the problem is, I will occasionally encounter people who can cite all the different statistics about him and make me feel slightly sheepish.

SETH: Sorry but yeah, Hammerin' Hank. So now baseball has been transformed due to "Moneyball." When I watched baseball growing up, the infielders all stood largely in the same place on every play. The third baseman was a few feet from third base, shortstop in between third base and second base, etc. Now infielders are all over the place. They're in the short outfield. They're in different places with different hitters. It seems crazy that you have these situations, the shifts where people are like, "You're leaving so much of the field open. How can that possibly be a good idea?" It feels crazy but it actually is right. So that motivated me. That's Moneyball baseball but I think, in our big life decisions, we tend to still go based on our intuition. So if we're picking a partner, deciding whom to marry, or trying to be happy or picking a career, I would argue that most of us largely rely on our intuition and don't do database things and don't do things that feel wrong (like shifting infielders) but might actually be right, be justified by the data. So I spent the last couple of years just talking to people, doing my own research but largely talking to other people, trying to find research in these big areas of life. We uncovered all the enormous data that we're now accumulating about human beings. What have we learned about these big life decisions that might help people make better life decisions and even do things that don't feel necessarily right but might actually be right?

SPENCER: I heard that you had some results about parenting and how we can use data to parent better.

SETH: Parenting is an interesting area. The first counterintuitive thing about parenting — this has been covered by a lot of people — is they've done a combination of twin studies and adoption studies. For twins, they compare identical twins and fraternal twins. You can break down and compare the outcomes, and you can use this. There's a formula where you can say how much of an individual's (let's say) income is driven by their genetics versus the environment they grow up in versus other factors. And then adoption studies, you can compare... There have been some programs where they've literally randomly assigned adoptees to households. You could say, "Okay, for different people who were just happily randomly assigned to the same household, how similar are their outcomes?" And it turns out, when you do this analysis, parenting matters, I think, a lot less than most people expected. There have been books written about this. Judith Rich Harris wrote a book about this. I think Bryan Caplan has written about it. The general studies found that where you grew up — the house you grew up in, the best possible place to grow up (like with the best family), whatever it is that makes a happy home great — it maybe adds 15% to your income, which isn't a huge amount. It's not as much as many would expect. And that includes everything that a parent does: where they're raising the kid, whether they read to the kid or how much they read to the kid, what school they send them to, whether they swaddled them when they're babies, how they feed them, everything goes into that about 15% additional income.

SPENCER: That result also applies across a whole bunch of outcomes. It's not just income, right?

SETH: It depends. Some outcomes, you can influence more. Education, parents seem to be able to influence more than income. Parents can help put their kid into a more elite university, but they have a harder time having their kid earn more money as adults. You could probably imagine reasons that may be true. They can help with the college essays maybe, but you can't necessarily make them a better employee for many years at a company. Parents influence different variables in different amounts. But generally in most things, parents influence kids less than we would expect.

SPENCER: My understanding is that you can break this down into three segments of influence. The first is genetics which you give to your kid. The second is the shared environment which, if you have two children, what's in common between their experiences: your parenting style, the house you live in, and the neighborhood they grew up in. And then there's the non-shared environment which is whatever is left over that's not part of genetics and not part of the shared environment. Is that the right way to break it down?

SETH: Yeah, and the general idea is that parenting, shared environment matter less than we think. Genetics matter a ton, and then also random things — the third group — matter more than people suspect, and we don't really understand why. For example, if someone is schizophrenic and they have an identical twin raised in the exact same household — that person shares the same genetics (they're identical twins) and the same shared environment (they're raised in the exact same place) — they're not 100% likely to be schizophrenic. I think it's something like 50%. If you have parents raising identical twins in the same place and one of them is schizophrenic, there's a 50% chance that the other one isn't going to be schizophrenic. So what caused that? That's a wildly different psychological outcome that was caused by something mysterious that's not genetics and not the place they were raised. So that's really interesting and mysterious. Psychologists still haven't really figured out exactly what's going on there.

SPENCER: I find that really interesting and I also wonder about this variable of non-shared environment because, for example, I am definitely aware of parents that treated their two children really differently. I think that would come into the non-shared environment rather than shared environment. Is that right?

