Everyone’s been in that situation wherein you just couldn’t sleep and you have your late night thoughts about the universe and everything in it.
You’ve actually self-diagnosed yourself time and time again, and you’ve had thoughts about extraterrestrial life outside planet Earth.
Well, Dr. Seth Stephens-Davidowitz, a former Google data scientist with a PhD in economics from Harvard tells VICE what our late-night search habits are and what they mean.
Check out the interview below:
VICE: How do our search habits change late at night?
Dr Seth Stephens-Davidowitz: Late at night there’s a lot of anxiety, panic attacks, and health concerns. People at 3 AM wake up in a cold sweat worried that they have cancer or Parkinson’s disease, or that they have ALS or brain tumors. There’s also a lot of late-night marijuana smoking. Between 2 AM and 4 AM, people ask the big questions, like: What is the meaning of consciousness? Does freewill exist? Is there life on other planets?
Perhaps one of the reasons for this is the use of marijuana, because people google “how to roll a joint” between 1 AM and 2 AM, so that’s one of the things you see. You also see late-night horniness—”porn” is most popular between midnight and 2 AM.
Presumably the data is pretty different depending on what day of the week it is?
Yeah. So at 3 AM on Saturday, you’re more likely to see people are drunk. People make a lot of spelling errors, and they forget their passwords.
Did your research show how gender affects our internet habits in the early hours of the morning?
One thing I did see is that Literotica [an erotic fiction website]—which is more popular among women—peaks more in the early morning. So maybe we can see women get hornier in the morning and men get hornier at night. We don’t break down the gender of the searches, so we don’t really know—on the internet, nobody knows you’re a dog.
What about spikes in searches related to taboo topics or illegal behavior?
This is more disturbing. At 3 AM, people make searches like “kill Muslims,” and these actually predict hate crimes later. These people are probably not the most sane members of society, but they reveal themselves in Google searches, and we can potentially use this information to learn what causes these unstable people to have violent, dangerous thoughts that they sometimes act on.
Do you ever hand any of your research over to the police?
I’ve talked a little bit about police departments potentially incorporating some of these findings. If there is an unusual number of searches for “kill Muslims” or “I hate Muslims,” then police departments would be wise to put extra security outside of mosques, for example.
“All the lonely people kind of reveal themselves on Google.”
Do different nations search for different things at certain times of the day?
One of the things I looked at was lunch breaks around the world. In the United States, there are various different things people do—at 12 PM on a weekday, people catch up on the news. In Japan, people plan for travel. In Belgium, people shop during lunch.
What can the data teach us about the human psyche?
With porn, health concerns, and metaphysical questions, I think what is happening is a combination of boredom, anxiety, and marijuana smoking during these late-night hours. More generally, what’s noticeable in this data are the clear patterns that reveal themselves when you’re dealing with 200 million people. A lot of people think that their situation is unique and that there’s something very particular about them, but when you put them altogether with everyone else, you see a large number of people are doing the same thing at the same time. In some sense, people are more predictable than we sometimes think.
So, in a way, some of this research could make us feel less alienated, as it shows we’re just one of millions of strangers fretting and pondering in the middle of the night.
Yeah. All the lonely people kind of reveal themselves on Google. What’s really exciting is that all of this data is unprecedented. For example, we didn’t really know previously—and some of this is preliminary—how many people were having a panic attack at 3 AM on a given Tuesday evening. But now we can start correlating this with what happens on Tuesday afternoon in New York City and why an unusual number of people had panic attacks later that night. That’s a question that really would have been impossible to answer for all of human history, but now it’s possible to answer—although I haven’t figured that out yet. You’ll have to read my follow-up book, I guess.