Keeping Tabs on ‘Hate’ Through Google Searches

IRA FLATOW: Hours after the December 2 shooting in San Bernardino, The New York Times reported that the top Google search in California with the word Muslims in it was kill Muslims. Economist and former Google data scientist, Seth Stephens-Davidowitz, calls Google searches a modern day confessional box. What can Google searches tell us about public sentiment on controversial views that polls and surveys cannot always measure?

Seth Stephens-Davidowitz is here in our New York studios. Welcome to Science Friday.

SETH STEPHENS-DAVIDOWITZ: Thanks so much for having me, Ira.

IRA FLATOW: Why do you call Google searches a modern day confessional box?

SETH STEPHENS-DAVIDOWITZ: Well, just to clarify, this is research with Evan Soltas, as well. And the idea is that sometimes when you type into Google, we think of it is an information source. So you type in a question you have. What’s the weather going to be tomorrow? Or, who won the game last night? Or, what’s a good recipe for apple pie?

But sometimes, people type just random thoughts they have into Google. I hate my boss. People are annoying. I am drunk. And these searches tend to feel like confessionals. They’re admitting things that they might not want to admit in polite company or might not tell to a survey. So it gives a different perspective and often, unfortunately, a darker perspective of the human psyche.

IRA FLATOW: Is there any predictive value in this?

SETH STEPHENS-DAVIDOWITZ: Yes, so that’s what this research was doing with Evan, is that we were predicting over the last 2004 to 2013, can you see by what people search into Google anonymous aggregate data, can you predict when hate crimes against Muslims will be high. So when people type things like these very disturbing searches, like kill Muslims or I hate Muslims, are hate crimes against Muslims higher?

And the answer is unambiguously yes. There’s a clear relationship on weeks when these searches are higher, there are going to be more anti-Muslim hate crimes. And what we also say is that this means we predict that anti-Muslim hate crimes are at levels not seen since the aftermath of September 11 right now because these nasty searches about Muslims have shot up recently.

IRA FLATOW: Were you able to see any trends in where this was all happening? Different populations?

SETH STEPHENS-DAVIDOWITZ: We did a little bit of that. We have to look into it more. I think that is obviously another power of the Google search data. If you take up a survey of let’s say 1,000 or 2,000 people, you’re not really going to have detail on different states, let alone cities or towns. But because Google searches are such big data, they have so many observations, you can really zoom in on very small communities and test a lot of hypotheses about hate.

IRA FLATOW: You would really need to get into a person’s head to gauge Islamophobia and hateful sentiment towards Muslims. You try to tackle this with key phrases, right?

SETH STEPHENS-DAVIDOWITZ: That’s exactly right. And I think one thing to clarify is it doesn’t mean– first of all, this is anonymous aggregate data. We don’t know the reason that anybody makes a search. It could be the most horrible search, the most horrific search possible. And someone could just be curious, or doing research, or there are plenty of other reasons to make a search. But on average, when these searches are high, we see that hate crimes against Muslims are high as well.

IRA FLATOW: After the attacks in San Bernardino, President Obama gave a speech asking for tolerance. And you used Google search as a real-time tracking of public opinion.

SETH STEPHENS-DAVIDOWITZ: That’s exactly right. So this is another advantage of the Google search, is you really do have it down to the minute. So Evan and I, right after this speech, we had already been working on the article and the research. We were excited, can we see, has this helped things?

Because the speech was very moving. He talked about the importance of not judging people based on their religion. Of taking people into this country, regardless of their religion. American values, freedom over fear. And we got very excited, let’s see how the searches have responded to these well-meaning words.

And we found out, disappointingly and disturbingly that if anything, it seemed to just provoke the angry mob or provoke intolerance. That searches for kill Muslims shot up threefold. And searches for how to help Syrian refugees dropped 30%. Searches against Syrian refugees, no Syrian refugees went up 60%. So if anything, it seemed like Obama’s speech on average backfired.

IRA FLATOW: I’m Ira Flatow. This is Science Friday, from PRI, Public Radio International, talking with Seth Stephens-Davidowitz about Google searches. He’s an economist here in New York. But there was one interesting thing about the speech. When the president turned a bit and started talking about Muslim American soldiers, Muslim Americans sports figures. Right.

SETH STEPHENS-DAVIDOWITZ: That’s exactly right.

IRA FLATOW: It’s like a whole different thing happened.