SETH: One thing they think, even with identical twins, is that parents treat their kids... Having the same parents is considered the shared environment, not the not-shared environment. But one of the reasons that a shared environment scores low and non-shared environment scores high is because, even though two people share the same parents or the same household, as you say, they are treated differently. So even identical twins, maybe they show slight differences for some random reason, and the parents start treating them differently because of those differences. And then the differences get bigger and bigger and it's a little bit like chaos theory, a slight initial difference can lead to big differences down the line and they get reinforced in important ways.

SPENCER: Right. And then they can end up in different friend groups and then all kinds of influence occurs from that and so on. But I have to say that I have some skepticism around this research. I think you describe it accurately and it seems to be the scientific consensus. But I'll tell you where my skepticism comes in. It seems really clear to me that a parent can ruin a child. A sufficiently bad parent, it seems, can damage a child permanently, or at least for many, many years. And I don't know how to square that away with this. I mean, unless the argument is that that just happens rarely enough that it doesn't affect the shared environment that much. Or the parents damage their children differently. Maybe they damage one of their children but not the other and that would explain it. What do you think about that?

SETH: I think that's totally the case. There are extreme examples where you have parents that raise their kids as slaves or something and those kids are, I think just about 100% of the time, completely messed up. Largely, they would say that that is included in the variance of anything that makes the positive effects smaller because the overall variance that's caused by parental outcome also includes these extremely negative outcomes. So there are some effects of parents that includes, presumably, the negative effects and includes some of the positive effects. To the extent that there are extreme negative effects, maybe the positive effects would be even smaller than a naive translation from variance to top outcomes would suggest, if that makes sense.

SPENCER: If both are true that parents can really ruin their children and that there's very little influence, relatively speaking, of the shared environment, it suggests that perhaps it's not that common that parents ruin their children. Is that right?

SETH: And the other ruining that we think — like my mom hugged me too much, or was too attached to me or things like that — probably don't have as big an effect as we think. I don't know. Again, I think there clearly are cases where literally extreme abuse does ruin people. But maybe it's not that common, that real extreme abuse that would ruin people.

SPENCER: Another thing about this, and this is more speculative, but it seems to me that it probably is possible to make kids much better off if we understood how to do it. Even if it's true that, right now, the shared environment doesn't have that huge a consequence, it doesn't mean that theoretically, it couldn't have a huge consequence. It doesn't mean that we couldn't, one day, discover how to do amazing parenting that leads to incredibly high well-being, emotionally stable children or something like that.

SETH: Totally agreed. One place to start is probably using some of these studies and zeroing in on the 99th percentile parents and figuring out what they did. That would definitely be a promising area of research. I totally agree.

SPENCER: This is an interesting distinction between what's the average variance caused by this versus what could it be or how long could we make it. And a small average variance may mean it's hard to do but it doesn't mean it can't be done. I think that's also worth flagging. One third thing before we switch to another topic; I just want to mention this. It also seems to be true that people are often indoctrinated by their parents. If their parents are diehard ex-believer, whether that's a religion or philosophy or whatever, clearly their children are way more likely to be a diehard ex-believer than a random person who wasn't raised in that household. I wonder how that squares with this? Is it just that the outcomes that we're looking at here, things like (let's say) income or personality or whatever, are just not that related to these things?

SETH: I'd have to check but I think political and religious views, parents can influence their kids more on those dimensions than they could in income or personalities or some other traits. Like a sports team, which sports team do you like? That would almost be a sanity check for some of these studies. Does a shared environment influence the sports team? Shared environment which includes both the place you're raised and who your dad may root for or your mom would root for. It would be hard to imagine that it wouldn't have an enormous effect. But there's actually a second point on parenting. The first general idea is that a shared environment doesn't matter that much. There's also recent literature (I don't know if you've followed it) by Raj Chetty and some of his co-authors, where they've gotten access to comprehensive IRS data for a long period of time in the United States. Raj Chetty has just published one groundbreaking paper after another using this data. It's such a powerful data set and really hadn't been used much before. He and his co-authors got their hands on it and they did this study on the effects of neighborhoods on children. We know that kids who grow up in certain neighborhoods earn more money or have better life outcomes in various dimensions, but it's hard to know how much of that is causal. So if someone grows up in Beverly Hills, we probably expect that they probably would have a higher income than someone who grows up in the boondocks of West Virginia. But there's probably a lot going on there, not just the actual place they grew up. So he did this interesting study where, because he has comprehensive data on so many people — we're talking about tens of millions of Americans — he could really zoom in on parents that moved while their kids were at different ages. So parents who moved from one location in the United States to another, particularly when kids were at different ages... So maybe they have two kids, one of them is 15 and one of them is five, and they move from one town to another. Basically, if you think about it, you can use that data to really disentangle exactly how much the neighborhood matters by seeing how the five-year-old on average compares to the 15-year-old. So if every time someone moves to a particular town, the five-year-old does way better than the 15-year-old, we'd say that that location has to be helping kids in ways. Does that make sense?

