41:18 Let's just say the Max is true here and the max is true about the illegal immigrants that are brought into the country to vote for the Democrats. If you could sway 10 million votes and have maybe, I don't know another ten million immigrants all voting for Hillary and she still loses? How much how many people really hate this woman or or how much did people decide or how much do people really want to vote for Trump? You could look at it either way So a little background on the professor. He's a PhD and senior research psychologist
41:57 Research scientist, media professional author of 15 books and reading from his own bio. More than 300 articles on psychology related topics including empirical studies in science nature psychological science and the proceedings of the National Academy of Science has a PhD from Harvard University under BF Skinner Dr Epstein is the founder. Yeah do you know him? BF Skinner Oh yeah he's the famous behaviorist okay well one of the most famous guys ever Epstein is the founder and the director emeritus of the Cambridge Center for Behavioral Studies. He's also hosted several radio shows, etc. He is a registered Democrat he voted for Hillary Clinton and he started a non-profit which is The American Institute For Behavioral Research And Technology 501c3 I checked
42:49 990 filings has about a hundred and fifty thousand dollars on hand raised about forty thousand dollars every year for the past couple of years, little bump in 2016. The only people in this organization were three Epstein himself Tyler Healy who is the The technology, in fact if you read his bio you know exactly what it is. Technology director, cybersecurity expert full stack developer so he basically put together all the tests and then Brian Meredith a managing director who came to the AIBRT after a much-honored advertising agency career spanning three continents former vice president of Benton Bowles founding member and director of the international creative team at McCann Ericsson these are big advertising agencies
43:39 He passed away in 2017. I think he was probably the original funder of this organization and clearly not a lot of funding has come in but, of course if you're fighting Google you might have a lot of enemies. The Google has...I think is the largest lobbyist in DC at the moment maybe battling with China who the hell knows? Maybe they work together. So I look at the research and he has a number of research beyond what we're talking about here, which is search engine manipulation effect. The SEME
44:17 He's also looked at the search suggestion effect, the answer bot effect. Anybody has done many other things that come to psychological behavior and I think having an advertising guy in early was probably hey how do we use today's technology to sell products? And that's probably where he stumbled yeah so Probably the most fair research you get because they were looking to figure out what manipulates people towards choices The research as published in the PNAS, which is an official journal not only was it accepted and reviewed. It has been replicated in Germany! This is a big deal if you go look at all these bullcrap studies everywhere Look and see if there was a replication of it The replication crisis is rampant certainly in the psychological sciences
45:06 This you know the people can't recreate these studies and but still they're what are you eating? What are you doing it's it's distracting me. Sorry Are you ripping paper or I am actually I'm just going through some looking for some notes go on don't tear paper I wasn't tearing paper, man. I mean I'm gonna name Mike So don't talk to me named the movie I looked at his research. It is large groups, tens of thousands of people double blind study... you're tearing paper again... I'm not tearing paper! I am just taking paper out of a pile. I am demiking it. Just keep talking. No don't de-mic i need you to be able to interrupt for a reason
45:48 Here's a pile of paper, I take the paper off. What are you looking for something? Are you doing crossword puzzles? Yes! I told you that already go on Is your chair squeaking to your paper and they're talking back to each other? Just do your bit Will you just keep going with it normally you're bitching and moaning my interrupting now you're just bitching and moaning because i'm not interrupting Please please go back All right um So the research is very deep. It is done with all of this, I mean...I'm not a scientist but I've seen a lot of research throughout 11 years of doing the show and it really looks like he did all the business the way it should be done and in 2014-15 but particularly in 2016 he did a number of studies and again in 2018 where he would look at
46:39 the results based upon political questions that came in from Google query, from Bing and from Yahoo. And the results for Google were significantly different I don't...I think that you know their algorithms are different but the research really focused on and this is what I found to be interesting is that there's the bias of what people click on in search results So when you search for, you propose a question to Google the top two results receive all of the clicks with 50% going to the top one 30 percent going to the second one it drops off quite dramatically. Interestingly the last the bottom one on a page gets more
47:30 more clicks than the five or six above it. And you can probably figure out why we've all done that. Let me scroll down to the bottom, I'll click this one and go to the next page but the click is pretty much always on the top one or the top two and depending on what is driving the results, he found that amongst undecided voters this is key people who are really on the fence and it could be you know undecided voters can be 10% can be 15%, can be 20 percent of an election. People on the fence that...that the choices people make I should mention usually over twenty percent
