S1: Shortly before the midterms in 2018, Facebook launched this tool. They call it the ad library. It’s a digital archive of political messaging. It’s also a Real-Time calculator tallying up how much each candidate is spending on the social network. Andrew Morans over at The New Yorker. He’s been spending a lot of time in this archive watching the candidates, especially President Trump, make a case for themselves.
S2: The general election hasn’t even started and the Trump campaign is already treating it as if the general election is in full swing.
S1: There are so many clips and posts catalogued here. Once you dive in, you could scroll pretty much forever.
S2: So there were many weeks last year, for example, where the Trump campaign was spending a million dollars a week on Facebook and a million dollars a week on Google, meaning search ads and YouTube ads. And the Democrats, even though there were five hundred million of them, were essentially spending nothing.
S3: Since the twenty eighteen midterms, the president has spent more than $44 million on Google and Facebook advertising alone. What do these ads look like? Like? Do you have a favorite digital ad?
S4: A favorite is maybe not the term I would use, but yeah, there they are. Some of them are very standard. You know, some of them are, you know, the president is keeping us safe or whatever.
S1: There are ads for Trump themed straws and ads pushing donor dinners with Melania, but also ads that ask you to hold the corrupt Democrats accountable. In the middle of impeachment, Trump’s campaign started pushing ads like this one.
S5: Prosecutors not fired. You’re not getting the money. Son of a bitch got fired.
S1: But when President Trump these ads claim President Trump is fighting corruption while Joe Biden encouraged it, the Biden campaign asked Facebook to take them down because they were full of objective lies.
S6: And Facebook said, no, we can’t do that, because when politicians lie, it’s not against our rules.
S7: And they read this one stat that sort of stopped me in my tracks.
S8: It was about Joe Biden, who’s now the Democratic frontrunner at this point, and he was looking at his spending on Google and Facebook and he was spending less than any of the other Democratic presidential candidates. And I just thought, whoa, yeah, yeah.
S9: It’s such great evidence, different messaging strategy.
S10: But yeah, I mean, there are many things to say about the ways in which the Biden campaign is promising a return to the past and to the policies of the Obama administration. And they also seem to be relying on media strategies that come out of 2008, which is not encouraging.
S3: Today on the show during the last presidential election. Trump’s campaign used Facebook and Google to nudge voters in ways no political operation had before. And in 2020, their digital strategy is only ramping up. So, Andrew, it’s going to take us inside Trump’s digital war room. I’m Mary Harris. You’re listening to what next. Stick with us.
S11: Trump campaign ads are famously bigoted, cynical, crude. But this is intentional, based on millions of little experiments to find the message that engages the desired audience. These experiments started in 2016 when Trump’s campaign set up a way to create and push out ads for Twitter, YouTube and most importantly, Facebook. There is a process for setting each ad’s performance, picking exactly who would see what and when. At first, the strategy was somewhat maligned at the time.
S12: This was all sort of dark arts as one of the Clinton campaign staff called it, which, you know, some people would say is hyperbolic because a lot of this stuff wasn’t breaking any laws or anything.
S13: It wasn’t dark arts in that sense. It was really just new. And one of the ways in which it was new is that it was very, very selectively targeted. So you could target a Facebook ad to a select group of people like literally by name. So as opposed to a TV ad where, you know, you could run it in Ithaca, New York, or you could run it in Albany, New York, which is, you know, a sort of crude kind of targeting on Facebook. You could just make an Excel spreadsheet with the exact forty nine people.
S1: You want to see the ad and what did these ads look like?
S13: Well, they were all different. So the ones that they sent to everyone were generally the most benign. They were Trump wants to keep us safe or Trump wants to build a wall. But there were always things in theory that they could do when people weren’t watching that were a little bit more ominous. Well, there were PACs supporting Facebook pages like there was one called Defeat Crooked Hillary that would pump out essentially disinformation, alleging that Hillary actually had a close relationship to Vladimir Putin or that she was actually on all kinds of steroids or, you know, illicit drugs because she had unspecified health conditions. These are things that you probably would not want to run a national TV ad about because there would be a lot of understandable blowback. But if you’re just sending it to people who, you know, are super gung ho fans of yours on Facebook and it’s you know, you’re targeting the red meat to the right people, then you might be much more kind of unrestrained about it, which it appears they were.
S1: And were these kind of darker ads, were they being amplified by the campaign itself?
S13: Yeah, they were. They they were. Yeah, for sure. They were sometimes coming from official campaign channels. Other times coming from PACs that were because of the nature of superPACs, officially separated from the campaign, but pretty clearly getting money from the same sources and pretty clearly affiliated in the way that it’s possible to see which superPAC is affiliated with which campaign. So yeah, these are definitely all intertwined.
S11: By the end of the campaign, the Trump team had released almost 6 million different ads. Many of them were just small variations on the same basic message tweaked to appeal to the maximum number of people.
S13: So it’s Trump looking up and pointing to the left’s Trump looking up and pointing slightly to the right. A green banner that says Donate here, a red banner that says contribute here. These are all considered different ads by that calculus. I think it’s easy to see those as sort of arbitrary minor distinctions. But the way social media works is that what they call a B testing is very important. Testing often through kind of machine learning and software. Which of these tiny variations is more effective? This is something that the Trump campaign was way ahead on and really good at and really, really willing to experiment with. And to the naked eye, it might seem like a green contribute button is the same as a red donate button or whatever, but when you actually apply machine learning to it at scale, it might turn out that one is 10 percent better at getting a desired result than another, and that can mean millions of dollars.
S1: The other thing is you really make it clear that it’s not just about Facebook ads. You could add one campaign staffer saying like Facebook was driving everything, like where to send canvassers, where to hold rallies, even what Trump should say at those rallies.
S14: Yeah. And the staffers went on to clarify that the advice they gave to Trump was not always heeded. But yeah, they were able to use this kind of data to suggest, you know, because remember, you’re not only pumping stuff out, you’re also getting data back on how people are responding. So you’re not only putting out ads and surveys and, you know, click here if you support the president, you’re also getting all kinds of data on what time of day is most effective to target people. Which kinds of people? You know, men over 65 who have not yet retired because their pensions haven’t kicked in. You can you can cut people into very, very tiny specific slivers. And then you can see how that sliver of the electorate is responding to a particular kind of message. So if men over 65 who haven’t been able to retire yet really want you to talk about how you’re going to protect Social Security, you can use that data to say, you know what, I’m going to bring up Social Security a lot when I’m in this or that city.
S1: Yeah. I mean, it stood out to me that you could. Kind of fragmentary information, like a cell phone number and email address. You could just have little bits. And Facebook could kind of fill in the rest. And then you could use that sort of pumped up list with more information and find more people like it. So it’s create this kind of snowball.
S13: Absolutely. And because elections are decided on such tiny margins in this time in our history. Finding specific people who might be not tuned into your candidacy or who might not be that motivated to vote. Finding people on little scraps of information is a huge part of the name of the game. And we talk about people switching sides from Democrat to Republican or vice versa. But there’s also a huge universe of voters who are switching from non-voters to voters, and that that was a huge part of the Trump campaign strategy, too. And a lot of those people you can only find them by buying consumer data on where they’ve shopped or which social networks they’ve logged on to. They might not be on other kinds of voter rolls or anything like that because maybe they’ve never voted before.
S11: So these are the rough contours of the Trump campaign’s digital media operation. It’s innovative, a bit rogue, even by the standards of political messaging. And it’s widely credited with helping Trump win in 2016. The guy calling the shots back then, his name is Brad Paşa Scale.
S1: People may be familiar with his name because now he’s Trump’s 20:20 campaign manager. But at the time, he was the digital director of the campaign. But he didn’t really come from the political world. Where did this guy come from?
S12: Yet, Brad? Preschool essentially had no political experience before Trump. He like a lot of people around Trump. He started as a kind of Trump family loyalist in a Trump business loyalist. So he was a guy in living in San Antonio. He’s originally from Kansas. But he moved to San Antonio, went to college there, started learning how to build websites. And one of the Web contracts he got. The biggest one by far at the time was for Trump Real Estate. And then from there, he just sort of stayed loyal to the family and built them Web sites for under market rate, which they liked. And he was just sort of the guy who was around and who they thought of as a guy who knew how to do computer stuff. And when they decided that they had to figure out how to do digital campaign operations, he was the guy who got tapped to do that.
S1: He started as a web designer. How did pass scale turned to Facebook?
S12: Around 2000. Ten eleven twelve. Web development starts to be more of an interactive and sort of it started being more about finding where people’s attention was likely to flow to and either targeting your messages to that attention or targeting people’s attention in the direction where you wanted it to go. So rather than just making these sites, he started getting good at things like SEO search engine optimization. He started getting good at word tagging. There were all these little tricks that he couldn’t sort of employ that would widen the appeal of the sites he was building.
S6: And he talks about, for instance, building a Web site for a local gun store and showing them how, you know, I can make it so that people all over the country are going to want to buy guns from you, which is something they had never really thought about because they were, you know, a traditional local business. And he was sort of starting to get into the habit of taking things that used to be local and sort of nationalizing them through these new ways of disseminating information.
S1: I mean, you just PA skill as a moral a couple of times in your piece. And I’m wondering what led you to that conclusion.
S12: Yeah, I don’t I don’t think I directly conclude that he’s amoral because I just I didn’t talk to him. So I don’t want to armchair diagnose him as a person. But he definitely ran an organization that seemed to act in amoral ways, similar to the way that social media companies often act in amoral ways. How so? Well, I just I think it should be clear that I didn’t say immoral. Amoral doesn’t necessarily imply that you are a bad person or that you’re setting out to harm people. What it implies to me is that through everyone I talked to about how personal operated, he was sort of mercenary. He was out to make a name for himself and be a good business guy. And that, you know, often comes from trying to win and dominate and do a good job for your client.
S15: It doesn’t translate to trying to think a few steps down the line to what are the moral implications of what I’m doing and will it, you know, destabilize global geopolitics and, you know, exacerbate climate change and, you know, all the stuff that matters to the world. It’s more about just crushing the competition.
S16: Part of the reason Pass Gill’s operation seems so insidious is because the rules for political advertising online, they aren’t as clear cut as the ones for television and radio. So targeting offensive ads to a very select group of people, it may seem wrong, but in reality, Trump’s campaign wasn’t breaking any laws, just generally agreed upon norms and institutions that are supposed to be keeping track of all this. Like, say, the FEC, they’re woefully unequipped.
S6: Yeah, I think that’s true. Not only of the FEC, but also of the FCC and also the FCC. And I mean, not even governmental bodies.
S12: Just all of us are kind of outgunned by this because it changes so quickly. And I think the challenge is to kind of do two things at once. Right, to maintain fidelity to these very longstanding cherished norms such as the First Amendment, while also trying to adapt and be really skillful about and observant of how these things change. It’s very hard to maintain the same exact set of rules and norms over hundreds of years when the means by which those norms are being implemented have completely changed.
S1: Well, the other question is what’s Facebooks role here? Because they’ve been so key to all of this. You talked about how a bunch of Facebook employees came together and had a list of here are things that we could do right now that would sort of gum up the works for someone looking to push this system to the max when it comes to politics. And they had there was a multi-point plan. But how did Facebook respond to that?
S12: They just brushed it off. I mean, they they took the sincere desires of hundreds of their employees and said, thank you, we hear you and no thanks. We’re not going to do any of it, which, you know, is pretty upsetting. I mean, they they make this argument that this is about free speech and everybody should have a voice and all that stuff, but it just doesn’t really hold up to scrutiny. This isn’t, you know, the people who signed that open letter. We’re not asking them to curtail free speech. They were asking them to label things properly, to make sure that things were a little more vetted, to invest a little bit more resources and attention in helping users pass information from disinformation.
S17: And Zuckerberg essentially said, no thanks.
S18: These companies, such as Facebook, are not really trying to help us get it right because they’re based on the ethos of hacking and moving fast and breaking things and just sort of innovating for innovation sake. So it’s not like when we’re talking about the use of the word amoral. The reason I apply it to the companies as much as I apply it to people like PayScale is that when you build a whole internal ethos on innovating and breaking things, not everything you do is going to have a beneficial moral outcome. And it’s just you sort of offload that and say, well, that’s not our job to worry about.
S11: And if the story of 2016 was the Trump campaign’s Facebook strategy, the story of 2020, that might be how the campaign has continued to evolve beyond Facebook. Now they want to be in your Demme’s.
S12: So they, for instance, have invested heavily in text messaging technology. One of the current digital employees of the campaign, Gary Kobie, has a text messaging startup and he’s very invested in that technology. And they that’s a smart investment because text messages are opened at a way, way higher rate than emails or social media posts or anything else.
S1: I mean, you talked about how in order to get into a Trump rally now, you need to give your cell phone number. That’s how you get your ticket. It’s free, but you have to give up a little data about yourself.
S12: Yeah, they’re they’re very adept at harvesting data. They brag about it all the time, but it’s not always clear exactly what they’re doing with it. And on some level, I think that should sound as creepy as it does on another level. It’s also kind of just becoming industry standard. And so part of the issue is, you know, it’s not really clear whether the Democrats are going to get their hands dirty with that kind of similar kind of data acquisition and manipulation, or if they’re going to stand by and say, you know, this isn’t really for us, you know, it creeps out our base too much. We’re going to sort of let this opportunity pass us by. You could argue that that’s the ethically upright thing to do.
S1: But it it certainly seems like it would put them at a disadvantage when Michael Bloomberg had been building this digital infrastructure, spending tons of money on Google and Facebook. And he’d said from the beginning, if I drop out, I will sort of hand over the keys to this operation, to whoever the nominee is. But can his operation keep up with what the president has been building over the last three or so years?
S12: Yes and no. I mean, $60 billion can certainly buy you a lot. And as much as the story last year was, no one can catch up with Trump’s spending. I think that’s no longer true when you have the 16th richest person in the world on your side, which it seems like the Democratic nominee will. That said, there are a lot of things that only time can do for you. I mean, the Trump campaign has just had a long headstart in terms of how repeated and relentless their messaging has been, and no amount of money can catch up with that.
S1: And they’ve merged their information with the RNC. Is information right? So you sort of expanded it’s a huge war chest.
S12: So I think, you know, Bloomberg can pour a lot of money into the Democratic operation. Things like time and things like data. You know, you can’t just buy data overnight. It takes time to acquire it because data is not just a fungible commodity. You have to get it from the right people in the right moments and make sure it’s clean and make sure it’s accurate. So there’s still very much an arms race here.
S1: Just a few days before the actual election itself, back then in 2016, Bloomberg BusinessWeek visited the Trump campaign when no one thought Trump would win. But everyone was still marveling that he was in this thing and the way they characterized the digital operation that Brad PayScale had built. They called it Trump’s Plan B. Yeah. And they they clearly thought that, well, he might lose this election, but now he’s going to have all of this information about people. What stood out to me about that now was that that’s still true. Even if Trump isn’t re-elected in the fall, he’s still going to have all this data. All of this information that he’s hoovered up.
S12: Yeah, I I don’t think he’s ever gonna go away.
S6: I just think, you know, look, a lot of these tactics that people use to advertise political campaigns on Facebook, those were borrowed from tactics that were used to advertise brands and products on Facebook. And in fact, if it hadn’t been Parkhill, it likely would have been someone else. I think it’s not that he has some singular history changing genius or something.
S19: He is someone who spent the time and money and attention to use Facebook and these other platforms to their fullest possible extent. And that’s what’s worth paying attention to that you can harness the energy and power of these platforms to do this stuff. If you have enough time and money, you don’t have to be some singular genius. They’ll kind of take the reins and do it for you.
S20: Drew morant’s, thank you so much for joining me. Thank you. Yeah, it was great.
S21: Andrew Morans is a staff writer at The New Yorker. And if you like this interview, you will love his new book. It’s all about how social media has twisted political discourse. It’s called Anti-social. All right. That’s the show. What next is produced by Mary Wilson. Jason De Leon. Daniel Hewitt and Maurice Silvers.
S22: This episode was recorded in a moving car, a closet and, yes, from underneath a hoodie. It’s pretty amazing technology we’ve got over here. I’m Mary Harris. I’ll talk to you tomorrow.