SPENCER: That's a cool methodology. I like it.

SETH: If you actually do the math, you can, not just find out how much different places matter, but exactly how much each year matters. So the year between zero and one, one to two, two to three, all the way up to whatever age... I think he stops at 20 or something. What he found when he did this were surprisingly large effects of the place you're raised. I've done the math and I think being raised in the best place, the best location, on average, would add about 10% to your income. If in the same family, one kid got raised in what would be the best place according to this methodology, and one kid was in an average place, that kid would get about 10% higher income. I think these two literatures: the one we talked about, first the overall effects of the shared environment which is small, and the Raj Chetty and co-authors' work which shows that neighborhood effects are big...

SPENCER: Neighborhood is part of the shared environment, right? It's one facet of it.

SETH: It's a subset of the shared environment. It's really interesting. It basically says that the biggest decision a parent makes by far, at least on income and maybe a couple others — he hasn't tested all the outcomes — is the neighborhood you raise a kid. So if you put everything you could put together into the 15% shared environment bucket — I mentioned them earlier. You can even think of a lot more. Everything you do as a parent. Everything our parents stressed about: talks to their friends, reads books. "What should I do? What should I do?" — I would argue that probably everything you do, besides where you live, adds up to less of an effect than just where you live, which is really interesting and probably something that parents don't think enough about. And I think one of the reasons for this... Chetty and his co-authors have also found adult role models seem to play a really big role in what kids end up doing. For example, they found that a young girl who moves to a neighborhood with a lot of adult female scientists — he measured that by combining the IRS data with patent records — is much more likely to become a scientist when she grows up. And Chetty's also found the effects to be really hyper-localized, really small neighborhoods. So if you're raising your daughter around a lot of female scientists, she's much more likely to become a scientist herself. African-American boys do much better if they grow up on a block of Black fathers, not even their own Black fathers. That matters but what matters seemingly even more is how many Black fathers are they growing up around? Are you giving your young Black boy great Black male role models? You could imagine that one of the reasons it's probably so difficult for parents to have a huge impact on our kids is that the parent-child relationship is so complicated. A lot of the advice you give your kids, kids will sometimes take it and sometimes they'll rebel against it. So if you're a scientist, they may go through a period where they say, "Well, I don't want to be like my mom. I want to be something different." Or a lot of kids have complicated relationships with their parents and can sometimes do the opposite of what their parents are trying to encourage them to do. But the relationship with your neighbors is pretty uncomplicated. I think most kids view their neighbors as pretty cool. So you can almost trick your kids in parenting style by outsourcing it in various ways. Don't try to lecture them on what to do. Get out and show them the way of other people you want them to emulate as adults.

SPENCER: That's really interesting stuff. I guess I'm wondering, how much skepticism do you think we should have of this research? For example, when I hear results around the non-shared environment vs shared environment, it does get my alarm bells tingling like, "Hmm, should I really believe this?" It's a super counterintuitive finding. And I love the methodology of this new research you've been describing. It sounds really promising because it allows you to control for a lot of variables. But I also do wonder, should we view this as a tentative finding before we jump to conclusions about it? What's your thought on that?

SETH: I think we should basically always be Bayesian, which I think is something you'd probably agree with. When I say that neighborhoods raise your income by 10%, and shared environment raises it by 15%, those are exactly the correct numbers. I probably would say no. Again, I agree with you. And maybe they're a little counterintuitive, a little stronger than I'm willing to believe. Would I update my priors based on this research? Yes, I'd say probably, based on those two literatures. I'd say the effects outside of the neighborhood you pick are probably smaller than I would have previously thought, had I not read this material. And the effects of your neighborhood are probably larger than I would have guessed, had Raj Chetty and his co-authors not done those studies.