SETH STEPHENS-DAVIDOWITZ: That’s exactly right, Ira. So there was one line where he talked about how Muslim Americans are our friends, our neighbors, our sports heroes. They’re the men and women who are willing to die for this country. And after that line, for the first time in a year, the top noun search with Muslims was not terrorists, extremists, or refugees. It was athletes.

People want to see, want to learn more about these Muslim athletes. And number two, after athletes, was soldiers. People want to learn more about the Muslims who have signed up to serve the nation. And athletes kept the top spot for more than a day afterwards.

And we said on Twitter there was also a lot of tweets, a lot of fans, people who had grown up rooting for Shaquille O’Neal the famous basketball player, were now tweeting, in a bit of shock, wait, Shaquille O’Neal is Muslim.

So that was a little bit different than the more generic statements that Obama was giving that seemed to just inflame people. This was more new information, piquing people’s curiosity and really changing what people thought about Muslim Americans. So that had more positive effects.

IRA FLATOW: Perhaps that shows the power of speech, doesn’t it, saying the right things.

SETH STEPHENS-DAVIDOWITZ: Exactly right. But it shows that, and it shows that it’s more subtle than we usually think. That a lot of people, if you say something like, we need to appeal to, we need to care more about freedom and not give into fear and that we can’t judge people based on religion.

When you say things like that, afterwards, you’re going to get rave reviews from all the serious people for saying these things. But that actually backfired. But there are subtle things you can do where you really pique people’s curiosity. And you can with words change how people think about a group.

IRA FLATOW: What kind of data would you like more of? What’s your holy grail that you’re looking for?

SETH STEPHENS-DAVIDOWITZ: I think Google searches are pretty good. I’ve actually done a whole bunch of research on this starting about three years ago. And I’m writing a book about it that everyone should look out for. And that’s coming out in a few months. But I think it is like a confessional. And people are telling things. I don’t think many people are going to tell a poll that right after San Bernardino they had an urge to kill Muslims. But that is something that people do type into Google. And when you look at it, anonymous aggregate, you really do get a really unprecedented view into the mind.

IRA FLATOW: Is there any way of making this more scientific about what people think and how they act afterwards?

SETH STEPHENS-DAVIDOWITZ: Well, we think our work was pretty scientific. But we are writing it up as a scientific publication. And there are more details where we flesh out a little more of the models we used and stuff. But we think that this will show. And we have already shown it. And in the paper, we will clarify in more detail that there is clearly a way to tie people’s thoughts– kill Muslims, I hate Muslims– to people’s actions– anti Islamic hate crimes.

IRA FLATOW: This all sounds so much a little bit like Minority Report, where in the movie they could predict a crime, a killing that was going to happen before hand.

SETH STEPHENS-DAVIDOWITZ: It raises some difficult questions. We didn’t really go there. But it does raise some questions.

IRA FLATOW: So what what’s next? Writing the paper? Where would it show up? Where would you publish it?

SETH STEPHENS-DAVIDOWITZ: Well, we’re going to send it to a variety of journals. So, it’ll be on our websites initially. Evan’s and my websites, initially. And then, eventually to an academic journal.

IRA FLATOW: Well, I want to thank you for taking the time to be with us. Well, this is a new age, isn’t it, following Google. Can you do anything with Facebook, any Facebook kind of things the same way with Google?

SETH STEPHENS-DAVIDOWITZ: I think Facebook data is a bit overrated. Because it’s public, people don’t type to Facebook. You usually type that you’re in the Caribbean, and how much you love your husband, and your family’s all great, and you love life. Whereas Google is a little more pure because it’s not broadcasting to all your friends.

IRA FLATOW: I got it. Thank you very much, and good luck. We’ll be following this.

SETH STEPHENS-DAVIDOWITZ: Thanks so much.

IRA FLATOW: Have a happy holiday.

SETH STEPHENS-DAVIDOWITZ: You too.

IRA FLATOW: Seth Stephens-Davidowitz is working on all this stuff. He’s an economist based here in New York.

Copyright © 2015 Science Friday Initiative. All rights reserved. Science Friday transcripts are produced on a tight deadline by 3Play Media. Fidelity to the original aired/published audio or video file might vary, and text might be updated or amended in the future. For the authoritative record of ScienceFriday’s programming, please visit the original aired/published recording. For terms of use and more information, visit our polices pages at http://www.sciencefriday.com/about/policies.