The Psychology of Search Engines (Archive Episode)

0:00Hello. Thank you for downloading Everything Psychology. This is an archive recording from last year of some of the most popular episodes that maybe you haven't heard. I'm now busy recording brand new episodes which I'll release in the spring. And if there's any topic you'd like me to cover, then please email me directly on paul at everythingspsychology.com. I hope you enjoy the show. Coming up in this episode...

0:30I've been doing this research for about 12 years now, but I'd say it's only within the last year or two that I finally realized that the search engine is a very peculiar form of control. The search engine is absolutely unique because it's training you like you're a rat in a Skinner box. One of the scariest leaks from Google is an eight-minute video which is called The Selfish Ledger. It was made by the Advanced Products Division. It was never meant to be seen outside

1:03the company, but it leaked. And it's very clear about re-sequencing human behavior according to... It's right there. It's right there in the narration according to Google's values. Well, you've scared me even more now. Thank you. I'm going to go off and put a little tinfoil hat on.

1:27The extent to which the listener judges the responses is true, valid or correct is governed by the extent to which comparable responses by the same speaker have proved useful in the past. That was a quote by B.F. Skinner in 1957, writing in his book Verbal Behavior. Skinner here refers to a really simple concept. We pay more credence to people who have, in the past, proven to give us correct answers. Now, imagine you were appearing on the TV quiz show

1:59Who Wants to Be a Millionaire? and were asked to choose someone for the Phone a Friend lifeline. You're just much more likely to choose someone who you have had past experience of giving you good answers to various questions. Of course you would. Now, when it comes to fact-based questions, where there is a right and a wrong answer, that's fine. But what about opinion-based questions, such as, what clothes should you wear? What car should you buy? Or who should you go and vote for?

2:29What if, unbeknownst to you, your friend had a hidden agenda and wanted to persuade you rather than simply give you facts? Now, hang on a minute. I hear you say, I thought this episode was about search engines. Now, it is. I'm so excited to be talking to Professor Robert Epstein about how the digital technology we all rely on every day might not be the friend you thought it was. Hello, Robert. Hello. Nice to see you. You too. I am really, truly delighted to be speaking with you because

3:05I think the work that you and your team have been doing, I think over the last decade or so, is truly startling and yet so few people know about it, at least maybe in the UK. And it shows how tech corporations, and I think specifically Google, Facebook, are employing techniques from behaviorism to manipulate our beliefs and opinions to influence global politics and beyond. Now, that's a massive statement. So, to start, can you summarize your hypothesis about that?

3:39Well, I think you said it quite brilliantly just a minute ago. If for any reason you come to trust a speaker, then you're very likely to believe that speaker when finally one day you ask a really tough question or you call it an opinion question. I would call it an open-ended question. And you're going to trust that person. Now, it turns out that about 84% of the queries that we type into the search

4:18bar on Google, but 84% are for simple facts. What is the capital of the UK? I'm guessing it's London. It is London. Good, good, good, good, good guess on my part. The point is that 84% of those queries give us the correct factual answer, at least on Google they do, right at the top of that list of search results. So, this is something I only figured out pretty recently. I've been doing this research for about

4:5412 years now, but I'd say it's only within the last year or two that I finally realized that the search search engine is a very peculiar form of control. It's very odd. It's giving you lists, list after list after list. And psychologists have been studying list effects for well over 100 years. A man named Ebbinghaus, a German, he worked on himself. He tried memorizing lists of nonsense syllables,

5:31and he found out generally that he would remember the first item on the list best, and then the last item on the list almost as often. But the stuff in the middle, you know, tougher to remember. Yeah, people also recognize that as the primacy and recency effect. It's quite sort of well-known in popular psychology, isn't it? Yes, it's one. I prefer to Google. Oh, very, very much so, because this kind of effect is one of many list effects that have

6:02been studied. There's one that's relevant to the issue you raised about elections. It turns out that if you list candidates in a fixed order, say alphabetical by last name, well, being at the top of that list gets you more votes. Quite a few, actually, 5% or more, which can easily flip a close selection. So there are many list effects. But the search engine is unique, absolutely unique in the kind of list effect it has, because it's training you like you're a rat in a Skinner box,

6:35because 84% of the time, it's reinforcing clicks to the top item. It's reinforcing attention to the top item, because it's giving you the correct answer at the top. So when you're Googling a fact-based question, as you said. Yes, and which is most of the time. So it's a list effect that is supplemented by a regimen of operant

7:05conditioning that never stops. Now, this normally doesn't happen with other list effects. In fact, I don't know any other list effect like this, where there's this constant training going on. So the point is, when you do finally say, who's the best candidate? Or you say, what's the best restaurant in London? The fact is, you are going to trust what's at the top. And I thought, well, you know what, if they have that much value, if people think that higher results are somehow better or truer, could I use high-ranking search results to shift people's thinking or opinions or votes?

7:44Could I do that? And so I set up, you know, I did what I always do. I set up a controlled experiment. I made a prediction in advance, which I always do. I thought that if I put favorable search results, that is, search results leading to web pages that make one candidate look really good compared to the other. So if I put favorable search results for one candidate at the top for one group of people, and for another group, and this is with random assignment, I put the opposing candidate at the top,

8:20could I shift people's opinions about the candidates? And could I even shift perhaps their voting preferences? I thought I could. I think I predicted that I could produce a shift of two or three percent, which doesn't sound like much. But as a matter of fact, in a very close election, and a lot of national elections are very close, that could flip an election. So that was my prediction, two or three percent. So we ran the experiment with a pretty good representative sample of voters in the U.S. In other words, we didn't do this in a classroom. We didn't do it with college

8:54students. We actually recruited, you know, representative sample of voters. And we got a shift in that first experiment of, I believe, 43 percent. Which is a little higher than the two, three percent you thought. Oh, how very British of you. A little higher, yeah. So I thought it was a mistake, because that happens all the time in research. So we replicated the experiment with another group, which is expensive, because we have to pay everybody. And we got a shift of,

9:30I think, over 60 percent. We did it a third time, again, got an enormous shift. And along the way, I began to wonder whether we could mask what we're doing. So... What do you mean by mask? Well, mask would mean that normally, without a mask, I have pro-candidate A, pro-candidate A, pro-candidate A, and so on. And that's for one group. For the other group, I have pro-candidate B, pro-candidate B, et cetera. But when I mask, it means I want to confuse people a little bit. I don't

10:03want them to see the bias. In these early experiments, people could see the bias. Usually about 25 percent or so could see the bias. We learned later that that actually did not help people at all. People who see the bias shift even farther in the direction of the bias. Even if they were consciously aware that, hang on, something's wrong here. I'm seeing the same candidate stories coming up again and again. And they twig. That still doesn't matter. They still are influenced because of that. Is that what you're saying? No, I'm saying that they shift even farther in the direction of the bias. I'm saying that they say