SPENCER: I think that's a great way to look at it. It's like, okay, you've got some priors and this is some more information. You shouldn't necessarily flip to believing it 100% if you originally strongly disagree with it, but you should be now more in the camp of like, okay, maybe this is true, and be less confident if you used to believe the opposite.

SETH: Again, I think that's the way to deal with all social science research, especially since there's so much difficulty replicating social science research, except the Spencer study. Spencer is the only one who fully moved my priors.

SPENCER: Well, no, I'm sure some of my studies are wrong. I wish I knew which ones and then I would track them. [laughs]

SETH: But I think that, anytime you read a study, I think you never go 100% toward that study if your prior is different, but you do adjust. And then how much do you adjust? Well, based on how much do you trust the researchers. How much incentive do you think they had for a splashy finding? I have people try to replicate it. You can do all kinds of factors that determine how much you adjust.

SPENCER: And the rigor of the research: is it a randomized controlled trial versus just a longitudinal study and so on?

SETH: Yeah, my friend once tried to do a machine learning analysis of whether psychology studies replicate and he found that only two variables explained everything. And it was basically social psychology studies — that was a negative variable — and studies by Kahneman and Tversky were a positive variable. And that was it.

SPENCER: [laughs] That's pretty funny. Yeah, there was some replication where they did 13 or something papers, and I think all the Kahneman and Tversky ones panned out, which was nice to see. On the question of, do parents have not that much effect on their children, I think one thing that's worth keeping in mind is, even if this research is true and you're not going to have a huge impact (let's say) on your child's income or something like that, it may well be the case that you will have a huge impact on your relationship with your child. So how you treat your child and the interactions you have with them could vastly affect whether you have a good relationship with them when they're older, and how well you get along when they're living with you, and so on. So this still could be a strong reason to care a great deal about how you interact with your child, even if it's not going to cause them to earn a significantly larger amount, etc.

SETH: Yeah, for sure, although even that's complicated. I have acquaintances who don't talk to their parents, where the things they've described to me don't sound like they were off-the-charts negative parents. I think they tried their best. Kids do go through angry, rebellious stages, sometimes independent of what you do. So that may be another area where the effects of shared environment turn out to be smaller than we would have predicted. But I think another thing about this research is, small effects don't mean it doesn't matter. They didn't find zero effects. So if you care about having a good relationship with your kids, and presumably most parents do, even if the effect isn't that huge, most parents would probably want to do just about everything they could to maximize the odds that they get the outcome they want. Same even with income, it could be that you can only increase your education or income or psychological health by a small amount. But I think parents care so much about how their kids turn out that they still would want to devote a lot of time to try to do everything right.

SPENCER: Good point.


SPENCER: What can you tell us about how to date better using data?

SETH: I'm still working on this chapter as we speak. This is largely summing up research from other scholars who've studied online dating sites. Again, I think the theme of this book is going to be how much new data there is to study people. We didn't have comprehensive IRS data 15 years ago, so we couldn't study how much each neighborhood matters for a kid. And we didn't have comprehensive online dating data, so we couldn't learn things about better ways to date. You might have already heard this one, but my favorite finding in the dating arena is, they've done studies of who does well in online dating measured in different ways, how many messages till they get the people they like to like them back. And if you look at the most successful online daters, I think they've done articles — I think they collaborated with OkCupid — and they've shown that the six or seven most successful online daters... And I'd say the absolute top of online dating success is exactly who you'd expect. They're conventionally gorgeous, gorgeous people, perfect shape, great bodies, aesthetic appeal. They have these quirky, fun, playful profiles. They check all the marks that I think we traditionally have thought make someone an attractive mate, so there's no real surprise there. But then, Christian Rudder wrote this book, "Dataclysm," (which is excellent) about what he learned from OKCupid dating data. He found that a group that does surprisingly well — not as well as the Natalie Portmans and Brad Pitts of the world — but better than just about anybody would expect, is women with shaved heads, which is pretty shocking. Heterosexual women with shaved heads do really well on online dating sites, which is pretty surprising. I don't think most people think of heterosexual women with shaved heads as sex symbols. And I don't think most people would think that that's a smart approach to dating. So what's going on? And I think what Rudder discovered was that a really important factor in how well you do in dating is not just the mean of how other people think of you, how attractive other people find you, but the variance. So having a high variance can be really good. Because really, in dating, you don't want a lot of people to think you're okay. And if you're not Natalie Portman or Brad Pitt, you're going to struggle for a huge number of people to think you're really attractive, but you can increase your variance so that some people will be really turned off with you. So a woman with a shaved head, probably a lot of people see them on OKCupid or Tinder or Zoom, just immediately say no, and kind of almost laugh at how unattractive they are and how ridiculous they look. But some small group of people is gonna be really, really, really into them, which is kind of underrated in dating, having a small group that you really appeal to. And I think that that is really counterintuitive. I think, to try to make themselves more attractive mates, people frequently try to make themselves more conventionally acceptable to a lot of people. So you try to dress in a more conventionally attractive way, or have a more conventionally attractive haircut, or have a more conventionally attractive personality. And I think people probably underestimate the success rate of a strategy that's more about getting an extreme reaction to you that could be extremely negative, but could also be extremely positive.

