136. How Russia Made Trump: Putin’s Troll Factory (Ep 5)

17:43Gordon:And then everyone leaves. And you can just imagine this video of very confused looking New Yorkers who thought they were going to have hot dogs. When it reverberates back in St. Petersburg, the Internet Research Agency trolls are like ecstatic, you know, hugging each other and crying because they've been able to demonstrate that they could get like seven New Yorkers to show up with the promise of free hot dogs. And we should say, I think, again, very much a theme of the Internet Research Agency effort. they are scoring themselves on inputs, not outcomes. Okay, that's going to be really important.

18:21Gordon:Okay, we got a few people to show up thinking they were going to have a hot dog. That's a big win. A lot of this stuff ends up getting packaged and sold up the chain in Moscow as great victories when in fact it's just sort of just activity.

18:39David:But it is wild when you stand back and you think that... Getting people to turn up for a non-existent hot dog giveaway is a Russian active measure. We're now putting that in the context of intelligence operations. It just gives you some idea that this is not spying as we used to know it or think about it, is it? It's something different. But that's the point of it, isn't it? I mean, that's the point of what they're trying to do.

19:02Gordon:They should have gotten one of those hot dog vending machines like we had at Langley. They would have been able to just have that sweet hot dogs roll out for those New Yorkers. Now, the staff who work in the American department, so the piece of the Internet Research Agency, this kind of translator project that's looking at the states, again, just kind of a further portrait that these are not the Pop-Tart eaters over at the GRU. They're pretty young. There's a great description of them as hipsters kind of taking smoking breaks, wearing stylish haircuts, having beards, which are apparently part of his sort of hipster culture in Russia.

19:43Gordon:The new employees, the employees of the Internet Research Agency break in. There's an internal document that they see. And it's very clear, though, that the goal of this effort, hot dogs and all, is to spread distrust toward the candidates, the American presidential candidates and the political system in general. So the election really is a core focus area for Prigozhin and his trolls right away. The head of the America Department is a 27-year-old named Zhekun Aslinov. And again, we apologize, as we always do profusely, on The Russia's Classified that we cannot, our pronunciations leave much to be desired.

20:23Gordon:He is an Azerbaijan-born guy who's apparently nicknamed Jay-Z. Which I think is just wonderful. These great details come from a fantastic book called Russian Roulette, written by Michael Isikoff and David Korn, all about Russia's intervention in the election. They've got just some wonderful details here. According to one former co-worker, Aslinov, Jay-Z, was more popular as a colleague than as a boss. He was, quote, a great guy, but frankly speaking, generally incompetent as a manager. So we've got really the crack team here in St. Petersburg. Now, Jay-Z has a budget of approximately$1 million a year.

21:12Gordon:Cue the Dr. Evil$1 million. It's not that much. It's strikingly low. It doesn't seem like that much money. But I guess the pay is pretty good.

21:24David:Yeah, I remember the pay was above average for people,

21:27Gordon:which is why they wanted to work there. Exactly. Now, the trolls, which I also feel like this term is really, it's very derogatory. Pejorative. Yeah, it is. Bearded hipsters in St. Petersburg who are attempting to sabotage our presidential election. They are paid bonuses on audience engagement and reactions in the US. So I would argue that this creates massive incentives for bullshit. It's like a self-licking ice cream cone, Gordon, is what this is. This is, again, a theme of the social media piece of the active measure. Lots of stuff going on, but as we'll see, I think pretty minimal impact. Now, the Internet Research Agency by 2016 had bought computing infrastructure and servers in the US.

22:15Gordon:They purchased space on the servers and set up dedicated VPNs to be able to route traffic into the US and make it look like it was coming from the US. And that tactic made it harder for social media companies in the States to actually uncover what was Russian disinformation operations on their platforms. Even long after, it had become clear that this influence operation was underway. By the time we get into this kind of summer fall of 2016, so as the election is going, the troll factory's audience online had grown to hundreds of thousands of direct followers. And I will note again, we'll use some large numbers in this section, but we are not trying to claim here.