10:38to themselves, oh, look, the search engine is preferring one candidate. It really must be good. It must be a wonderful candidate because people trust, which is very, very poor thinking on our part. But people trust computer output. They trust algorithms because they don't know what computer output is and they don't know what algorithms are. So they think that computer output is inherently impartial and objective, which it is not. It is positively controlled by programmers. And as you say, you started to talk about, you introduced the ideas with the words conditioning,

11:13reinforcement, skinner box, all these words from behaviorism. And you wrote something in one of your papers in 2024. And it said, people will always be unaware that the process by which they make both trivial and important decisions is being affected by a perpetual regimen of operant conditioning, as if they were rats trapped forever in an operant chamber. That's really strong, isn't it? And if that effect can happen through that conditioning process,

11:44we believe the top results are the best results. And the people in charge of these algorithms are able to manipulate them. Is there a motive for them to do so, though? Isn't it their job to try and give us a balanced opinion in these areas? No, no, no. In fact, that's another thing I didn't realize until I was well into this research. They're not motivated to give you a balanced anything. In fact, quite the opposite. The search engine

12:16is designed to be biased. It must be biased. You want it to be biased or it will be worthless. In other words, you want it to show you the best dog food. If you're asking about the best dog food, you don't want it to balance all the different dog foods out and make them equivalent. So it does that for everything, including political candidates. Now you get to the particular platform or company that's

12:47involved. When it comes to Google, well, they lean in the US terms, they lean very far to the left. I lean left myself. So, I mean, I sympathize, but I don't like the fact that they have so much control over elections and people's thinking and that they can indoctrinate children and that they are a threat to human autonomy, you know, all that stuff. But they do lean strongly left. So 95, 96% of their donations go to Democrats in the US. Okay. Uh, the head of Google, uh, volunteered in 2015 to

13:25literally to head up Hillary Clinton's political campaign. I mean, that those, we learned that through leaked emails, uh, 2016, uh, Google was, uh, Clinton's largest donor, uh, 2020, uh, Google was Biden's largest donor and so on. So they have very strong, uh, political bent. And so they have lots and lots of, uh, of reason to want to shift votes. And we learned over the, over a period of years with

13:59lots and lots of research. And finally, by setting up a nationwide monitoring system, the first in the world, but we learned not only that they have access to these techniques, but we learned that they actually use them. And so we learned that. Yeah. Have Google admitted that that's what they do? Uh, no Google, uh, their PR people and their CEO, uh, even under oath, even testifying before Congress, they straighten, straighten out, just lie. They absolutely tell blatant lies about many things

14:34that they do. And, you know, there's no penalty. I mean, uh, in theory, it's a felony to, uh, to lie to Congress, but Congress for many, many years now has not enforced, uh, that, that rule, that law. So they lie. Uh, I mean, I, I'll give you a quick example. First time I testified before Congress, uh, there was a high ranking Google executive testifying before me, he was asked under oath by a U S Senator, uh, does Google have blacklists? And he said, no, Senator, we do not. Now, just a few weeks later,

15:12what do you mean by a blacklist? Sorry, Robert. Oh, well, get to that in a second. Okay. First, I want to just tell you just a few weeks later, uh, a senior software engineer at Google who'd been there more than eight years, he turned him, turned into a whistleblower and he left the company, walked out with 950 pages of documents and a video. And three of the documents were labeled blacklist. Now, now there, there's an arrogance here. There's an arrogance here because if I had blacklists at my

15:42company, I would not label them blacklists. I'm might call them shopping lists or something, but I would not call them blacklists. The point is, uh, what, what are they for? Blacklists, uh, is a cheap and easy way to tell an algorithm what to suppress. So instead of reprogramming your whole algorithm, when you want to support one candidate, let's say, all you do is you add that candidate's name to either a blacklist or a whitelist. And the program automatically always is checking the blacklists and the whitelists

16:15uh, so it knows what it should suppress at the last second or what it should elevate at the last second. So it's a very, very simple, easy way. And in other words, you don't need a programmer to make the change. You just need somebody who is, is either decides or is, or is directed to add, let's say, Hillary Clinton to a blacklist. Uh, we have a Senator here in Massachusetts named Elizabeth Warren was a Democrat, one of the few Democrats who has spoken out against the tech companies

16:49and it was called for their breakup. Well, we learned from our monitoring system, which I'll describe to you shortly if you like, but we learned from our monitoring system to our absolute shock. We learned that Google had added Elizabeth Warren to a blacklist. Google was trying to push her out of office, even though normally they're very strong supporters of Democrats, but because she had spoken out against Google, they were trying to take her down. And the simplest way to do it, add her name

17:26to one or more blacklists. That way the algorithm before it shows people some content, it's going to, it's going to do very bad things to poor Elizabeth Warren. It's going to take people to websites that make her look like the devil. I was going to say, it seems to work two ways, doesn't it? One, removing the positive stories, her own, um, uh, narratives that she puts out there so that in Googling that you do not see the positives, but you do see the negatives. And when you trust Google,

18:00like you trust your friend who's giving you answers to fact-based questions in the past, suddenly you're seeing a narrative which seeks, seeks into people. Well, let me, let me be clear on one point that we, cause we haven't gotten to this yet. Uh, you can't do this with everybody. Yeah. Okay. You can do it with the people who are vulnerable, who are undecided, who are trying to make up their minds. Uh, six months before an election, that's a lot of people that could be 40% or more of the electorate. As you get closer to the election,

18:34that, that proportion of course shrinks. Yeah. But even last minute, even the day of the election, sometimes you still have 10, even 20% of, of people who are still not quite sure. Uh, there was one election in, uh, Israel a couple of years ago where researchers were looking into this and they found, I think that more than 10% of the people were still undecided as they walked into the voting booths. Yeah. Uh, so there are a lot of people out there who can be influenced. Now, one thing that's again, very, very unusual about Google is that they know exactly who those people are because

19:09they're the most, uh, aggressive, uh, surveillance company, uh, or, or surveillance entity of any sort, including governments. They're the most aggressive surveillance tool that's ever been created. They're, they're literally tracking every single thing that more than 5 billion people around the world do pretty much every second of the day. Yeah. Uh, not just by. Can sound like a conspiracy theory. When you think about people are using Chrome as their web browser,

19:41Gmail as their email, and Google maps as the way to move places, then Google will know exactly what you searched for and bought exactly who you're contacting and what you're saying and where you've been and where you're going. That's massive. Well, Paul, Paul, you know, as long as you focus on those items, you're going to, you're going to be unintentionally, you're going to be misleading people. Okay. Because sure, we all know about Gmail and Google maps and maybe Google docs and so on. But the fact is that Google is actually monitoring us, all of us and our kids, uh, through about 200