SPENCER: So one thing that that reminds me of is this idea that the lower the mean is, the higher the variance would actually be desirable. If you were super, super attractive to the average person, increasing variance wouldn't actually help you. But if on average, people didn't find you that attractive, then increasing the variance would actually substantially increase the number of people that find you attractive. Is that right?

SETH: That's a great point. Although Natalie Portman, I'm pretty sure, did also shave her head for one period of life, and I think she was still incredibly attractive to everybody. But yeah, I think that is a great point. That makes a lot of sense. I've also analyzed data from Pornhub, which I talk about in "Everybody Lies." And I think one of the things that is striking — which gives an unvarnished view of physical attraction, what people really desire — I was kind of shocked by how much variance there is in what people find attractive, because that plays into why a woman with a shaved head does well. We might think heterosexual men, when they're at their computer, really saying what their deepest fantasies are, would search for a woman with a full head of hair. Well, no, a good percentage is searching for a woman with a shaved head. And you see this in many dimensions. Weight, we think conventionally attractive is being thin. Well, a large number of people are searching for really heavy people. And on anything you can think of, there's more variation than we sometimes think of in what people find attractive. I think there's also a sense in which people are attracted to characters in interesting ways, like almost extremes. So someone with a shaved head is showing with their appearance and possibly with their personality as well, a really extreme personality. I think when you look at what people watch in porn or what people fall in love with, you have a better chance if you're a character in some sense, or you're extremely something instead of just kind of 'meh.' That's another reason that playing extreme outcomes can be effective.

SPENCER: I think this applies in the reverse direction, too. If you're trying to find someone that's really, really well suited to you, if you're trying to select on the things that everyone's looking for, like if you're trying to select on, "Oh, I want them to be extremely conventionally attractive," then it's actually going to be really, really hard to find someone because you're going to try to find someone 99th percentile conventionally attractive, but so is everyone else, and the competition is just incredibly demanding. But if you can find some trait that you find really, really attractive — and higher and higher levels of it are even better for you — that most people don't care about, then you might actually be able to find someone in the 99th or even higher percentile on that trait, and that's incredibly good for you with very low competition.

SETH: Yeah, my friend Ted said this thing I thought was really interesting. Who are people you're unusually attracted to, have traits that you are unusually into? And who are people who find you unusually attractive? And then finding that sweet spot where that Venn diagram overlaps is a great place to be to date. Yeah, I think I totally agree. There have been studies which are looking at what correlates with long-term happiness and relationships. And I think it's pretty clear in the data that conventional attractiveness does not correlate with long-term happiness, that two, three, four years down the road, you're not any more likely to be happy if you're with a partner who's in the 99th percentile of looks (or the 98th percentile or 70th percentile), that other things seem to correlate much more with happiness. Findings like that, I can get a little skeptical of. They strike me as very politically correct. It's very nice to say looks don't matter and don't think about it. And advice like that is incredibly difficult to follow. If you look for what most people look for in a partner, people put huge emphasis on looks. People can tell you until they're blue in the face that looks don't matter and people have a hard time following that advice. I think it does make people feel good to say that it doesn't matter. I'm always a little skeptical of results that make people feel good or sound good. But I think legitimately there is strong evidence that in a long-term relationship, looks are really overrated. So independent of just avoiding the competition, I think anybody would probably be wise to care less about looks. Even if you are Natalie Portman or you are Brad Pitt, probably paying attention to the factors that correlate more with long-term relationship success — like the attachment style that a person has, or whether they have a growth mindset — would be wise, compared to trying to maximize someone's looks.