22:59Gordon:I think we should probably say this right up front, that these numbers are not going to end up becoming particularly impressive when we talk about what actually happened. But the point is, is that there were a lot of Americans who were exposed in some way, shape or form to what the Internet Research Agency was up to. I would say that the managers Gordard inside Pregosian's shop, Jay-Z and his friends, were operating, even after that little trip that their friends took to the U.S., they were operating with some pretty rough and bizarre assumptions about what was actually going on in the U.S. So for example, they decided that infographics worked better with liberals than with conservatives.

23:41Gordon:They also decided that liberals were more active at night while conservatives got up earlier in the morning. I'm not sure if that's - Sweeping statements. Yeah, I'm not sure it's true.

23:51David:It's just sweeping statements. But yes, it doesn't seem like they'd done that much research, but there you go.

23:57Gordon:And the basic strategy was that they would create personas that impersonated activists or sometimes organizations and then boost those through paid ads to grow their following. And as many potential sort of aspiring influencers will discover, it can be kind of hard. It can be long, tedious work to build an audience for your social media account.

24:24David:Tell me about it, David. Have you seen my Instagram feed?

24:27Gordon:As evidenced by Gordon Carrera's Instagram feed.

24:32David:How many Instagram followers do you have, Gordon? Not that many. I only post when I'm told to post by, I shouldn't say this, but by colleagues at Goalhanger and Emma and others from our social media team. Post this and collaborate on this. You shouldn't say that. No, I shouldn't. It should sound much more organic. It's like an expression of my life and my real authentic self. Is that what it should be? 703 followers, I'm being told. 703.

25:01Gordon:I just looked it up. Gordon has 703.

25:04David:I should take some lessons from these internet research people. And some of them are probably bots. You're right. Thank you.

25:11Gordon:Some of them are probably bots. They're mostly... Gordon has bought nearly all of his 703 followers.

25:19David:Yes, exactly.

25:19Gordon:Yeah. Yeah. If Purgosian were still around, you could hire him to kind of whip your social media profile.

25:26David:Let's get back from my social media feed to the Internet Research Agency social media feeds. I mean, it is interesting, isn't it? Because they're creating these personas. And I mean, they're quite targeted in what they're trying to do, which I think is interesting. And they try lots of different things. You know, some of them work, some don't. But you can see that one of the interesting areas is they, for instance, target certain communities. So we talked about liberals and conservatives, but they target black voters specifically, don't they? And they are essentially trying to send a message to them to say, don't bother voting.

26:02David:I mean, it's really interesting because they're not saying vote for this candidate or vote for that candidate, but it's like, don't bother voting. It's not worth it. There's an element in which they're just trying to suppress the vote and make people disillusioned with politics. That was one of the tactics they used, wasn't it? It was.

26:20Gordon:And I think it's worth looking at potentially a particular persona because the Internet Research Agency created someone named Crystal Johnson. They chose a picture of a young black woman who appears to be in her early 20s. And by mid-2016, after some hard work by the bearded hipsters at the Internet Research Agency, the account had about 7 ,000 followers. So roughly 10 times the number of followers as Gordon Carrera's Instagram account. And Crystal's bio said, quote, it is our responsibility to promote the positive things that happen in our communities. She's supposedly from Richmond, Virginia.

26:57Gordon:In one of her more popular posts in June of 2016, she puts up a picture of the boxer Muhammad Ali and his star on the Hollywood Walk of Fame. And she jokes that Muhammad Ali's star is the only one hanging on a wall, not for anyone to step on. Now, that post has more than 22 ,000 engagements, but as I just said, not political. It has no impact on the election. It is designed, because the trolls understand this, they're trying to build an audience for Crystal Johnson, right? And it's designed to kind of attract people to her persona. There's another example. In late September of 2016, a Twitter account at Black to Live, it was one of the Internet Research Agency's most important fake black activist accounts that had a follower count of 11 ,200, but kind of mediocre engagement figures.

27:55Gordon:The account accumulated fewer than 190 ,000 social interactions in about a year and only 16 of the accounts, more than 2 ,600 posts during the run up mentioned Hillary Clinton. And most of those mentions were supportive. No, they didn't post at all in the weeks before the election about voter suppression. So you you can you kind of have like, yeah, there's a lot of people who are kind of seeing this stuff, but some of it's not political. Most of what they're seeing is just not political at all.

28:27David:It's audience building rather than political.

28:29Gordon:Yeah, that's right. Now, the Internet Research Agency, Jay-Z, and his crew in St. Petersburg were probably more successful with some of their conservative personas. And their most successful English language social media account was trying to mimic the Tennessee GOP, the grand old party, the Tennessee Republican Party. And by the end of September 2016, at 10 underscore GOP had just under 36 ,000 followers.

29:04David:That's more than me.

29:05Gordon:It's way more than you. It's way more than me. It's way more than the rest is classified. So shame on all of the tens of thousands of people who are listening to our podcast who haven't gone and checked us out on Instagram or Twitter or TikTok. Or subscribed on YouTube. Or subscribed on YouTube. Where is our bot army, Gordon? It should be there. We want a bot army. Alistair, Becky is typing into the chat, Alistair Campbell, Pregosian lookalike, has 195 ,000 followers.

29:37David:On Instagram, is that?

29:38Gordon:Is it on Instagram? I don't know how many of those are paid bots.

29:42David:How many inherited from people thought they were following Pregosian and then actually got Alistair Campbell.

29:47Gordon:If you're listening to this, go to at restdisclassified on Instagram and subscribe, please. Because these numbers -

29:55David:This is the worst sell. This is the worst sell.

29:58Gordon:This is a pity sell because when we compare our followers to this troll farm, it's terrible. Back to 2016 at Tennessee GOP.

30:06David:Yeah, 36 ,000 followers. And they get a lot of engagement.

30:11Gordon:Lots of engagement. Now, the account generates 3.2 million shares in the year before the election. It's unknown exactly how many of those interactions are with actual Americans, but it's probably the majority. And I think it's worth kind of zooming in on just a couple of the posts because among that account's top 10 pre-election posts on Twitter, five attempted to undermine the legitimacy of the outcome and flagged voter fraud. And one post, for example, published just the day before the election, earned more than 10 ,000 engagements. And it said, wow, another proof of hashtag voter fraud. Machine refuses to allow vote for Trump.

30:54Gordon:Retweet because media will never report this. That obviously is bad. Shame on the Russians. But again, it is striking that there's just a bunch of, you know, because I tend to think, you know, we tell these stories and you think about Russian disinformation. You sort of think about some kind of evil mastermind sitting in a room somewhere puppeteering the election or kind of reprogramming American minds. And the reality is there is basically a kind of factory-like vibe at the Internet Research Agency with all of our bearded hipster friends.

31:33David:Churning it out.

31:34Gordon:Just churning it out, right? One worker said they needed 135, 200-character tweets per shift. The workers on the different floors apparently don't talk much to each other. They only interact over smoke breaks and over lunch and coffee. So they're very isolated. And in a February 2018 interview with the Washington Post, so a couple years after the election, a former internet research agency employee said, I immediately felt like a character in the book 1984. The agency was a place where you have to write that white is black and black is white. Your first feeling when you ended up there was that you were in some kind of factory that turned lying, telling untruths into an industrial assembly line.

32:17Gordon:And I think maybe it's worth taking a break here, Gordon. It's worth coming back and discussing whether or not Evgeny Pregosian's troll empire accomplished anything. We'll see you after the break.

32:30David:When you want your spring break to feel like and your kid's pool day to feel like and your hotel bed to feel like

32:43Gordon:and room service to feel like because at Hilton, hospitality feels like

32:49David:your cabana's ready. Would you like fresh towels?

32:52Gordon:It matters where you stay.

32:55David:Book now at Hilton.com. Hilton for this day.

33:01Gordon:We're leaving today and entering a world of Mickey Mouse waving, princess meeting and greeting, lightsaber clashing, the Twilight Zone Tower of Terror dropping, banshee flooding, Space Mountain launching Galaxy rewinding What's the whole death of galaxy rewinding? Fireworks igniting World of Other Worlds For whatever you love Infinite worlds await at the most magical place on Earth Walt Disney World Resort Right now, get up to 20 % off select online storage solutions Put heavy-duty HDX totes to good use Protecting what's important to you

33:39David:The solid, impact-resistant design prevents cracking and the clear base and sides make items easy to find, even when the totes are stacked. Find select online shelving and tote storage up to 20 % off at the Home Depot to organize every room in your home, from your garage to your attic. Visit HomeDepot.com.

33:59Gordon:How doers get more done.

34:06David:Welcome back. So Evgeny Prigozhin's bearded hipster trolls have been busy and they're busy right through the election, aren't they, David? Right until November and Election Day. But the question is, what impact did they really have? There's always a figure which is cited by I think Facebook after the election says that the Internet Research Agency reached 126 million Americans. Now, that sounds extraordinary and led people to go, wow, that's nearly half of all Americans were reached by these posts. But it's not quite the same as saying it had an impact, is it? No.

34:48Gordon:And I think the bottom line up front here is that these efforts didn't have much of an impact. And the first reason is that those numbers are the 126 Americans, a lot of them are reached by really anodyne, benign posts made by these accounts that had nothing to do with politics or the election whatsoever. So that's one piece. Another piece is only 8.4 % of the Internet Research Agency's activity was election-related at all. And the pieces that were mostly stayed very much within echo chambers and did not cross over into networks of undecided voters. That's a really important point is a lot of this was conservatives or Democrats amplifying stuff that they already heard and believed and saw elsewhere.

35:42David:Which is unlikely to change someone's vote. It might increase the intensity of their view. And, you know, you could have some second order effects, but it's definitely different from actually persuading someone. It's very hard to measure what's reaching someone and what actually changes them. They managed to do a few things in the real world, though, didn't they, where they managed to create events. We talked about this earlier with the hot dogs. And they managed to do that during the election campaign, which, again, is interesting, I think, but not necessarily impactful.

36:11Gordon:There does seem to be evidence that in at least two cases, the Internet Research Agency was able to actually arrange in-person kind of rallies or events. And in one case, they paid a Republican political activist in Florida. And I don't think she knew who was paying them, but they paid her to dress up as Hillary Clinton. And this woman happily did. She had a blonde wig. She had an orange prison uniform that she bought at Goodwill. Because you remember, of course, one of the kind of major MAGA chants during the election was to lock her up. Lock her up. And this woman, when she's confronted with the fact that it's actually the Russians who had paid her and set this thing up, she said, well, I would have done it for Trump anyway.

36:50Gordon:There's a lot of excitement around this, and Russians didn't have any part of that. This wasn't a trick for me. So I think, again, it's this idea that if anyone else had reached out and said, go ahead and do this, she would have done it. And she was already voting, and everyone who's showing up at that rally, they're already voting for Trump, presumably. So that's another piece of it. I think another slice is that we talked, we joked about this a little bit with the way that the Internet Research Agency employees were getting to know the target. There wasn't a lot of regional or cultural specialization and it prevented, in other words, a lot of the trolls didn't know anything, much of anything, about the communities, the states that their personas were alleging to come from or out of.

37:37David:Despite the road trip.

37:38Gordon:Despite sending two people on a road trip across the United States, somehow they weren't able to pass themselves off as natives of many of these states and communities, right? So one internet research agency member says, he said, first, you got to be a redneck from Kentucky. Then you need to be a white guy from Minnesota. That's actually me, Gordon. I'm a white guy from Minnesota. And then 15 minutes later, you're from New York posting in some slang. That's what this guy said. So it's like, yeah, you probably are going to struggle as a Russian in St. Petersburg who's just on this assembly line to really kind of write copy for your accounts that's going to be credible and that other people are going to want to interact with.

38:22Gordon:Right. The other piece of this is that there were about a billion tweets related to the campaigns and that were posted in the 15 months leading up to the election.

38:34David:By everyone, you mean? Yeah. By everybody.

38:37Gordon:By everybody.

38:38David:Everything election-related.

38:39Gordon:Everything election-related. The internet research agency generated less than 0.05 % of all election-related posts. When I read that, I'm kind of like, that's a small number. Then I thought, it's actually not an insignificant number for one operation in St. Petersburg, Russia. But the point is, is that it was absolutely drowned out by the organic tweets created by Americans during the election. And to just kind of seal the point, the most engaged content was designed not to polarize, but to build communities around these personas so that more and more people would follow them. Now, the House Permanent Select Committee on Intelligence released about 2 ,300 of the Facebook ads that the trolls had placed over the two years preceding the election.

39:29Gordon:And the top 10 most popular ads, which accounted for a quarter of all views, none of them had sharp or corrosive disinformation of them. So again, it's all kind of trying to drive people to these personas to build their profile. Now, the best known, the most widely covered Facebook ad depicted Satan in an arm wrestling match with a white robe Jesus. And the caption says, Satan is saying, if I win, Clinton wins. And Jesus says, not if I can help it. The New York Times reported on and reproduced that ad after the election, which obviously is, you know, it's famous.

40:07David:It's iconic as one of the more famous ones.

40:09Gordon:The original ad was one of the internet research agencies least successful. It displayed for one day. It cost them 64 rubles, which at that point was$1. And it had 71 impressions, which is probably about the same number of impressions that you're getting on some of your Instagram posts, Gordon. I was going to say, I'd be pretty happy with that. 14 Americans clicked on it.

40:39David:Intel. And yet that becomes the iconic one, which everyone points to, and it's on the New York Times and everywhere else. I think it's a really good example about how outsized the media reporting of it was compared to the potential actual impact of it, which isn't to say that it's meaningless, but it's just, I think that perspective is important. That's right.

41:00Gordon:I mean, the median number of impressions for all of the pre-election political ads thrown out there by the internet Research Agency, the median number of impressions was 199. And impressions is literally, I looked at it. I scrolled past it. I didn't actually read or engage with it. So -

41:19David:Yeah, it doesn't mean it did anything. Yeah. Exactly. But I do think what's interesting is the legacy of this operation will be enormous because particularly in the long term, the press coverage of it, the questions around, was their targeting through social media and remember the whole Cambridge Analytica thing and the ability to manipulate people. It became a real thing and a real obsession. And that arguably overstates the impact of it and creates this sense that the Russians are there and almost all powerful and have the ability to manipulate American politics. When in this case, I think the actual evidence is much weaker.

42:00David:So it's an active measure, which it doesn't necessarily have a huge impact on the election, on people's views, but it does have a much longer tail in terms of the kind of cultural, social, and political impact after the election, which maybe we'll get to when we get to the end of the series, in terms of what it means. So I think there's an interesting dichotomy there, isn't there?

42:21Gordon:And I think it's worth summing up that this piece of the Russian sort of disorganized active measure, right? The siloed active measure. This was, I think, the least effective component of what the Russians did in 2016. But the hack and leak -

42:38David:Yeah, it's much more significant.

42:39Gordon:Is much more significant. And by October of 2016, the Trump campaign is beginning to signal that another dump of information is coming. The hack and leak is really, unlike the social media piece, is going to be really integrated into the TikTok, the chronology of events in the run-up to the election. And of course, once again, Gordon, it's Roger Stone, Trump sort of campaign advisor who is in the middle of the chaos. Because on Sunday, October 2nd at 12.52 AM, I think in Trump world, you have to tweet in the middle of the night. It's important that important tweets go out.

43:17David:But I thought conservatives are up early, though.

43:20Gordon:Stone's up. He got up at midnight. No, he didn't stay up late. He got up early, Gordon. He got up really early. He got up at 12.45.

43:29David:But he's tweeting.

43:30Gordon:That's the point. He's tweeting and he tweets, Wednesday, at Hillary Clinton is done, hashtag WikiLeaks. So again, we have Roger Stone's got this weird foreknowledge of what Julian Assange is up to. Then later that day, Stone appears on the show of Infowars conspiracy theorist, Alex Jones. Did Alex Jones, did he ever break into the UK market, Gordon?

43:53David:I think I'm aware of Alex Jones, but I don't think he was as big over here. I think it's fair to say.

43:58Gordon:OK. And Stone says, you know, I'm assured the mother load is coming on Wednesday, says the release will be devastating. Also says that Assange was scared that the globalists and the Clintonites are trying to figure out how to kill him. And then he puts out more tweets on the 3rd of October saying that the payload is coming. The following day, Stone is contacted by a senior Trump campaign official who wants to know what on earth WikiLeaks is planning. And Stone writes back that Julian Assange is going to release a load every week going forward. Donald Trump Jr. had intermittently been in private contact with WikiLeaks.

44:39Gordon:And we should say for those listeners who want to go deeper into some of the campaign connections with WikiLeaks and the Russians and the kind of whole Trump-Russia nexus, what's fact and what's fiction, do go and check out our mini-series for club members on this. And you can join at therestisclassified.com. But Donald Trump Jr., who's on again, off again in private contact with WikiLeaks, he wants to know more about what Stone knows. Now, a couple weeks earlier on September 20th, WikiLeaks' Twitter account had actually sent Donald Trump Jr. a private message asking him about a new election blog called PutinTrump.org.

45:23Gordon:I wonder if that website is still up, that had been created by a political action committee that had been funded by an internet entrepreneur. And Donald Trump Jr. replies, he says, off the record, I don't know who that is, but I'll ask around. And then he emails Steve Bannon, campaign advisor, Jared Kushner, Killian Conway, who's another advisor about his exchange with WikiLeaks. So on the 3rd of October, WikiLeaks reaches out to Donald Trump Jr. again with a private request and says, hiya, it'd be great if you guys could comment on or push this story. And then WikiLeaks attaches an article alleging, I think falsely, that Clinton had once suggested she wanted to, quote, just drone Assange.

46:07Gordon:And then Donald Trump Jr. says, already did that earlier today. It's amazing what she can get away with. And then a couple minutes later, Donald Trump Jr. is, you know, he's kind of trying to figure out what's going on with WikiLeaks. He messages WikiLeaks referring to the Stone tweet and says, what's behind this Wednesday leak? I keep reading about. And so basically what we have is the Trump campaign realizing that a new and potentially very damaging leak is coming that could be fodder that they could use against the Clinton campaign.

46:40David:So there, David, let's stop as this dramatic election campaign in 2016 reaches its final stages, looking incredibly tight. Trump versus Clinton with the Russians lurking in the background. And we'll see in the final stages what the Russians get up to and how the US national security community responds. And a reminder, of course, you can listen to that episode right now. If you're a club member, join at therestisclassified.com and you'll also get access to that bonus series. but otherwise we'll see you next time see you next time

47:19David:rinse knows that greatness takes time but so does laundry so rinse will take your laundry and hand deliver it to your door expertly cleaned and you can take the time pursuing your passions time once spent sorting and waiting folding and queuing now spent challenging and innovating and

47:35Gordon:pushing your way to greatness. So pick up the Irish flute or those calligraphy pens or that daunting Beef Wellington recipe card and leave the laundry to us. Rinse. It's time to be great.