48:13 The undecided voters are always very high right into the election. Oh really? What kind of percentage we talking? I've seen as high as 40 So he claims that given an A-B choice, that the top two links determine people's choice. And regardless of what that content is... well obviously it's pointing towards a or b if its a 20 to 90 percent will choose a over B just because they were the top 2 links so we understand the research because that's really what all his research is saying. His research is saying
48:51 Unlike anything else, when you have a choice between two candidates and you pose the question the top-two link answers that you click on that are on the top of the page will determine who we're going to vote for in aggregate over your research. I should stop you? Mmhmm. Because what he's doing is maybe deconstructing what has already been done at Google, because I will remind myself that Sergey Brin used to come on the Silicon Spin show a lot and he one time said to me you know we have the most PhDs of any company in the world. And what are those PhDs doing there? They're probably trying to figure out well they can do stuff to manipulate things
49:41 Or is that really what they're doing? Because I have some thoughts about where the biases come from. But, I have a bunch of clips most of them are about 50 to 60 seconds long but they do tell the story as from different interviews that i've put together and the first one is Epstein Introducing himself and giving a brief overview of SEME, which is the Search Engine Manipulation Effect. This is what he proved in his research. By the way the whole PDF of this research is in the show notes I've been researching all kinds of new methods of online influence that internet has made possible My first discovery was something called SEME, S-E-M-E The search engine manipulation effect
50:26 which is the impact that biased search results have on opinions and votes. When I first started doing experiments on this, which was more than six years ago, I thought the impact would be very small. It turns out the impact It's enormous, it is one of the largest behavioral effects ever discovered in the behavioral sciences. I published my first report on this effect in the proceedings for the National Academy of Sciences that was in 2015 and that report has since been accessed or downloaded more than 200 thousand times and that's a lot for scientific paper and since then I've discovered about seven other effects
51:09 neither tax are so powerful and that if they're in the hands of people who have uh... particular political leanings altogether they can shift up with a fifteen million votes in a presidential election uh... without anyone knowing that they're being manipulated him about leaving a paper trail for authorities to trace So that's kind of the same background or I gave but in maybe more succinct language so in 2016 he decided to monitor searches This is before the election leading up to the election and to compare if Google Was delivering biased search results to those top two positions versus competitors 2016. I actually set up the first
51:54 ever uh... project to monitor the search results that google being and yahoo were showing users prior to the election when they conducted election related searches uh... and i found The search results were strongly biased in favor of Hillary Clinton, whom I supported by the way. I am not a conservative and so they shifted votes lots of votes away from Donald Trump toward Hillary Clinton but in a way that people couldn't see because the way this works is People trust and click on search results that are higher in the list. So 50% of all clicks go to the top two items in the list. Wow!
52:41 Sure, and what Google was doing was putting items high in the list that led people to webpages that made Hillary Clinton look a lot better than Donald Trump. And over time that shifts the opinions and votes of undecided voters decided voters and of course in close elections it's undecided voters who determine the winner. In this particular case we calculated based on the bias that we found, uh... that Google could quite easily have shifted two to three million votes toward Hillary Clinton just using this manipulation without anyone knowing but they were doing it
53:23 And that the part of without anyone knowing they're doing it is important because this is not just a clear bias that stands out. It has a name. They can shift millions of votes using what they themselves call ephemeral experiences, in other words things like news feeds and and search suggestions uh... in search results answer boxes even ephemeral because they appear only for a second or two the effect your thinking they disappear there are stored anywhere no one can go back in time and retrace them and google employees and you know we've seen leaks recently today hero they're well aware that they can use a femoral experiences to shift votes of opinions and they do this
54:14 Deliberately I've proven it with my monitoring projects so ephemeral Experiences means it's just for an instant. It's just there your search results something a search box Your your auto complete all these things are really not trackable by you even in your mind because it's the way You've done used the Google product for ever since you've been using it now Now there was a political article that he wrote explaining all this after the election which was followed up by the top research scientist at Google search