SPENCER: It seems to me there are a few different reasons why people try to date someone who's good-looking. One of them, which seems like one of the usually least valuable to me, is because they want to impress other people. They want others to see that they're with someone attractive. Okay, that's fine. We all want to look good to other people. And so I don't think there's anything inherently bad about that, but maybe it's not the greatest reason to want to date someone good-looking. A second is a sense of identity. Some people might just say, "Oh, I'm the sort of person that dates good-looking people," or something like that. So it's more about yourself, and it's not directly about how others are going to view your partner. The third is for sexual attractiveness reasons where you want to have a good sexual relationship and being with someone that you find sexually attractive may enhance that. And then the fourth is, in the same way that we like being around beautiful nature or beautiful art, some people can just find it enjoyable to be around people that they find beautiful in a similar way. So I would be really interested — I doubt anyone would say this — but I'm really interested to see if it actually matters to what extent those four different motivators are coming into play, and if that affects the effect of someone's physical attractiveness on their marital or relationship satisfaction.

SETH: Kids will also be another one since there is a genetic component to attractiveness. The looking good in front of other people is an interesting one, too, at least for me personally... Larry David always said, "I'd never trust someone whose wife is too beautiful." Sometimes I see someone with a ridiculously attractive person where I'm like, they met them in another country, maybe they're happy with them (I don't know), maybe they have a lot of great qualities. But to me, it just at least makes me think that they're really insecure in some way. Whereas if I see someone with someone who may not be the most conventionally beautiful person but they seem to have a really good dynamic, I always think more highly of them. I think it's interesting that we think we're impressing other people — and obviously, we do impress people to some degree — but I think how people react to the conventional attractiveness of your mate is a little more subtle than I think some people might think or realize.

SPENCER: There's this rule of thumb if you go to a cocktail party with impressive people: the person you want to talk to is the person who's both poor and not good-looking. That's probably the most interesting person in the room

SETH: Because they had to overcome?

SPENCER: Yeah, because if they're both poor and not good looking, and they're in this room of impressive, high-roller people, there's probably something really amazing about them. Because they don't have the superficial attributes, they hopefully have the deeper attributes.

SETH: Yeah, I've heard that. I think Nassim Taleb said, when you're in a top hospital, find the doctor who looks the least like a doctor, like someone who looks like a butcher or looks like a schmuck. Because we are so superficial, he or she would have to have been a much better doctor to rise to that level than someone who looked like George Clooney.

SPENCER: That's really interesting. One day, when I was in college, I'd started a new class in the math building. And there was a man standing at the front of the class right before the class was to begin. He was wearing no shoes. He was literally chewing on his tongue and he had chalk all over his face. And I was like, "Who is this person?" And it turned out, of course, he was our teacher, who's a great mathematician, but he literally seemed like a crazy person who walked in off the street. But he was just a great mathematician and didn't give a shit about anything else.

SETH: Math's an interesting one, because there are even examples where mathematicians have intentionally adopted the weird traits of John Nash. I think in math, there's kind of an idea that coming across as a weirdo or disheveled can be an advantage. Certainly, it plays into...

SPENCER: That's my excuse.

SETH: [laughs] Come on, you know it's a show. But maybe you actually want to hire that mathematician who looks like Mitt Romney, because looking like Mitt Romney might be a bad disadvantage in the world of math, where people wouldn't take him seriously because how could somebody look like that and be a really good mathematician? I don't know.

SPENCER: Do you want to talk about what big data can tell us about happiness?