20:19different platforms, most of which people have never heard of. So the, the, the, the, the surveillance is much more extensive than people are aware of. Can you give an example? I can give you many. We could go an hour and I could just keep going. But point is that, uh, for example, a few years ago, Google bought, uh, Fitbit. Why? Because it gave them physiological data about people 24 hours a day. Even more odd. Google bought, I should, I should point out Google buys a new company on average

20:52once a week. So they buy a lot of companies, but they bought, uh, a, in the U S company called Nest. I don't know if it's in the UK. Yes, it is. Oh, they make smart thermostats among other things. And without telling anyone, Google started putting microphones into, into some Nest devices. And then some, some, some wise guy like me, some professor actually figured out that there were these mics inside of the Nest devices and went public with that. Google did not deny it. They

21:25just said, Oh yes, we did put them in there, but we're not using them. So why would they put them in? Yeah. And why, by the way, would they not only put them in, why at the same time would they have applied for and been granted multiple patents on, uh, novel ways of analyzing, uh, sounds within the home? So for example, if you're, if you have very sensitive microphones in someone's house, you can figure out whether the sex life is good or bad, whether the relationships are good or bad,

22:00whether the kids are brushing their teeth enough and you can monetize all of that kind of information. So of course, when they put in microphones, of course they're, they're listening, they're getting more data. So we're, um, the point is this is happening, uh, over about 200 different platforms, not, not just the three or four people are aware of. Yeah. Uh, so when you go to almost any website in the world, for example, those websites are almost always using Google

22:34analytics to track the traffic to their websites. Why are they using Google analytics? First of all, because it works really well, like most Google products do. There are exceptions, but second, it's free. It's free. Everything is free. Supposedly. I wrote a piece about this called free isn't freedom. And which I explained that there is no free, uh, at least not with a company like Google. The point is it's free. So people use Google analytics and under Google's terms of service,

23:07which I, I have read many times and I read all their updates under their terms of service. If you are using a Google, any kind of Google software, a Google service, whether you know it or not, it doesn't matter. Google has the right to track you. So there are millions of websites where Google is tracking every single thing you do because those websites have Google analytics running in the background. Yeah. So they're gathering this data and then it goes back to your point of,

23:40it doesn't affect everyone. Then they can pick, are you saying they can pick who they put things in front of? Oh, absolutely. In fact, they, they, uh, they positively do that because once we starting around 26, uh, 2016, once we began to figure out how to monitor their content, uh, then we could see for sure what they're doing, what they're showing people, what techniques they're using. So in 2016, uh, we set up a very small system. We recruited, uh, 95, uh, people whom we called

24:18field agents. Uh, they had to sign NDAs. We had to vet them. We had to create very special software, install it on their computers and this. So, and now with their permission, we are from that point on looking over their shoulders. Okay. And seeing what content Google and other tech companies are sending them. And then we're grabbing that content. Uh, we do this without violating anyone's privacy, by the way, we don't, when we transmit the data, there's no identifying information.

24:49And now we accumulate, we aggregate the data coming in, in this case with 95 field agents. Uh, and we also, if, if they're sending people to web pages, we grab those web pages as well. So we could see whether there was bias in the content that they were sending people, uh, in the weeks leading up to the 2016 presidential election in the U S well, it turns out, yeah, they were, they had a very strong, uh, political bias favoring Hillary Clinton. Uh, we knew from our

25:20experiments on seam, uh, search engine manipulation that with that level of bias going to people nationwide, that would have shifted between 2.6 and 10.4 million votes to Hillary Clinton with no one having the slightest idea. Uh, because as I was saying a little bit earlier, we did learn that we could mask, uh, this technique. We could mask it just by mixing things up a little bit. So if you're in the candidate A group, the pro Hillary Clinton group, it means we go pro Clinton, pro Clinton,

25:56pro Clinton, pro Trump, pro Clinton, pro Clinton, pro Clinton. So we just mix things up a little bit and that dramatically reduces people, people's awareness of bias just by mixing things up just by doing a little masking. Yeah. So now at this point we had a small monitoring system, 2018, a larger one, 2020, a much larger one. Uh, we, and then finally 2023, I decided the time had come to

26:28make a permanent system that runs 24 hours a day. And so right this second, we are monitoring content coming to registered voters and to children 24 hours a day through the computers of more than 16,000 registered voters in all 50 U.S. states. Uh, massive amounts of data, terabytes of data. And we have also learned how to analyze the data in real time, not all of it, but how to analyze

26:58key data in real time. And if people want to see this in motion, they can go to our public dashboard. That's at americasdigitalshield.com, americasdigitalshield.com. And you'll see the data coming in. We're past 121 million ephemeral experiences preserved. In fact, they're not only ephemeral experiences, which I can say more about, but they're personalized ephemeral experiences. In other words, we're grabbing the real content that Google and to a lesser extent other companies are using to

27:33manipulate people and we're preserving it permanently. This is, this is a very serious threat to Google very serious because, and do they know that you're doing it? Oh, of course we, we, we make them aware. We, we, we do everything possible to make them aware. And, and as a result, we've, we've been able to have quite an impact on the content that they're showing people. Okay. I will get to that because I am very interested in that bit. You mentioned ephemeral experiences and that's, that's

28:07crucial, isn't it? To what you're recording and why you're recording it. Could you expand on it? Yes. Well, we have to take a look at this word ephemeral. We also have to take a look at a different, another word, which is personalized. Those are both very critical terms here and trying and understanding what's going on. Because what's going on, by the way, is it's obscene. It's, I mean, it's, it's, it's a, it's a, it's a, it's a tragedy for humanity. The fact that this is going on on such a large scale and that people are unaware, UK, for example, has been very,

28:41very aggressive in taking on big tech companies. The EU, the European Commission in particular, has been even more aggressive and they've levied huge fines and so on, but they have no monitoring system. You don't have a monitoring system in the UK. There's no monitoring system anywhere in the EU. So they, they can't track compliance with all these rules and laws they're passing. They have no idea. And just so if any listeners aren't aware, in 2017, the European Commission fined Google

29:15$2.7 billion for filtering and ordering search results as a way that favored its own products and services. And that hearing was upheld in 2023 or 24, can't remember. But yeah, so there is definite evidence and proof that Google have been manipulating things. Oh, there's, there's, there's plenty, plenty more. Yeah. Let's, let's get back to these two very important terms that people need to know about. One is ephemeral. Ephemeral means, uh, temporary fleeting, uh, on, on the internet. That means

29:52that you do something like type in a search term and then they flash search suggestions at you, but they, they flash them really fast and then they disappear and then they may impact you. I can tell you about the fact that exactly how they influence you and then they disappear and they're gone forever. There's no record cap. Search results are ephemeral. News feeds are ephemeral. Um, really, uh, most of the really important, uh, content that we run into on the internet that is intended to manipulate us is

30:23ephemeral. So, uh, do they really think this way? Well, 2018, there was a leak of emails from Google to the wall street journal, uh, in which the reporter quoted a bunch of emails, which basically said, so these are Google employees now chatting. How can we use ephemeral experiences? Just when I saw that my, my head just spun, how can we use ephemeral experiences to change people's views about Trump's

30:58travel ban? Now this is coming right out of Google. Now I had been studying ephemeral experiences at that point since 2013. This was 2018 that those emails leaked. And I thought, this is incredible. They actually use that terminology within the company. And of course, they're going to try to use ephemeral experiences for their manipulations because they leave no paper trail. Normally they disappear forever. So that when we started setting up monitoring systems, we, we accomplished the impossible. We started

31:36learning to learn how to preserve ephemeral content. And now we're, we're, if you get it, if you go to americasdigitalshield.com, you'll see we're past having, uh, preserved 121 million of these ephemeral experiences and they're, they're preserved and they're protected. And we, we know how to analyze them now very, very quickly. And we're seeing the manipulations that are occurring in real time, which brings me to the second term, which is maybe even more scary than the word ephemeral.

32:11And that's personalized. We're not just capturing ephemeral experiences. We're capturing ephemeral experiences on the computers of real, real voters and real kids. It's as if we're looking over their shoulders with their permission, of course, and we're preserved, we're preserving, protecting their privacy. The point is we're, we're pulling real content that's being sent to real people and Google knows who those people are. Ah, now some people are aware of and comfortable with the fact that a lot of the content

32:46that you get from companies like Google and Facebook, uh, is personalized, meaning it's, it's customized just for you. So my wife who, who passed away a few years ago, unfortunately, she loved the fact that her, her, her Google results were personalized because she said they're like my personal shopper. I've heard the same. I've heard people say, why would I want adverts, which are random for things that I

33:16don't want? I'd rather have adverts that I've maybe mentioned or I'm on my mind at the moment. And so they see it as a positive thing. You don't do. No, not at all. Because it's not just advertisements. It's also answers to questions about anything. The answers you're getting are personalized. And here's the problem. One of our most, uh, recent series of experiments, which was just published in a peer reviewed journal. In fact, if people want to see a technical study, they can go to, uh, digital

33:49personalization effect.com, digital personalization effect.com. And in those experiments, we, you could, you could say we, people are getting assigned, generally randomly assigned to one of two groups in one group, they're getting bias content, like they get in many of our experiments going back years. But in the other group, they're getting exactly the same bias content. And that's the key to this experiment. It's exactly the same bias content, but it's coming. It appears it's coming

34:22from news sources that they trust. And what happens here, which again, blew my mind. And this has happened over and over again, is that the impact of the bias content, uh, overall tripled, tripled. Well, again, this is not an increase of 2%. We're talking about massive, massive difference, which told me that many, many, many of the studies that we've done over the years really have been

34:53underestimating the power that Google and to a lesser extent other companies have to manipulate people online because we weren't using personalized content. We were just using bias content. But when you personalize people pay very close attention and they're much more likely to believe the content that you're sending them. So just to make that concrete, I think in the listeners' minds, it would be that you Google, maybe you Google, who should I vote for in the upcoming election, be it UK, US,

35:27or what have you. And not only are the results that you're getting swayed one way or the other, they're coming from sources that you personally have trusted in the past, maybe in the UK, the BBC, the Guardian, the Times, rather than a blog you've never heard of. The messenger effect is so powerful that you're saying that the results you've got are even more powerful than you've ever known. It's not just that, because it's, it's not just news sources that can be used to personalize

35:59content. If you've been using the internet for 20 years, Google has the equivalent of more than 3 million pages of information about you. So personalization for Google can mean anything. It can mean, because they know so much about you. They literally know more about you than you know about yourself. And so they can personalize content in very subtle ways that, again, you're just

36:32completely unaware of. And in that, in those experiments that I just mentioned, we're only doing one kind of personalization. We're just personalizing the news source. But imagine having that much information about someone, and in subtle and not so subtle ways, personalizing content so that you are highly, highly likely to be influenced by what they're sending you. By the way, none of this stuff is in my imagination, okay? Because this has been confirmed in recent years, not in the

37:08beginning when I first started, but in recent years, the things I've been talking about have been confirmed in leaks from Google and Facebook and Twitter and other companies. They've been confirmed by whistleblowers, people who have gotten so fed up with the company and their dirty tricks that they've quit the company or they've been fired and they've gone public. Every single thing that I'm telling you about, there are people who know that this is the truth.

37:42In one of your papers, you quote, I think it's Tristan Harris, who was former design ethicist at Google, which is scary in the fact that they had one, and now he's not that, that they don't have one anymore. So that itself is scary. But he said, the tech industry is creating the largest political actor in the world, influencing a billion people's attention and thoughts every day. Now, that is somebody within Google said that. Yes. And at that time, the number was a billion, but now it's over 5 billion.

38:14Okay. Yeah. Okay. So we've seen and heard your thoughts, your experiments, and we started to talk about the monitoring system as well. But you've listed quite a few detailed techniques. And I'd quite like to get into the nitty gritty, just to go through them to show people how this is working. Now, the first one you've mentioned a few times already is sort of the lead one, which is the search engine manipulation effect. The order of the search of results will change people's behavior. And as you

38:44said, in psychology, we've been studying serial position effects for decades, really. And an example which I think resonates with people quite a lot is that experiments have been done when they change the order of wine on a wine list from the least expensive to the most expensive, and then swap it around. So the most expenses at the top, people on average spend more on wine simply because the order of the list, it's that simple. And I, and I use that as an example, because it's so simple to think about. And we've all been in a restaurant looking at the wine list, having no idea what any of

39:18these wines are, apart from, you know, they're red, white, and maybe from France or from California or something. And so we rely on something, a rule of thumb, which is, well, I'll, I'll pick not the top one, not the bottom one, I'll go maybe a third down. And that's an order effect. Now, search engine manipulation, as you say, you can start to see that working in that way. But that isn't the only thing you found, is it? You've also found something called the search suggestion effect. Can you briefly talk about that? Sure. When I first testified before the US Congress, and I was trying to explain

39:54search suggestion effect, SSE, I asked Senator Cruz, I asked him to please, I said, do you have your your cell phone with you? And he said, yes. I kind of wondering what, why I would ask such a silly question. I said, would you mind taking it out and just going to google.com? And by the way, I, I said to him, I hope you never go to google.com again. I hope this is the very last time that you do this.

40:26And people laughed. And I said, and just up in the search box, just type the letter A. And if you, if it's not too embarrassing, if there's no porn sites or anything, I said, just tell us what you, what it's showing you on search suggestions. And he, he read out the five search suggestions and four of them were for Amazon. Google is showing you and everyone else Amazon when you type the letter A, because Amazon is one of Google's largest advertisers. It's either number one

41:02or two. And also because Google is Amazon's single largest source of traffic. In other words, these are, they're not just business partners. These are bosom buddies. Okay. So Google is, is manipulating you with search suggestions from the very first character that you type. And with each successive character, you know, of course that, that limits them a little bit. They're still manipulating you. And so we've done experiments showing unbelievable that just by manipulating search suggestions,

41:38at least in, in the, in laboratory kind of situation, we can turn a 50, 50 split among undecided voters into a 90, 10 split with no one having the slightest idea that they have been manipulated in this case by search suggestions. Wow. Yes. And so how does that work? Is it to the fact that I might be searching for something innocuous? I might be searching for, you know, Humpty Dumpty, what's the rhyme? So I put H in and then straight away it comes up Hillary

42:11Clinton is a crook. Yeah. Well, that's the point is that when you start to type, you are getting distracted by these search suggestions. You might continue to type your, your entire search term and hit enter, you might ignore the search suggestion. Of course. But it turns out that a fair chunk of the time, people end up clicking on one of their suggestions, which is what they want. They want you to click on one of their suggestions because then they're in control. Once you click on a suggestion,

42:46that is going to give, that's going to lead them to give you a particular set of search results. They're in control. And of course they control the filtering that is the selection of search results and the ordering they're in control. So they're, they're, they're trying to get you to whatever it is they want to get you to for whatever reason. And it could be, uh, to make money or it could be to shift votes or it could be just believe it or not to impact people's values. One of the scariest leaks

43:16from Google, uh, is an eight minute video, which is called, has a funny name, the selfish ledger. If you type in my name online, Robert Epstein, and then you type in the, the selfish ledger. Uh, and if you're not using google.com, you're using a, a sensible search engine, such as brave search. And you look for the selfish ledger and put my name next to it. You might get lucky. And you might find my annotated transcript of this eight minute film. What is this film about? It is about

43:50Google's ability to re-engineer humanity. It was, it was made by their advanced products division. It was never meant to be seen outside the company, but it leaked. And it's very clear about re they say, re-sequencing human behavior, according to it's right there. It's right there in the narration, according to Google's values. Yeah. They're, they're not kidding. They, I mean,

44:23they think they know better. They think they know how everyone in the world should live. Yeah. And they have very powerful tools to make that happen. People will be thinking of, you know, George Orwell's books, 1984, mind control. I'm thinking, you know, Skinner wrote Walden too, as his view of how behavior techniques can be used for good and for child rearing development and creating a nation of kind, caring people. It's almost like, it's almost like a Bond villain has read that book and said,

44:59well, I love the techniques. I'm just going to use it for my own nefarious means. Is that fair? I think you're being too kind because the problem with Google is that they,

45:13everything they do is secret. See, I, I, in Walden too, the, everyone knew that there were some psychologists who were trying to, trying to make everyone happier and more productive and more creative. Everyone knew that. And they knew who the psychologists were and they rotated people through different positions. Everything was out in the open, but with Google, everything is secret. It's hidden. And also remember Walden too, is a community of a couple hundred people, all of whom knew each other. And the impact was kind of restricted to that little, little commune,

45:49if you want to call it that Google is operating on a different scale. Yeah. We're talking about most of the people in the world, 24 hours a day, having, having more information, more surveillance than any dictator has ever had in history and having more power to shift people's thinking, behavior, attitudes, beliefs, votes, purchases than anyone has ever had. There's just, there's nothing that's ever come close in history. I mean, again, something that I realized only

46:26recently. Let's say that, that the actual personnel at Google were, were not interested in making the world a better place by their standards. Let's say they were not interested. And then let's say they're not interested in getting mixed up in elections. Let's just say they're interested in making a darn good search engine, period. And they leave their algorithms alone. They don't mess with them. They don't, they don't add blacklists and whitelists. And the search algorithm just does its thing day after day

47:00after day. You know what? It would still be a terrible threat to humanity because it has no equal time rules built in. It would still be shifting the thinking and behavior of billions of people every day without their awareness. It's true. In this case, there are no human, you know, overlords, you know, telling it how to shift people's thinking and behavior, but it would still be occurring. It's a, it's a machine that is built to do exactly that. And I don't, I'm sure I'm positive because I know some of

47:38these people that when they started out, they never had that in mind, but that's what the search engine is. It is the most powerful mind control machine ever built. And we've been picking on Google so far and rightly so. I'm not saying we shouldn't have, but it's not that just, is it? And the next effect is the targeted message effect. And you've looked at that and with Facebook in particular, haven't you? And what that is. So could you just explain what the targeted

48:09message effect is and how Facebook has been using that? I can. And as a bonus, I can tell you about a brand new experiment we just completed on our Facebook simulator, which I've never spoken about publicly yet. We haven't submitted it yet for publication. We will be presenting it at a scientific conference in May, but it's brand new. Absolutely. Great. Yeah. Exclusive. So here's what we have found in the past. And then I'll tell you what we just found. Well, in the past, we found that on a platform like Facebook, if certain messages

48:45are inserted in just the right way, and if they're sent to some people and not others, then you have a tremendous impact. Again, especially if you were looking for undecided people, you can have a tremendous impact on just that group. Now, that group could be, let's say, Democrats or Republicans or conservatives, or it could be any kind of group. If you target your message to one group, then you can have an enormous impact on that group. And no one knows. That's the interesting thing.

49:22But the people who don't get it don't know that they should have ever had it. How could you know? Yes. You know, you're getting a message. Your neighbor's not getting a message. What are you getting together with your neighbor every day and comparing messages? Yeah. What messages did you get on Facebook today? Yeah. No, we just don't do that, do we? But our monitoring system, our nationwide monitoring system is preserving all that content. We know when they're targeting. Can you give a concrete example of that effect using Facebook? Well, I have a better example than Facebook because that is, in my opinion, very, very startling and

49:59very consequential when it came to this past election and Trump's win. As election day was approaching, we were getting better and better and better and faster and faster at looking at announcements that Google was putting on its home page. Now, in the U.S., Google's home page, which is normally kind of blank, right? Now and then they put a message there or some cool art thing. Yeah, the Google doodles. The doodles, exactly. And so we noticed that Google was sending out

50:35go vote reminders, mainly to Democrats. This is highly consequential, what happened here, because they were sending out 50 percent more go vote reminders to Democrats on their home page, which is seen 500 million times a day in the United States than to Republicans. So that can shift a lot of votes because there are people who just sit around saying, I'm going to go vote now. And

51:06then they just don't. But if you keep reminding them over and over again, and we know this from rock solid research that was actually done by Facebook employees working with my colleagues at the University of California, San Diego, published in Nature, which is a British highly respected journal. Probably the most respected scientific journal in the world. And they showed that when you send out go vote reminders, that gets a lot of people to vote who otherwise would not. So here we are

51:37documenting the fact on a very large scale in all 50 states that Google is stacking the deck. Google is sending out those vote reminders mainly to Democrats. Now, why wouldn't they do it exclusively to Democrats? Well, that's masking. That's what masking is. It confuses people so that they can't see what's going on. So October 31st, which was, let's see, the election was a few days after, I think it was on

52:07the 4th or the 5th of September, November rather, sorry. On October 31st, where the numbers were so clear and so, so scary, we decided to go public and to let Google know and to let members of Congress know. And that night, we let everybody know that this was happening. Now, meanwhile, we're still collecting data. The system's still running. It's running right now, 24 hours a day. And the next day, they stopped. So literally for three days before the election, plus on election day, they stopped. They stopped.

52:46Yeah. It was almost 50-50. Those go vote reminders were going out from that point on, almost 50-50 from the time that we exposed this manipulation, they stopped. Now, what that means is that they, perhaps fearing retribution, fearing the law, maybe, because this is technically illegal under our Federal Election Commission law, what they were doing, sending out those kinds of partisan go vote reminders. So they stopped. Now, when they stopped, well, think about what they're

53:22doing. They had been shifting lots and lots of votes to Kamala Harris, and then they stopped that manipulation. We got them to stop two others as well. But the point is, this is a big manipulation, and they stopped. So what does that mean? That means that instead of sending millions of extra votes to Harris, they were willing to let Trump win. And by this time, the CEO of Google had already called up

53:52Trump and wished him well. And then they, when Trump won, they immediately donated a million dollars to Trump's inauguration. And they showed up for it and sat in the front row. In other words, they did an about face, not just because of the monitoring, but there were other factors, too. But the point is, our work, our research, and our system, combined with other factors, got Google to back off.

54:27To be crystal clear, are you saying that with your monitoring system last year, you changed Google's behavior? That then changed the flow and the result, possibly, of the U.S. election? Yes. Now, I hate saying that because, of course, I was never, well, I should say, of course, but I was never a Trump supporter. I don't think he's a good president, and I don't think he has the temperament or the abilities, the cognitive abilities to be president. And so I've always been very clear on

55:02that. I'm not a Trump supporter. What I'm trying to do is put our country and our system of government ahead of any particular candidate or party. And I think everyone should do that. I mean, the system we have is meaningless. If we're allowing the system to be run by private companies, it's meaningless. There is no democracy. I get that. And it's a very laudable position to hold. The balance is more important. However,

55:36I'm just going to be devil's advocate here because it's interesting that the emotion that's attached to the presidential election and Trump in particular, more than anything else, is very high. We now have, I say we now have, we now have a president of the United States who is a self-confessed sex offender, a multiple felon, twice been impeached. He's now pushing an aggressive policy of discrimination and hatred, especially of foreign nationals. Is there an argument to say that Google

56:09and Facebook were actually serving a purpose to rally the liberal minded voters to stop this from happening? And you called it, you know, to make a world better by their standards. And whilst it isn't a balance, what my mind goes back to is when I did philosophy in my undergrad, Plato wrote a book called Gorgias. And in the play, a chap called Gorgias comes to town who's a skilled orator and rhetorician. And his

56:40argument against Socrates in the play is that he can persuade the forum to give him the contract for boat building, because he can stand up and talk, whereas the actual boat builder with the skills cannot do that. And therefore, oratory and rhetoric is the best thing to be. And in the play, Socrates, as he does, talks around the subject and proves that that isn't the just way. However, I think a lot of people will think now

57:11that resonates, that the orator is now somebody who gets attention. And it's almost that when you're arguing in politics, and you stand up and say, emotional lies, that you want people to just to vote you in, as opposed to realistic truths. If somebody stands up and says, we do need to raise the taxes, because we do need to pay for things. And you know, it's going to be a hard four years ahead of us. That's not going to get people to vote for you, even though that may be the best medicine for

57:46the country. If somebody stands up and says, I'm going to lower taxes, make the economy great again, bring everything back into this country and give people jobs again, however empty those promises are, it's more likely that people are going to do that. So politicians making those false claims and lying in order to gain power, isn't true balance anymore, when the actual other argument is actually more realistic, and people are not going to vote for that. So could you argue, and it's philosophical

58:18more than anything else, that Google and Facebook were doing a service to try and balance against that sway of people who will be, like you say, the floating voters, who will go towards these empty emotional promises, rather than the realistic politics? Of course, of course, of course, you can make that argument. You did it very well.

58:44I still will do keep, I'll still keep doing what I'm doing, though, because the, you know, we, we, we all get caught up in short term kinds of concerns. I mean, we all do, every one of us. It's, it's, it's irresistible. You know, what's going to happen in the next five minutes is that's all that matters. And the fact is, that's, when it comes to our lives, that might actually be true. But when it comes to a system that's in place, and it's been in place for several hundred years, and

59:15that works pretty well, I mean, a democracy has its flaws, but I, I personally can't come up with a better system of government. So I think a system, when you talk about systems, you, you really do have to hold on to your ideals. I, I think you have to hold the line. You have to say, wait, this system has so much value, not just for us right this minute, but for maybe for humankind, that it's worth

59:50preserving. And that when we see threats to this system, we should fight. And that's what my team and I are doing, because what we are doing, we believe is very important. We're, we're not trying to impose our way of life on anybody, but we are trying to preserve a rather unique system of government. I, I learned over the course of doing this research that our founding fathers, George

1:00:20Washington, John Adams, Thomas Jefferson, and others, they were vehemently, vehemently opposed to having a party system here. Uh, John Adams said, a, the two-party system is the greatest possible threat to our constitution. Uh, Thomas Jefferson, who was a kind of a, a joker, uh, of, of a man said, if the only way I could get to heaven would be to join a party, I wouldn't go. My own friends and family

1:00:52members have kind of distanced themselves from me because they think I'm trying to help Republicans and conservatives and Trump. And I'm not, I'm not, I'm trying to help a system. And, and I can tell you this, that without a monitoring system in place, if you, if we allow companies like Google to exist and just do all the things they're doing, there is no democracy. Democracy becomes meaningless. Yeah. As they are determining who wins. I think we've talked about the historical side, you know, the founding fathers. I'm going to ask

1:01:27you to get your crystal ball out or your divining rods or whatever technique you use to predict the future. Cause we now find that tech owners are lining up behind Donald Trump. You potentially or probably fearing what he will do to hamper their businesses if they don't. Right. With Trump's proven track record of manipulating American public, you know, such as, um, now controlling which journalists are allowed into the white house press pool, which has never happened before, which is, you know,

1:01:59obviously manipulating, um, you know, who he wants in front to ask the questions or influencing Jeff Bezos to stipulate what the Washington post can or cannot publish about Trump and his government. Do you think the search engine marketing effect will swing from pushing that liberal side to pushing more right wing views? Well, we're, we're watching very carefully and we are seeing, uh, some shifting in that direction, but so far overall, Google is still doing what it does. In other words,

1:02:35it's still favoring, uh, the left, but things are a little bit complicated right now because Trump, God, I hate the idea that he, he, he, he's in charge of our nuclear weapons, but when it comes to, when it comes to tech, he is literally holding the sword of Damocles in his hand and it's poised over the heads of these executives at these, because last summer in the case called the

1:03:05Department of Justice versus Google, uh, a federal court ruled, this is the first time that's ever happened in this country ruled that Google is a monopoly. And that was August. Now, August, September, October, November, there's the election. Since, since August, that case has still been in its penalty phase. And this is, this is one of the factors I think that led to Trump's win. It led to, that led to

1:03:39Google backing off because there's still, to this day, there is still in the penalty phase of that trial. So the, the verdict has been rendered, but they're still in the, which means that the, the, the, the court is working now with members of the Department of Justice to try to figure out what the penalty should be. And of course, one thing they're considering is breakup. So, you know, the folks at Google are in fact cozying up to the administration in many, many ways.

1:04:15Because they're in the penalty phase of that trial. Now, could they really be broken up? No. And they know that. And I could talk about that issue at length, but the point is that they could be forced to sell off some of the companies they've bought. Like Google Chrome. Well, Chrome is the number one candidate actually. But if they were forced to sell off Chrome, you know, they'd get $500 billion in cash for it and wish they could be used for whatever nefarious purposes they like, but also they

1:04:49would still have access to all the Chrome data. It turns out. So they'd be thrilled if that's what they were forced to do. And at the moment, a fat wallet, and they wouldn't have really lost anything. They would have lost no data at all because whoever bought Chrome would still have to use Google, what Google calls their quarantine list, to check on the safety of a website before Chrome takes anyone to the website. So Google's getting all that information. Google also would still be

1:05:20the default search engine on Chrome. Okay. Right now, Google pays Apple $20 billion a year to be the default search engine on Apple's Safari, which is their browser. Where did that $20 billion come from? And where did we learn that? We learned it from that trial that I just mentioned, Department of Justice versus Google. So if they're forced to sell off Chrome, that would be a huge win for them. And they know that the government, no government can ever break up the search engine

1:05:55itself. And that's their biggest weapon. And you can't break up the search engine because it won't work. Yeah. Same with Facebook. You can't break up the social media platform because that would be like putting the Berlin Wall, you know, through the houses of billions of people around the world. You can't do that. No. So they're not really scared of breakups, but they are, they're a little nervous now and they are cozying up to the Trump administration. But there's still, if you look at americasdigitalshield.com, you'll see they're still favoring the left. So they're still pretty

1:06:32much doing what they've been doing for a long time. And obviously the digital monitoring system, America's Digital Shield, will continue to monitor that and see if it does go the other way if Trump's administration maybe puts more pressure on them. Well, we can monitor anything. And now we're starting to look at the possibility that they've been manipulating financial markets. We know they've been indoctrinating children. We're trying to find ways to quantify that. So, you know, we're gathering so much data that in fact we will,

1:07:04we're providing a kind of resource for researchers of all sorts and government officials that they'll be able to dig into and look back in time and see what these companies were doing. We're, we're collecting data from, not just from Google, Google, but from seven, seven platforms, one of which is YouTube. And again, what's happening there is very interesting. We, we got them to back off on political bias on YouTube by 50% between August and November of last year. So YouTube's recommendations

1:07:39are often highly politically biased, but because of our monitoring, we got them to back off considerably on that level of bias. So monitoring, I have just become a real believer in monitoring whether, whether I'm involved in, in, in those projects, you know, as they're set up one by one or not, it must exist. And we have been contacted by the way, from people, from officials from eight other countries, including in the EU, and they want us to help them set up monitoring systems.

1:08:12That was my question. Yeah. Is it going to go beyond America? Well, of course, people can do that. It can do it on their own. We've learned so much that I, I, I hope we can help people do it well because we, it's been very, very long journey, very difficult. And we talked about, so the monitoring system, it sounds absolutely crucial that going forward, good. You've gone through a huge amount of work, getting awareness of people to know that this is happening. And awareness is always that first step, isn't it? In any change of behavior, they need to be aware that it's happening. Now you've got the America's digital shield, the monitoring system

1:08:47to police the people who were unpoliced before the search engines, the Facebooks and the YouTubes and so on and so forth. What about at a personal level? What can somebody who's listening to this and goes, Oh, I, I am now aware. What can I do on a personal level to help protect me and my privacy and from this manipulation?

1:09:10I have two very concrete suggestions. One is to go to America's digital shield.com America's with an S and there's a, there's a little box there where you can say sponsor a field agent and you can sign up. And we've had thousands of people do this, not enough, but the numbers are growing and you can, you can pay $25 a month off your credit card and that will sponsor one field agent. So that takes burden off of us running the system for this to be permanent and self-sustaining. We're going to need

1:09:46a lot of people to, to sign up voluntarily to sponsor field agents. You're probably thinking, well, why don't we just ask for volunteer field agents? Can't do that. Can't do that because you're going to be something to everyone. Yeah. Well, because Google would send us people. Yeah. Google has done this repeatedly. We know we've caught them and they not only have a hundred thousand employees, they have 120,000 outside consultants. It would mean we get useless data because they'd be sending, uh, you know, uh, basically, uh, laundered, laundered data. Number two,

1:10:20this is probably as important. Go to myprivacytips.com and you'll come to an article of mine. Again, it's myprivacytips.com and you can learn how to protect yourself and your family so that you can use all the good tech that's out there, but without being surveilled and without being manipulated, you can cut back on those terrible threats and, and you learn how to use new technology in a smarter

1:10:53way. Would you be as bold as to say somebody with a Google phone running Android using Chrome, Gmail, maps, got a Fitbit on their arm that they need to have a rethink if they want to avoid being manipulated in this way? No, I think they need brain surgery. You always go a step further than me. I like that. Well, the point is you have to be seriously, if you really understood what's going on here, uh, what information is being collected and how the information is being used. I, I, I, I think

1:11:28a lot of people would run into the forest and, and really try to get off the grid because what's happening is, should not be allowed. It should be illegal. And, and the government's not- And that's why I find it fascinating because it does come across sometimes like a film plot. Like, I think it was one of the James Bond films that Elliot Carver was in, in control of the press and how much he could manipulate. Now that's a fictional narrative, but we're reading in your papers

1:12:00and in the articles that you're writing, the type of effects like the search engine manipulation effect, the search suggestion effect, and ones we've not covered yet, you know, or yet I won't cover them, but we'll put a link to them. The answer bot effect and how it's moving into digital assistance and potentially expanding into AI as that's ever marching towards our lives. Once you read all of this and you feel how real it is from the work that you've done and the data that you've presented, you can't help but think, I do need to have a little rethink about what I'm

1:12:33using, how I'm using it, and then who's listening and using that at the other end. At the moment, again, the awareness, you, you're, you spoke about the importance of awareness. The moment is, there's just not enough awareness. There's, there's a complacency. I, I, I, I had a meltdown on Joe Rogan's show. I saw that, yes. And the fact is, that's where, that is where we are now, where we're kind of, we're stuck. Our heads are in the, we're deep in the sand and we, we don't really

1:13:07know. So, so. The work you're doing is waking people up. Being on Joe Rogan, talking to me and many, many others and writing the articles that you write and sharing the journals is waking people up. And that's why I wanted almost to finish on what can we do at the individual level. Monitoring level, yes, it's fantastic that you set that up and other countries are interesting on taking that on. At a government level, I suppose, you, you talked about to make their index public commons as well.

1:13:38Yes. Or allowing other companies to create searches and create a competitive marketplace. You've talked about not using Google as a browser, but use Brave Search instead because it looks after your privacy and therefore you can be sure that actually what you're searching for is what you're going to get back rather than what the company wants you to see. There's lots of little ways. And I think once people have woken up, they can start to put these into place and then tell their friends and then slowly but surely things can change. You've also written a book which collates, I think,

1:14:13all of the work that you've done over the last decade to 15 years on this subject. When's that out? Well, I, I, I have no idea. I, I, there it is. Here is a, is a kind of preprint of it and it's called, um, preventing the misuse of digital influence. But the, the sad truth is I don't know when it'll come out or whether it would ever come out because I've published, before I started studying Google, I published 15 books. One of which was cited by, uh, our, our, our Supreme Court in the United States.

1:14:48Since I've started studying Google, I've not been able to get a single book contract because the companies are terrified of Google because Google can not only suppress a book, that's pretty easy. Just, just demote it or delete it from search results and it's gone. But they can do that for all the books that a publisher publishes and puts the publisher out of business. And a, a, uh, the man who's head of Europe's largest publishing conglomerate actually wrote a piece, an open letter to the leaders of Google was called fear of Google.

1:15:24Google. And he's talking, he's talking about how Google without even meaning to perhaps now and then shuts down one of, one of his, his publishers or comes close to shutting it down because they, because they make it disappear or they demote it from search results. And so there are a lot of companies that live in constant fear of, of what Google might do to them. And that's made it tough for me to publish any more books.

1:15:55Yep. And I would urge people to keep an eye on your website because you not only link to all of the websites that you've mentioned in this podcast, but you will keep people updated on where they can get that book when it's published. And I guess if it's not published, they'll, you'll get it out to them somehow. It is important. I've read it. I've had a copy. Um, and it is important. Um, and it's very clearly written and it's very well articulated. The three things that need to happen. And so I do urge people who are interested in, like you say, not only their own privacy, but the privacy of their children going on in the world to have a look at that. Robert, that just leaves me to say, thank you so much for your time. It's been mind boggling. And I hope people are listening and haven't gone away. No, I'm going to be bold like you. I hope they're scared. And then they go and do something because of the awareness that you've raised with that. Thank you very much, Robert.

1:16:48Uh, thank you, Paul. Uh, again, to the extent that there is any awareness, you're the one who's making that happen. And I'm extremely grateful for that. It's a pleasure. Bye bye. Thanks. Bye bye. Here's a little extra time with Robert, who after our main chat mentioned a new study about Facebook manipulation, which is yet to be published, and gave his view on how Google influenced the vote on the UK referendum. Well, uh, I didn't have a chance to say it, but we, you know, we really have just finished another, uh, shocking Facebook experiment.

1:17:22Go on, tell me. Well, we've, uh, we don't really understand the results yet, but we basically by, by throwing in some, some occasional, uh, notices into people's feeds. This is a perfect Facebook simulator. Uh, we've not only raised awareness of an upcoming election, but we actually dramatically influence, uh, change the way people are going to vote, which we're not even trying to do.

1:17:58So we don't even understand that, but we're talking about a, uh, 51% increase in people saying, I'm going to vote in favor. And we don't, we don't know why. Uh, so there's a lot there. We're about, we're about to start a second experiment in which we're going to explore this further. But we, we're learning that social media platforms, uh, can have a tremendous effect on people with, again, without their understanding what's happening, because it's, these messages are just kind of thrown in to the, or among the, the organic items in the feed.

1:18:36So that's something we're trying to figure out now. Um, in the UK, we had, you know, the, the referendum for, uh, should we leave the EU or not, uh, in 2016. And the team who were saying we should leave were very emotional. Things on the side of buses saying how much money we spend on the EU. And it will all come back to the NHS. And that basically by being the EU, we are taking money away from our nurses and our doctors and our healthcare system.

1:19:06And they were very, very passionate, screaming down in political broadcasts at people watching. Whereas the people arguing that we should stay were economists a lot of the time. And they were being very rational and calm about their argument. What happened? Everyone voted for to leave the EU. Say everyone. It was 48-52. So not everyone, but it swung it. Well, uh, I can tell, I can give you a couple of details here that might surprise you. I was contacted by a reporter from the Times of London a couple of weeks before that vote, the referendum.

1:19:43And, uh, he, uh, he had read about my work and he was kind of curious whether I thought it was relevant to the Brexit vote. And I said, oh yeah, absolutely. He said, well, how so? I said, well, Google needs to destabilize the, uh, the EU because the, the European Commission is the most aggressive agency in the world. That's trying to, trying to mess with Google. And I said, so they want, they're, they, they're, they're going to want Brexit to happen.

1:20:14So they're going to shift as many votes as they can. He goes, whoa, whoa, what are you talking about? So of course I'm trying to explain to him. And he said, well, he said that, well, the, the, the, uh, the polling is showing that this is just neck and neck, that just too close to call. I said, oh, well that, I said, that's interesting. I said, well, if you, if you get me some numbers, I will tell you how many votes Google can shift at the last minute. And he, he went right along with it.

1:20:45And I told him all the stuff I needed. He got me all the information and I did some calculations and I told him that, uh, Brexit was going to pass positively. Brexit was going to win and it was going to win by at least 400,000 votes. And he said, that's impossible. And I, he said, I said, no, no, absolutely. I said, because these, these techniques that they use, you know, people can't see them, but they're very, very powerful and they impact the people who are undecided, blah, blah, blah. So Carol Cudwallader from the Guardian, when it came to Brexit and Brexit won, she wrote an article that was kind of crazy saying, I think Google made this happen.

1:21:27He said, I can't prove it because all the data are gone. Now she didn't use the word ephemeral. I can't prove it. And what a shame that we can't go back in time and we'll never know for sure what Google did. Well, you scared me even more now. Thank you. Um, I think I'm going to go off and put a little tinfoil hat on. Thank you. Thank you. Thank you. Thank you.