SETH: Yeah, there's this really interesting project. I don't know if you know it. Do you know the Mappiness Project? George MacKerron, a professor in England, led this project, where people basically download an app on their iPhone, and they report what they're doing at different times, and how happy they are. They also report, I think, how anxious they are, maybe how tired they are, how stressed they are. They're pinged at random times and it's this amazing dataset, both because they know what people are doing and how happy they are, and also because it's an iPhone, they can easily get data on where the person is exactly, their latitude and longitude. They can find it and compare it in interesting ways to what's going on in that particular environment. But I think it's this totally revolutionary study of happiness. It just found all these really amazing things about what makes people happy with this dataset. So actually, you mentioned that people like to be around beautiful nature and beautiful stuff. I guess you figured that out without data. That's actually one of the things that they found in the data when you compare the same person doing the same thing at the same time. If they're in a natural environment, particularly near water — water is really valuable — they're much happier than if they're in an unnatural environment. And there's another website where people have rated the beauty of every inch of the planet, basically. And they found that if people are in more beautiful environments, they're more likely to be happy, again, doing the exact same thing at the exact same time with the exact same people. One of the studies I found most interesting, because I'm an enormous sports fan, is their study of sports. They basically studied how sports fans react to a team's performance, whether the team wins or loses — they were actually doing soccer so it was a win, lose or draw — and they found that, when your sports team wins a match, you gain about four points of happiness. And when your team loses, you lose about ten points of happiness.

SPENCER: Oh, man, so it's asymmetrical,

SETH: Totally asymmetrical and it's actually Interestingly similar to a drug in various ways, that people seem to get really excited before a sporting event and then they get this, on average, horrible outcome. The most miserable activity, not surprisingly, is being sick in bed. If someone's sick in bed, that's the lowest happiness they'll report. Being a fan of three teams — let's say, I'm a Mets fan, I'm a Knicks fan, I'm a Jets fan — so if I'm a fan of three teams, over the course of a year, it's basically the equivalent of being sick in bed for an extra week.

SPENCER: Because of every time they lose, basically? [laughs]

SETH: Yeah, their wins aren't giving me much pleasure and their losses are giving me a huge loss. But if you actually do the math, okay, that's a three-point loss on average. Compare that to other activities. I'm basically giving myself some severe headache for a week of a year or something, which is a pretty huge effect. And I go through huge lengths to avoid that level of misery. Certainly, I'm locking myself in my apartment to avoid COVID, which probably would be the equivalent of a sick week in bed for somebody of my age.

SPENCER: Yeah, well, at least your team losing doesn't have a small chance of killing you, so that's an advantage.

SETH: Or killing my parents, although my dad's as big a fan as I am so...

SPENCER: Do you think that when people root for teams that are less likely to win, even though they're losing more, which hurts more, it might make up a little bit that, when they win, maybe it's even more exciting and you get more than those four points?

SETH: Yeah, they did that analysis. But basically, all that means is, there's no way to outsmart the system. My dad, for example, he was a Mets fan for most of our lives growing up, a huge Mets fan. There are pictures of him storming the field at Shea Stadium. And one of his proudest achievements when he was a boy, was when he won banner day, which was this competition to come up with the most clever banner in supporting the Mets. Just an enormous Mets fan. One of my earliest childhood memories is having him jumping up and down in our house when the Mets won the 1986 World Series. I was four years old. Just huge, huge Mets fan. And one day — he was probably in his 50s then — he switched from a Mets fan to a Yankees fan. This was during the 1990s when the Yankees were a dynasty. He goes, "Seth, life's too short to root for a shitty team." It's just like, "Yeah, I can't do this anymore. The Mets are causing too much pain." Basically thinking — okay, kind of realizing — "I'm losing all these happiness points when the Mets lose as they consistently do. Let me try the Yankees." But if you actually do the analysis, you explained that the negative hit is smaller when they lose, it also means the negative hit is bigger when a good team loses. So when the Yankees are really good, the wins give you less pleasure, and the losses give you more pain. So basically, what it means is, there's no way to escape the trap that rooting for a sports team is going to, on average, hurt your happiness.

SPENCER: If only we could believe our team will always lose, and then they're actually a good team, and so they win, that would probably be the ideal.

SETH: Yeah, that's a good point. Maybe you could trick yourself.

SPENCER: Seth, thanks so much for coming out. This was a lot of fun.

SETH: Thanks, Spencer.





Click here to return to the list of all episodes.


Sign up to receive one helpful idea and one brand-new podcast episode each week!

Contact Us

We'd love to hear from you! To give us your feedback on the podcast, or to tell us about how the ideas from the podcast have impacted you, send us an email at:

Or connect with us on social media: