WEBVTT 00:00:00.000 --> 00:00:02.259 Welcome back to The Deep Dive. Today, we are 00:00:02.259 --> 00:00:06.320 wading into waters that I think most people believe 00:00:06.320 --> 00:00:08.599 they know how to navigate, but in reality, they're 00:00:08.599 --> 00:00:10.259 probably browning without even realizing it. 00:00:10.480 --> 00:00:14.500 We are talking about media bias. And I know as 00:00:14.500 --> 00:00:16.940 soon as those words leave my mouth, about half 00:00:16.940 --> 00:00:18.940 the audience is probably picturing a specific 00:00:18.940 --> 00:00:21.940 cable news anchor screaming at the camera. Right. 00:00:21.980 --> 00:00:23.960 And the other half is picturing a different anchor 00:00:23.960 --> 00:00:26.519 screaming at a different camera. It's the immediate 00:00:26.519 --> 00:00:28.920 reflex, isn't it? We hear media bias. Right. 00:00:28.980 --> 00:00:31.320 And our brains just snap right into that culture 00:00:31.320 --> 00:00:35.460 war framework. Red versus blue, Fox versus MSNBC. 00:00:35.600 --> 00:00:38.320 It's like we treat it as a team sport where the 00:00:38.320 --> 00:00:40.820 other side is always cheating and our side is 00:00:40.820 --> 00:00:44.700 just. And that's the whole problem. We're so 00:00:44.700 --> 00:00:47.380 focused on the players and the teams that we 00:00:47.380 --> 00:00:49.780 completely miss the fact that the entire stadium 00:00:49.780 --> 00:00:52.399 is tilted. The game itself is rigged, but not 00:00:52.399 --> 00:00:54.880 in the way we think. Right. And if we just spend 00:00:54.880 --> 00:00:57.780 the next hour bashing pundits or, you know, complaining 00:00:57.780 --> 00:01:01.189 about fake news. We are wasting everyone's time. 00:01:01.310 --> 00:01:03.329 Because the research we have in front of us today 00:01:03.329 --> 00:01:05.930 suggests that the political stuff, that partisan 00:01:05.930 --> 00:01:08.989 shouting, is just the tip of the iceberg. A very, 00:01:08.989 --> 00:01:11.909 very noisy tip. A very noisy tip, for sure. But 00:01:11.909 --> 00:01:14.230 the really dangerous part is the stuff we don't 00:01:14.230 --> 00:01:16.670 see. The stuff that's happening below the surface. 00:01:16.810 --> 00:01:19.450 Exactly. The mission today is to look at the 00:01:19.450 --> 00:01:22.180 mechanics under the hood. We aren't really looking 00:01:22.180 --> 00:01:25.140 at who is biased so much as how the machine is 00:01:25.140 --> 00:01:27.780 built to produce bias almost automatically. We 00:01:27.780 --> 00:01:30.019 need to move beyond this idea that bias is just 00:01:30.019 --> 00:01:33.040 some angry editor in a back room plotting to 00:01:33.040 --> 00:01:36.980 destroy the country. The reality is much more 00:01:36.980 --> 00:01:39.319 boring. And because of that, it's actually much 00:01:39.319 --> 00:01:41.579 more terrifying. How so? Because it's structural, 00:01:41.780 --> 00:01:43.859 it's economic, and it's deeply psychological. 00:01:43.900 --> 00:01:46.000 It's not a conspiracy. It's a business model. 00:01:46.329 --> 00:01:48.329 We are going to talk about ski resorts lying 00:01:48.329 --> 00:01:52.290 about snow, baseball fans in the 1890s, and maybe 00:01:52.290 --> 00:01:55.430 the most uncomfortable part, why your own brain 00:01:55.430 --> 00:01:57.790 is actually the one demanding to be lied to. 00:01:58.069 --> 00:01:59.810 That's the part people don't like to hear. Yeah, 00:01:59.849 --> 00:02:02.730 I bet. But before we get to all the weird history. 00:02:03.420 --> 00:02:06.299 We need a baseline. How are we even defining 00:02:06.299 --> 00:02:08.620 bias here? Because I feel like that word has 00:02:08.620 --> 00:02:10.740 completely lost all meaning. It's just become 00:02:10.740 --> 00:02:13.300 a synonym for something I disagree with. Well, 00:02:13.379 --> 00:02:15.780 that's definitely the colloquial use. Yeah. But 00:02:15.780 --> 00:02:18.379 if we look at the sort of academic definition 00:02:18.379 --> 00:02:21.340 provided in the research we're looking at, media 00:02:21.340 --> 00:02:24.099 bias is defined very specifically. It's when 00:02:24.099 --> 00:02:27.300 journalists and news producers contravene the 00:02:27.300 --> 00:02:29.909 standards of journalism. Things like accuracy, 00:02:30.229 --> 00:02:33.689 fairness, and objectivity in a way that implies 00:02:33.689 --> 00:02:36.969 a pervasive or widespread drift. A drift. I like 00:02:36.969 --> 00:02:39.669 that word. It's key. It's not just one guy having 00:02:39.669 --> 00:02:42.150 a bad day or, you know, a cranky opinion column. 00:02:42.490 --> 00:02:44.689 It's a systemic violation of the rules of the 00:02:44.689 --> 00:02:46.990 game. It's a tilt in the floorboards that makes 00:02:46.990 --> 00:02:49.080 everything slide in one direction. That's a great 00:02:49.080 --> 00:02:51.360 way to visualize it. And I think the most interesting 00:02:51.360 --> 00:02:53.039 part of the research we're diving into is that 00:02:53.039 --> 00:02:55.759 this tilt isn't always political. In fact, some 00:02:55.759 --> 00:02:57.860 of the strongest, most reality -warping biases 00:02:57.860 --> 00:03:00.319 have absolutely nothing to do with who you vote 00:03:00.319 --> 00:03:02.699 for. No, not at all. They are what you might 00:03:02.699 --> 00:03:05.539 call invisible biases that distort reality for 00:03:05.539 --> 00:03:08.020 everyone, regardless of their politics. Okay, 00:03:08.060 --> 00:03:10.180 so let's start there then. The outline calls 00:03:10.180 --> 00:03:13.180 it the taxonomy of bias. We all know partisan 00:03:13.180 --> 00:03:15.060 bias. That's the my team good, your team bad 00:03:15.060 --> 00:03:17.080 stuff. We can put that to the side for a minute. 00:03:17.879 --> 00:03:20.699 But the research lists this whole zoo of other 00:03:20.699 --> 00:03:24.020 biases. Where do we even begin? Which one is 00:03:24.020 --> 00:03:27.159 the most pervasive? I would argue the most pervasive 00:03:27.159 --> 00:03:29.979 and certainly the least discussed is something 00:03:29.979 --> 00:03:33.240 called concision bias. Concision, as in being 00:03:33.240 --> 00:03:37.060 concise, being brief. Yes, exactly. OK, I'm going 00:03:37.060 --> 00:03:40.060 to play devil's advocate here. Why is that bad? 00:03:40.599 --> 00:03:42.719 I don't have three hours to read a manifesto 00:03:42.719 --> 00:03:44.580 every morning. I want the headlines. I want the 00:03:44.580 --> 00:03:47.659 bullet points. Isn't brevity a virtue? Brevity 00:03:47.659 --> 00:03:50.240 is absolutely a virtue for communication, but 00:03:50.240 --> 00:03:52.819 it can be a poison for understanding. Concision 00:03:52.819 --> 00:03:55.460 bias is the tendency of the media to report views 00:03:55.460 --> 00:03:58.340 that can be summarized succinctly while crowding 00:03:58.340 --> 00:04:00.800 out unconventional or complex views that take 00:04:00.800 --> 00:04:03.159 time and effort to explain. Give me a concrete 00:04:03.159 --> 00:04:05.240 example of that. Think about a broadcast news 00:04:05.240 --> 00:04:08.319 format or even a TikTok video or a Twitter thread. 00:04:08.560 --> 00:04:12.000 You have seconds. Right. Second, if I ask a guest 00:04:12.000 --> 00:04:14.599 on a show, why is the economy struggling? And 00:04:14.599 --> 00:04:17.199 they say, because taxes are too high. That is 00:04:17.199 --> 00:04:20.620 punchy. It fits on a bumper sticker. It feels 00:04:20.620 --> 00:04:23.579 intuitively true because it's so simple and clean. 00:04:23.939 --> 00:04:27.620 But if the real answer is, well, we have a lag 00:04:27.620 --> 00:04:29.899 in supply chain logistics from three years ago, 00:04:29.980 --> 00:04:33.259 combined with an inverted yield curve and shifting 00:04:33.259 --> 00:04:35.540 consumer confidence in the housing sector. Exactly. 00:04:35.540 --> 00:04:37.439 You've already lost the audience. And more importantly, 00:04:37.579 --> 00:04:40.730 you've lost the producer. The producer is in 00:04:40.730 --> 00:04:42.569 your ear screaming, wrap it up. We have commercial 00:04:42.569 --> 00:04:44.750 break for insurance coming up. So the complex 00:04:44.750 --> 00:04:46.430 truth just gets left on the cutting room floor 00:04:46.430 --> 00:04:49.069 and the simple, incomplete explanation gets aired. 00:04:49.310 --> 00:04:52.189 Precisely. The bias is toward simplicity. It 00:04:52.189 --> 00:04:54.730 privileges the soundbite over the syllabus. And 00:04:54.730 --> 00:04:57.550 the real danger is that over time, we start to 00:04:57.550 --> 00:05:00.870 equate simple with correct. We start to think 00:05:00.870 --> 00:05:03.250 that if something can't be explained in two sentences. 00:05:03.899 --> 00:05:06.639 It's probably nonsense or, you know, some academic 00:05:06.639 --> 00:05:09.759 gobbledygook. It filters out all nuance. It makes 00:05:09.759 --> 00:05:11.519 the world look black and white when it's actually 00:05:11.519 --> 00:05:14.079 a million shades of gray. It's like we're being 00:05:14.079 --> 00:05:17.139 fed a diet of intellectual fast food because 00:05:17.139 --> 00:05:19.439 a three course meal takes too long to cook and 00:05:19.439 --> 00:05:23.420 eat. We feel full, but we aren't actually nourished. 00:05:23.439 --> 00:05:25.680 That is the perfect metaphor. And that leads 00:05:25.680 --> 00:05:28.410 right into the. Next big one on the list, bad 00:05:28.410 --> 00:05:31.370 news bias, or as it's more formally known, negativity 00:05:31.370 --> 00:05:33.689 bias. This is the whole if it bleeds, it leads 00:05:33.689 --> 00:05:36.990 thing, the doom scrolling phenomenon. It is, 00:05:36.990 --> 00:05:40.910 but it's deeper than just sad stories get clicks. 00:05:41.089 --> 00:05:44.089 It's about the fundamental asymmetry of how events 00:05:44.089 --> 00:05:46.610 happen in the world. Asymmetry? Yeah, the researchers 00:05:46.610 --> 00:05:48.810 point out something called big news bias. Think 00:05:48.810 --> 00:05:50.509 about how the world actually improves. Slowly, 00:05:50.850 --> 00:05:55.490 painfully, painfully slowly, inch by inch. Right. 00:05:55.920 --> 00:05:58.819 A new vaccine reduces child mortality by 0 .5 00:05:58.819 --> 00:06:01.939 % a year. Incomes rise by a few dollars a month. 00:06:02.480 --> 00:06:04.699 Poverty declines by a fraction of a percent. 00:06:04.879 --> 00:06:07.720 It's incremental. It's a trend, not an event. 00:06:07.779 --> 00:06:09.600 You can't take a picture of it. It doesn't break. 00:06:09.860 --> 00:06:12.899 But a plane crash, a terrorist attack, a bank 00:06:12.899 --> 00:06:15.879 collapse. Those are sudden, they're visual, and 00:06:15.879 --> 00:06:17.879 they are catastrophic. They're discrete events. 00:06:18.319 --> 00:06:21.959 So the news, by its very nature, covers the sudden 00:06:21.959 --> 00:06:25.279 disaster, but it misses. The slow miracle. So 00:06:25.279 --> 00:06:27.879 the news covers every single car crash, but it 00:06:27.879 --> 00:06:30.019 never covers the millions of cars that arrive 00:06:30.019 --> 00:06:32.259 safely at their destination. And this creates 00:06:32.259 --> 00:06:35.060 a completely distorted map of risk in our heads. 00:06:35.379 --> 00:06:37.279 The source material talks about the hierarchy 00:06:37.279 --> 00:06:40.220 of death. We are constantly bombarded with sensational 00:06:40.220 --> 00:06:43.160 rare events. We see a shark attack on the front 00:06:43.160 --> 00:06:45.139 page of every paper. Right. Makes you afraid 00:06:45.139 --> 00:06:47.060 to go in the water. Yeah, but you're more likely 00:06:47.060 --> 00:06:49.470 to be killed by a fallen coconut. We don't see 00:06:49.470 --> 00:06:51.370 front page stories about heart disease, which 00:06:51.370 --> 00:06:53.449 is killing thousands of times more people every 00:06:53.449 --> 00:06:55.910 single day because heart disease is boring. It's 00:06:55.910 --> 00:06:58.310 slow. It's statistical. This reminds me of that 00:06:58.310 --> 00:07:00.250 concept that's been talked about a lot lately, 00:07:00.430 --> 00:07:03.149 missing white woman syndrome. Yes, that is a 00:07:03.149 --> 00:07:06.269 perfect and tragic specific subset of this kind 00:07:06.269 --> 00:07:09.560 of sensationalism. The media has a template for 00:07:09.560 --> 00:07:12.620 what a victim looks like to generate maximum 00:07:12.620 --> 00:07:15.620 sympathy and viewership. And that template is 00:07:15.620 --> 00:07:18.399 usually young, white, attracted, and middle class. 00:07:18.680 --> 00:07:21.519 So when someone fitting that profile goes missing, 00:07:21.579 --> 00:07:24.339 it becomes wall -to -wall coverage. It's a national 00:07:24.339 --> 00:07:28.240 soap opera. Exactly. Meanwhile... thousands of 00:07:28.240 --> 00:07:30.579 people of color or men or older people or poor 00:07:30.579 --> 00:07:32.740 people go missing. And it's a footnote in the 00:07:32.740 --> 00:07:34.819 local paper. If it's covered at all. If at all. 00:07:35.180 --> 00:07:37.939 And the cumulative effect on you, the listener, 00:07:38.100 --> 00:07:40.420 is that you end up thinking the world is much, 00:07:40.420 --> 00:07:42.800 much more dangerous than it actually is, but 00:07:42.800 --> 00:07:45.019 you're also afraid of all the wrong things. You're 00:07:45.019 --> 00:07:46.980 afraid of being kidnapped by a stranger in a 00:07:46.980 --> 00:07:50.019 white van, which is incredibly rare, rather than 00:07:50.019 --> 00:07:51.939 being afraid of not wearing your seatbelt or 00:07:51.939 --> 00:07:54.060 eating too much processed food. It completely 00:07:54.060 --> 00:07:56.160 screws up our internal risk assessment software. 00:07:56.670 --> 00:07:58.589 OK, I want to distinguish between two others 00:07:58.589 --> 00:08:01.230 mentioned in the notes here, mainstream bias 00:08:01.230 --> 00:08:04.470 and normalcy bias. They sound really similar, 00:08:04.629 --> 00:08:07.290 but the distinction seems important. They're 00:08:07.290 --> 00:08:09.550 like siblings, but they behave very differently. 00:08:09.889 --> 00:08:12.269 Mainstream bias is essentially peer pressure 00:08:12.269 --> 00:08:15.389 for newsrooms. It's the herd mentality. Chasing 00:08:15.389 --> 00:08:17.930 the same story. Right. If The New York Times 00:08:17.930 --> 00:08:20.470 and CNN and The Wall Street Journal are all covering 00:08:20.470 --> 00:08:23.449 a particular story, every smaller outlet feels 00:08:23.449 --> 00:08:26.149 this immense pressure to cover it, too, just 00:08:26.149 --> 00:08:28.490 to prove that they are serious journalists. So 00:08:28.490 --> 00:08:30.509 you get this homogenization. Everybody is reporting 00:08:30.509 --> 00:08:32.730 the same angle on the same story because they're 00:08:32.730 --> 00:08:35.230 terrified of being the outlier, of missing the 00:08:35.230 --> 00:08:38.870 big one. Exactly. Which means important but less 00:08:38.870 --> 00:08:42.370 trendy stories get ignored. Now, normalcy bias 00:08:42.370 --> 00:08:44.639 is almost the opposite. or maybe it's better 00:08:44.639 --> 00:08:47.139 to call it a coping mechanism. It's the tendency 00:08:47.139 --> 00:08:50.179 to try to frame abnormal, catastrophic, or unprecedented 00:08:50.179 --> 00:08:53.259 events as if they are, well, normal. Give me 00:08:53.259 --> 00:08:54.980 a concrete example of that. What does that look 00:08:54.980 --> 00:08:57.460 like in practice? Think about the early days 00:08:57.460 --> 00:09:01.299 of a massive crisis like the onset of a pandemic 00:09:01.299 --> 00:09:04.899 or the rise of a truly radical political movement 00:09:04.899 --> 00:09:07.519 that doesn't play by the old rules. The media 00:09:07.519 --> 00:09:10.679 often tries to fit these square pegs into round 00:09:10.679 --> 00:09:13.440 holes. They'll cover historic breakdown of democratic 00:09:13.440 --> 00:09:16.860 norms as if it's just normal horse race politics. 00:09:17.179 --> 00:09:19.440 You know, who won the week? Or they'll cover 00:09:19.440 --> 00:09:22.480 a looming climate catastrophe as just a series 00:09:22.480 --> 00:09:25.740 of unusual weather events. A wacky heat wave 00:09:25.740 --> 00:09:29.580 here, a big flood there. Right. Because admitting 00:09:29.580 --> 00:09:31.840 that the entire system is breaking or that something 00:09:31.840 --> 00:09:35.080 genuinely unprecedented is happening is, well, 00:09:35.159 --> 00:09:37.309 it's too frightening. And it's also just too 00:09:37.309 --> 00:09:39.509 hard to categorize. So it's a failure of imagination 00:09:39.509 --> 00:09:41.669 as much as anything else. It is. Journalists 00:09:41.669 --> 00:09:43.990 are trained to report on the status quo. When 00:09:43.990 --> 00:09:46.389 the status quo itself is threatened, the instinct 00:09:46.389 --> 00:09:48.570 is to just pretend everything is still functioning 00:09:48.570 --> 00:09:50.870 according to the old rules. It's a psychological 00:09:50.870 --> 00:09:53.309 defense mechanism. That seems like it links directly 00:09:53.309 --> 00:09:55.450 to false balance. This is one I really struggle 00:09:55.450 --> 00:09:57.690 with because, you know, I want to be fair. I 00:09:57.690 --> 00:09:59.509 want to hear both sides. When does presenting 00:09:59.509 --> 00:10:03.039 both sides become a form of bias itself? It becomes 00:10:03.039 --> 00:10:06.620 a bias when the evidence is not 50 -50. The classic 00:10:06.620 --> 00:10:08.799 example is climate change, though we see it in 00:10:08.799 --> 00:10:10.840 other scientific debates too, like vaccines. 00:10:11.120 --> 00:10:14.100 Right. If 99 % of climate scientists agree that 00:10:14.100 --> 00:10:16.720 the Earth is warming due to human activity, and 00:10:16.720 --> 00:10:19.720 1 % of scientists, often funded by interested 00:10:19.720 --> 00:10:22.720 parties, disagree. And you put one of each on 00:10:22.720 --> 00:10:25.759 a split screen on cable news. Right. Visually, 00:10:25.820 --> 00:10:29.240 you have just created an equivalence. You have 00:10:29.240 --> 00:10:32.200 implicitly told the viewer this is a legitimate 00:10:32.200 --> 00:10:35.940 debate between two equal viewpoints, but factually 00:10:35.940 --> 00:10:38.399 you are lying to them. Because it's not a 50 00:10:38.399 --> 00:10:42.299 -50 debate in reality. It's a 991 consensus versus 00:10:42.299 --> 00:10:45.139 a fringe opinion. By giving them equal time, 00:10:45.279 --> 00:10:47.399 you are elevating the fringe opinion to the same 00:10:47.399 --> 00:10:49.879 level as the overwhelming scientific consensus. 00:10:50.139 --> 00:10:52.600 That isn't objectivity. It's a distortion of 00:10:52.600 --> 00:10:54.840 reality in the name of fairness. It's performative 00:10:54.840 --> 00:10:57.340 neutrality. Exactly. And finally, in our little 00:10:57.340 --> 00:11:00.220 taxonomy here, we have structural bias. This 00:11:00.220 --> 00:11:01.820 is the one that bores people at dinner parties 00:11:01.820 --> 00:11:04.259 but gets media scholars really excited. Why is 00:11:04.259 --> 00:11:06.580 that? Because it's completely non -ideological. 00:11:06.679 --> 00:11:09.000 It has nothing to do with left or right. It's 00:11:09.000 --> 00:11:11.399 all about logistics and routines. It's about 00:11:11.399 --> 00:11:14.860 the how of news gathering. Take the incumbency 00:11:14.860 --> 00:11:16.840 bonus, for example. This is why the president 00:11:16.840 --> 00:11:20.259 is on TV every single night. Yes. But it's not 00:11:20.259 --> 00:11:22.179 because the media loves the president no matter 00:11:22.179 --> 00:11:24.960 who it is. It's because the president is a content 00:11:24.960 --> 00:11:27.779 machine. A content machine. He has a press secretary. 00:11:28.059 --> 00:11:31.120 He has a dedicated briefing room. He has a giant 00:11:31.120 --> 00:11:34.240 staff that produces policy papers and announcements 00:11:34.240 --> 00:11:37.899 every single day. It is easy to cover the president. 00:11:38.059 --> 00:11:40.840 It is logistically simple. And it's hard to cover 00:11:40.840 --> 00:11:43.220 a challenger who is, you know, traveling around 00:11:43.220 --> 00:11:45.840 in a bus in Iowa. It's incredibly hard. You have 00:11:45.840 --> 00:11:48.620 to send a reporter, a camera crew. It's expensive 00:11:48.620 --> 00:11:51.440 and difficult. So the bias is toward convenience. 00:11:52.120 --> 00:11:54.259 It's the path of least resistance. Journalists 00:11:54.259 --> 00:11:56.519 are overworked and underpaid like a lot of people. 00:11:56.820 --> 00:11:59.500 That's a huge part of it. If one source like 00:11:59.500 --> 00:12:01.419 the White House or the police department spoon 00:12:01.419 --> 00:12:04.240 feeds them a prepackaged story and another source 00:12:04.240 --> 00:12:06.559 makes them dig for it, guess which one gets the 00:12:06.559 --> 00:12:09.139 coverage? The spoon feeder wins every time. OK, 00:12:09.220 --> 00:12:11.519 so we've established that the whole machine is 00:12:11.519 --> 00:12:15.059 tilted by laziness, by the need for speed, by 00:12:15.059 --> 00:12:17.799 our psychological wiring for sensationalism. 00:12:18.200 --> 00:12:19.779 But now as we get to the part of the research 00:12:19.779 --> 00:12:23.320 that really surprised me. The economics. Because 00:12:23.320 --> 00:12:26.080 usually when people say media bias, they follow 00:12:26.080 --> 00:12:28.980 it with follow the money. And they should. They're 00:12:28.980 --> 00:12:31.620 not wrong to think that. But they often follow 00:12:31.620 --> 00:12:33.440 the money to the wrong place. What do you mean? 00:12:33.620 --> 00:12:36.320 They assume the money leads to some shadowy billionaire 00:12:36.320 --> 00:12:39.559 in a volcano lair, you know, paying for propaganda. 00:12:39.899 --> 00:12:42.320 And sometimes it does. That happens. But much 00:12:42.320 --> 00:12:45.769 more often it leads to. Well, to ski resorts. 00:12:45.769 --> 00:12:47.730 I have to say, when I saw ski resorts in the 00:12:47.730 --> 00:12:49.649 outline for an episode on political polarization, 00:12:50.049 --> 00:12:52.450 I was sure there was a typo. It's honestly my 00:12:52.450 --> 00:12:54.950 favorite study in this entire stack. It was conducted 00:12:54.950 --> 00:12:57.789 by Zinman and Zitzowitz. They wanted to test 00:12:57.789 --> 00:13:00.629 a concept called supply -driven bias. Okay, explain 00:13:00.629 --> 00:13:03.049 that concept first. What is supply -driven bias? 00:13:03.309 --> 00:13:05.870 Supply -driven bias is when the person or organization 00:13:05.870 --> 00:13:09.149 providing the information, the supplier, distorts 00:13:09.149 --> 00:13:11.049 it on purpose to manipulate the consumer into 00:13:11.049 --> 00:13:13.450 doing something specific. It's a distortion for 00:13:13.450 --> 00:13:15.970 profit. Okay, that makes sense. So back to the 00:13:15.970 --> 00:13:19.049 ski resorts. So Zinman and Zitzewitz collected 00:13:19.049 --> 00:13:20.950 the daily snow reports that are issued by the 00:13:20.950 --> 00:13:23.269 ski resorts themselves. You know, six inches 00:13:23.269 --> 00:13:25.289 of fresh powder overnight. And they compared 00:13:25.289 --> 00:13:27.809 those reports to the official government weather 00:13:27.809 --> 00:13:30.690 data from meteorological stations at the exact 00:13:30.690 --> 00:13:32.970 same locations. And I'm guessing the resorts 00:13:32.970 --> 00:13:36.850 were a little optimistic. Significantly so. They 00:13:36.850 --> 00:13:39.309 found a clear, measurable inflation of snowfall 00:13:39.309 --> 00:13:42.129 amounts. But here is the really clever part of 00:13:42.129 --> 00:13:44.830 the study, the nuance. Go on. They were most 00:13:44.830 --> 00:13:47.750 optimistic on weekends. Of course. Of course 00:13:47.750 --> 00:13:49.570 they were. That's when the customers can actually 00:13:49.570 --> 00:13:52.690 come. Exactly. On a Tuesday morning, the report 00:13:52.690 --> 00:13:55.049 might be pretty honest. But on a Saturday morning, 00:13:55.230 --> 00:13:56.870 when people in the city are deciding whether 00:13:56.870 --> 00:13:59.870 to make the drive up, that two inches of slush 00:13:59.870 --> 00:14:03.009 miraculously becomes four to six inches of fresh 00:14:03.009 --> 00:14:05.029 packed powder. Wait, hold on a second. Is that 00:14:05.029 --> 00:14:07.330 really bias or is that just, you know, marketing? 00:14:07.429 --> 00:14:09.690 If a used car dealer tells me a 20 -year -old 00:14:09.690 --> 00:14:12.049 Honda is pristine, I expect him to be lying. 00:14:12.759 --> 00:14:15.799 Why do we classify this as media bias? That's 00:14:15.799 --> 00:14:17.980 a great question. It's because in many of these 00:14:17.980 --> 00:14:20.539 ski towns, these reports are treated as news. 00:14:20.740 --> 00:14:23.039 They're distributed through channels that people 00:14:23.039 --> 00:14:25.700 trust for factual information, local radio, websites, 00:14:25.980 --> 00:14:28.720 newspapers. And it perfectly illustrates the 00:14:28.720 --> 00:14:32.100 mechanism on a small scale. The supplier, the 00:14:32.100 --> 00:14:35.000 resort, has a clear financial incentive to distort 00:14:35.000 --> 00:14:38.000 the truth. And their goal is to change the behavior 00:14:38.000 --> 00:14:40.940 of the demander, the skier, to get them to buy 00:14:40.940 --> 00:14:43.139 a lift ticket. OK, I see that. It's a perfect, 00:14:43.279 --> 00:14:46.659 clean, small, stale model of a much larger and 00:14:46.659 --> 00:14:48.860 more complicated problem. Right. Now scale that 00:14:48.860 --> 00:14:51.980 up. Think of the theory posed way back in 1988 00:14:51.980 --> 00:14:55.059 by Noam Chomsky and Edward Herman in their book 00:14:55.059 --> 00:14:57.799 Manufacturing Consent. They argued that mass 00:14:57.799 --> 00:15:00.559 media is essentially a giant ski resort for corporate 00:15:00.559 --> 00:15:02.460 America. They called it the propaganda model, 00:15:02.639 --> 00:15:05.179 right, with the five filters? Yes, exactly. One 00:15:05.179 --> 00:15:07.740 of their key filters was advertising. They argued 00:15:07.740 --> 00:15:09.840 that because mainstream media relies so heavily 00:15:09.840 --> 00:15:12.440 on advertising revenue, They simply cannot afford 00:15:12.440 --> 00:15:15.120 to seriously offend their advertisers. So if 00:15:15.120 --> 00:15:17.320 a network is funded primarily by pharmaceutical 00:15:17.320 --> 00:15:20.860 ads, are they going to run a hard -hitting, multi 00:15:20.860 --> 00:15:23.519 -part investigative series on price gouging in 00:15:23.519 --> 00:15:26.120 the pharmaceutical industry? Probably not. Or 00:15:26.120 --> 00:15:28.840 if they do, you can bet it'll be watered down. 00:15:29.019 --> 00:15:31.519 Herman and Chomsky called this flack. Flack? 00:15:31.519 --> 00:15:34.879 Yeah. If you step out of line and report something 00:15:34.879 --> 00:15:36.740 that threatens powerful corporate interests, 00:15:37.000 --> 00:15:39.769 those interests generate flack. They pull their 00:15:39.769 --> 00:15:41.870 funding. Their executives complain to the board. 00:15:41.990 --> 00:15:45.149 They fund PR campaigns against you. So the supply 00:15:45.149 --> 00:15:48.730 of news is filtered to protect the economic interests 00:15:48.730 --> 00:15:51.149 of the owners and their advertisers. That makes 00:15:51.149 --> 00:15:53.470 total sense. That's the cynical view we're all 00:15:53.470 --> 00:15:55.409 kind of used to. The corporations are controlling 00:15:55.409 --> 00:15:58.870 our minds. But then the research pivots, and 00:15:58.870 --> 00:16:01.370 it points the finger right back at us. This is 00:16:01.370 --> 00:16:03.269 the demand -driven bias. And this is the part 00:16:03.269 --> 00:16:04.929 of the conversation that gets really uncomfortable 00:16:04.929 --> 00:16:07.070 for a lot of people. Because it's easier to blame 00:16:07.070 --> 00:16:10.120 the man behind the curtain. Much easier. Demand 00:16:10.120 --> 00:16:12.639 -driven bias argues that, in many cases, media 00:16:12.639 --> 00:16:15.080 organizations aren't manipulating us. They are 00:16:15.080 --> 00:16:17.559 catering to us. They are businesses. They sell 00:16:17.559 --> 00:16:19.659 a product. And if the product that's in high 00:16:19.659 --> 00:16:22.980 demand is news that makes you feel good and smart 00:16:22.980 --> 00:16:25.100 about your political team, then they will sell 00:16:25.100 --> 00:16:27.559 that product by the truckload. The customer is 00:16:27.559 --> 00:16:29.799 always right, even when the customer is factually 00:16:29.799 --> 00:16:32.600 wrong. Exactly. And to prove this isn't just 00:16:32.600 --> 00:16:35.080 some modern internet phenomenon, the research 00:16:35.080 --> 00:16:38.240 takes us all the way back to the 1890s. The Gilded 00:16:38.240 --> 00:16:39.980 Age. This is the New York Times weather study 00:16:39.980 --> 00:16:42.080 you meant. Yes. Researchers Raymond and Taylor 00:16:42.080 --> 00:16:44.159 did this incredible deep dive into the weather 00:16:44.159 --> 00:16:46.419 forecasts printed in the New York Times between 00:16:46.419 --> 00:16:50.299 1890 and 1899. Now, you have to think about that 00:16:50.299 --> 00:16:53.500 era. No TV, no radio, no internet. The newspaper 00:16:53.500 --> 00:16:57.100 was everything. And baseball, the New York Giants, 00:16:57.159 --> 00:17:00.320 was a citywide obsession. So what did they find? 00:17:00.399 --> 00:17:03.139 What was the anomaly? They found a statistically 00:17:03.139 --> 00:17:06.059 significant anomaly. On days when the Giants 00:17:06.059 --> 00:17:07.640 were scheduled to play a home game in Manhattan, 00:17:07.900 --> 00:17:09.880 the New York Times was significantly more likely 00:17:09.880 --> 00:17:12.279 to predict sunny weather. You're kidding me. 00:17:12.299 --> 00:17:14.240 Come on. They were fudging the weather report 00:17:14.240 --> 00:17:17.140 for a baseball game. The data is clear. If the 00:17:17.140 --> 00:17:19.000 team was playing away or if there was no game 00:17:19.000 --> 00:17:21.380 scheduled, the forecast was much more accurate. 00:17:21.420 --> 00:17:23.779 It lined up with the actual weather. But if there 00:17:23.779 --> 00:17:26.559 was a home game on the schedule, sunny skies 00:17:26.559 --> 00:17:31.079 ahead. That is absolutely wild. But why? Was 00:17:31.079 --> 00:17:33.359 the Times being paid off by the Giants? Was there 00:17:33.359 --> 00:17:35.819 some kind of conspiracy? There's no evidence 00:17:35.819 --> 00:17:37.859 of that at all. The implication is much more 00:17:37.859 --> 00:17:40.779 subtle and I think much more powerful. OK. The 00:17:40.779 --> 00:17:42.700 paper knew its readers wanted to go to the game. 00:17:42.839 --> 00:17:45.920 The readers were hoping for Sun. So the paper, 00:17:46.019 --> 00:17:48.759 perhaps unconsciously or perhaps just to sell 00:17:48.759 --> 00:17:51.440 more copies to hopeful fans, leaned into that 00:17:51.440 --> 00:17:53.779 hope. They gave the people the forecast they 00:17:53.779 --> 00:17:55.960 wanted, not the forecast that was scientifically 00:17:55.960 --> 00:17:58.839 probable. That is profoundly disturbing. I mean. 00:17:59.160 --> 00:18:01.480 More than the ski resorts, because weather is 00:18:01.480 --> 00:18:03.680 supposed to be objective. It's physics. It's 00:18:03.680 --> 00:18:06.380 barometric pressure. If you can bias the weather 00:18:06.380 --> 00:18:08.539 report just to make people happy about a baseball 00:18:08.539 --> 00:18:11.440 game, you can bias literally anything. Exactly. 00:18:11.660 --> 00:18:13.880 And this brings us right into the modern era. 00:18:14.000 --> 00:18:17.380 Two economists, Melanathan and Schleifer at Harvard, 00:18:17.500 --> 00:18:19.839 built a behavioral model around this exact idea 00:18:19.839 --> 00:18:22.500 back in 2005. They basically proved mathematically 00:18:22.500 --> 00:18:24.940 that news providers can maximize their profit 00:18:24.940 --> 00:18:27.279 by confirming their viewers' existing beliefs. 00:18:27.819 --> 00:18:30.420 So if I'm a liberal, I will pay more for news 00:18:30.420 --> 00:18:32.720 that tells me liberals are smart and conservatives 00:18:32.720 --> 00:18:35.420 are evil. And if a newspaper tries to be perfectly 00:18:35.420 --> 00:18:38.039 neutral, I'll just stop buying it. You'll switch 00:18:38.039 --> 00:18:39.720 to the competitor who tells you what you want 00:18:39.720 --> 00:18:42.579 to hear. In a competitive market, neutrality 00:18:42.579 --> 00:18:45.619 can be bad for business. You're selling validation, 00:18:45.940 --> 00:18:48.480 not just information. There was also a study 00:18:48.480 --> 00:18:50.900 on Weibo, which is like the Chinese version of 00:18:50.900 --> 00:18:53.670 Twitter, that backed this up, right? Yes. Dong, 00:18:53.930 --> 00:18:56.829 Ren, and Nickerson tracked millions of posts 00:18:56.829 --> 00:19:00.450 related to stock market news. They found, unequivocally, 00:19:00.470 --> 00:19:02.869 that news which aligned with a user's existing 00:19:02.869 --> 00:19:05.470 belief, whether they were bullish or bearish 00:19:05.470 --> 00:19:07.970 on a particular stock, got significantly more 00:19:07.970 --> 00:19:10.309 traction, more likes, more shares. We don't want 00:19:10.309 --> 00:19:13.130 facts. We want validation. It seems to be a very 00:19:13.130 --> 00:19:15.769 powerful drive. So. Okay, let me sum this up. 00:19:15.970 --> 00:19:19.569 We have the supply side corporations and advertisers 00:19:19.569 --> 00:19:22.309 protecting their interests. And we have the demand 00:19:22.309 --> 00:19:25.369 side, us, the audience, begging to be lied to 00:19:25.369 --> 00:19:27.849 so we can feel smart and validated. That is a 00:19:27.849 --> 00:19:30.250 truly toxic cocktail. And we haven't even added 00:19:30.250 --> 00:19:33.130 the accelerant yet. Social media. The great accelerant. 00:19:33.250 --> 00:19:36.049 We have to talk about the algorithm. Because 00:19:36.049 --> 00:19:39.150 if newspapers were biased by profit motives in 00:19:39.150 --> 00:19:41.430 the 1890s, at least there were humans in the 00:19:41.430 --> 00:19:45.210 loop. Editors, journalists. They had standards, 00:19:45.309 --> 00:19:47.410 hopefully. Now we have machines deciding what 00:19:47.410 --> 00:19:49.369 we see. And the machine has a very, very simple 00:19:49.369 --> 00:19:52.269 goal. Engagement. That's it. It doesn't care 00:19:52.269 --> 00:19:53.890 if a story is true. It doesn't care if it's good 00:19:53.890 --> 00:19:55.470 for democracy. It doesn't care if it's kind. 00:19:55.789 --> 00:19:58.009 It cares if you look at it and if you share it. 00:19:58.269 --> 00:20:00.349 And as we already discussed with bad news bias, 00:20:00.569 --> 00:20:02.730 the stuff we look at, the stuff we share is usually 00:20:02.730 --> 00:20:05.150 the stuff that makes us angry or afraid. Anger 00:20:05.150 --> 00:20:07.269 is the highest engagement emotion online. It 00:20:07.269 --> 00:20:09.250 generates more clicks, more comments, more shares 00:20:09.250 --> 00:20:11.750 than anything else. So the algorithm learns to 00:20:11.750 --> 00:20:14.950 prioritize content that triggers outrage. It's 00:20:14.950 --> 00:20:17.930 automated demand driven bias on steroids. Now, 00:20:17.990 --> 00:20:20.430 there's a study here from Nature Communications 00:20:20.430 --> 00:20:24.329 in 2021 about echo chambers and a concept they 00:20:24.329 --> 00:20:27.779 called drifting. This one seemed a bit more nuanced 00:20:27.779 --> 00:20:30.599 than the usual, everyone is in a bubble take 00:20:30.599 --> 00:20:33.039 we always hear. It is. It's really fascinating. 00:20:33.119 --> 00:20:35.720 They looked at how users drift ideologically 00:20:35.720 --> 00:20:38.420 based on the content they see, and they found 00:20:38.420 --> 00:20:41.160 a really interesting asymmetry. Asymmetry. You 00:20:41.160 --> 00:20:42.799 mean the left and the right behave differently 00:20:42.799 --> 00:20:45.599 in these online spaces? According to this particular 00:20:45.599 --> 00:20:48.500 study, yes. They found that right -leaning users 00:20:48.500 --> 00:20:51.509 tend to stay on the right. They are effectively 00:20:51.509 --> 00:20:55.049 locked in to their ecosystem and they are exposed 00:20:55.049 --> 00:20:57.210 to a higher proportion of what the researchers 00:20:57.210 --> 00:21:00.789 labeled low credibility content. So what we would 00:21:00.789 --> 00:21:04.190 colloquially call fake news or just heavily spun 00:21:04.190 --> 00:21:06.809 news. It's a mix of both. Hyper -partisan sources, 00:21:07.069 --> 00:21:08.890 conspiracy sites, things that don't adhere to 00:21:08.890 --> 00:21:11.009 basic journalistic standards. But here's the 00:21:11.009 --> 00:21:12.589 really interesting finding. It was about the 00:21:12.589 --> 00:21:15.069 left. When left -leaning users were exposed to 00:21:15.069 --> 00:21:17.869 opposing views. So when the algorithm showed 00:21:17.869 --> 00:21:20.670 them conservative content, they didn't just dig 00:21:20.670 --> 00:21:23.250 in deeper to their own side. They actually tended 00:21:23.250 --> 00:21:25.829 to drift toward the center. They moderated their 00:21:25.829 --> 00:21:29.410 views. That is fascinating. So exposure to the 00:21:29.410 --> 00:21:31.950 other side actually worked for one group in terms 00:21:31.950 --> 00:21:34.470 of moderation, but not the other. In that specific 00:21:34.470 --> 00:21:38.009 data set, yes. It's a complex finding, and there 00:21:38.009 --> 00:21:40.490 are a lot of potential reasons for it, but it 00:21:40.490 --> 00:21:42.549 suggests that the echo chamber might have different 00:21:42.549 --> 00:21:44.470 kinds of walls depending where you stand politically. 00:21:44.690 --> 00:21:47.829 Wow. Now, speaking of left and right on social 00:21:47.829 --> 00:21:50.609 media, we have to address the elephant in the 00:21:50.609 --> 00:21:53.289 room. If I go on Twitter or X, whatever we're 00:21:53.289 --> 00:21:55.390 calling it this week right now, I will see a 00:21:55.390 --> 00:21:58.430 thousand posts saying the algorithm is censoring 00:21:58.430 --> 00:22:01.750 conservatives. Big tech is woke. They are suppressing 00:22:01.750 --> 00:22:04.089 the truth. It's a massive narrative. It's become 00:22:04.089 --> 00:22:06.369 a core part of the modern conservative identity. 00:22:06.549 --> 00:22:08.890 This idea that they are being persecuted and 00:22:08.890 --> 00:22:11.650 silenced by a biased Silicon Valley. Is it true? 00:22:11.710 --> 00:22:13.589 What does the data say? Well, let's look at the 00:22:13.589 --> 00:22:16.630 data, not the anecdotes. We have a major 2021 00:22:16.630 --> 00:22:19.869 report from the NYU Stern Center for Business 00:22:19.869 --> 00:22:22.849 and Human Rights and another big study from 2022 00:22:22.849 --> 00:22:26.029 published in PNAS, the Proceedings of the National 00:22:26.029 --> 00:22:29.309 Academy of Sciences. These are serious, heavy 00:22:29.309 --> 00:22:31.369 hitting analyses. OK, so what did they find? 00:22:31.410 --> 00:22:34.130 Did they find a systemic shadow ban on conservatives? 00:22:34.569 --> 00:22:36.609 They found the exact opposite. The opposite. 00:22:36.940 --> 00:22:40.180 The PNAS study, for example, looked at algorithmic 00:22:40.180 --> 00:22:42.039 amplification across seven different countries. 00:22:42.200 --> 00:22:44.720 In six of those countries, including the United 00:22:44.720 --> 00:22:47.759 States, the algorithm favored the political right. 00:22:48.289 --> 00:22:51.049 It amplified right -wing content more than left 00:22:51.049 --> 00:22:52.829 -wing content. Wait, hang on. Let me process 00:22:52.829 --> 00:22:54.569 that. You're telling me that the data shows the 00:22:54.569 --> 00:22:57.170 algorithm boosts right -wing content more than 00:22:57.170 --> 00:23:00.029 left -wing content? Yes. The data consistently 00:23:00.029 --> 00:23:02.410 shows that right -wing voices are often dominant 00:23:02.410 --> 00:23:04.990 in terms of reach, engagement, and amplification 00:23:04.990 --> 00:23:07.490 on these platforms. But why? If the stereotype 00:23:07.490 --> 00:23:09.750 is that the engineers are all liberals in San 00:23:09.750 --> 00:23:12.650 Francisco, why would their machine be boosting 00:23:12.650 --> 00:23:15.450 the... Right. Go back to the core mechanic. The 00:23:15.450 --> 00:23:16.849 algorithm doesn't have a political ideology. 00:23:17.609 --> 00:23:20.529 It has an appetite. It loves engagement, loves 00:23:20.529 --> 00:23:23.230 high emotion. It loves us versus them framing. 00:23:23.369 --> 00:23:25.769 And a lot of populist political communication, 00:23:25.950 --> 00:23:29.190 which is currently more prevalent on the political 00:23:29.190 --> 00:23:31.869 right in many of these countries, tends to be 00:23:31.869 --> 00:23:34.869 more emotive, more sensational and more high 00:23:34.869 --> 00:23:37.849 arousal. It's better food for the machine. So 00:23:37.849 --> 00:23:39.710 the algorithm isn't ideologically conservative. 00:23:39.849 --> 00:23:42.549 It's just, for lack of a better word, thirsty 00:23:42.549 --> 00:23:45.519 for drama. And the right is currently providing 00:23:45.519 --> 00:23:48.099 better, more engaging drama for it to feed on. 00:23:48.200 --> 00:23:50.119 That is a perfect way to put it. It's not a conspiracy. 00:23:50.299 --> 00:23:52.619 It's a feedback loop. The claim of being censored 00:23:52.619 --> 00:23:55.619 is itself a high engagement, emotionally charged 00:23:55.619 --> 00:23:57.940 narrative that the algorithm loves. That is a 00:23:57.940 --> 00:24:01.339 huge aha moment. The victim complex is actually 00:24:01.339 --> 00:24:03.880 a very effective marketing strategy that the 00:24:03.880 --> 00:24:07.359 system itself rewards. In a way, yes. The data 00:24:07.359 --> 00:24:10.099 suggests the machine love the very content that 00:24:10.099 --> 00:24:12.099 people claim is being suppressed. There's one 00:24:12.099 --> 00:24:14.079 last piece of this social media puzzle, and it 00:24:14.079 --> 00:24:16.039 might be the most insidious, the comment section, 00:24:16.259 --> 00:24:18.480 the bottom half of the Internet. The place where 00:24:18.480 --> 00:24:21.180 hope goes to die. But it turns out it also kills 00:24:21.180 --> 00:24:23.700 our perception of neutrality. This is something 00:24:23.700 --> 00:24:26.519 called the hostile media effect and the Gearheart 00:24:26.519 --> 00:24:30.259 study. This study is genuinely frightening. Gearheart 00:24:30.259 --> 00:24:32.200 and their team found that a viewer's perception 00:24:32.200 --> 00:24:35.059 of bias in a news article changes dramatically. 00:24:35.680 --> 00:24:37.779 based on the user comments they read below it. 00:24:37.839 --> 00:24:40.319 So I can read a perfectly neutral, just -the 00:24:40.319 --> 00:24:42.799 -facts news story, but if the comments below 00:24:42.799 --> 00:24:45.339 it are all screaming, this is fake news, or the 00:24:45.339 --> 00:24:48.359 author is a communist, I will start to believe 00:24:48.359 --> 00:24:51.619 the article itself is biased. Yes, you subconsciously 00:24:51.619 --> 00:24:54.440 project the hostility and bias of the comments 00:24:54.440 --> 00:24:57.500 section onto the journalism itself. The room 00:24:57.500 --> 00:25:00.779 biases the content. The conversation around the 00:25:00.779 --> 00:25:03.099 article becomes the article in your mind. That 00:25:03.099 --> 00:25:05.240 explains so much about why everything feels so 00:25:05.240 --> 00:25:07.759 polarized. Even if the reporting itself is straight 00:25:07.759 --> 00:25:09.660 down the middle, the conversation around it is 00:25:09.660 --> 00:25:11.500 almost always toxic, so the whole experience 00:25:11.500 --> 00:25:14.299 feels toxic. Exactly. We aren't consuming news 00:25:14.299 --> 00:25:16.799 in a vacuum anymore. We're consuming it in the 00:25:16.799 --> 00:25:19.720 middle of a riot. Okay, we've covered the taxonomy 00:25:19.720 --> 00:25:22.519 of invisible biases, the economics of supply 00:25:22.519 --> 00:25:26.180 and demand, and the algorithmic accelerant. It 00:25:26.180 --> 00:25:28.619 really feels like we are living in a uniquely 00:25:28.619 --> 00:25:32.160 terrible and confusing time. But I want to step 00:25:32.160 --> 00:25:37.259 back. Is any of this new? Or have we always been 00:25:37.259 --> 00:25:39.460 this messy? Oh, we have always been this messy. 00:25:39.579 --> 00:25:41.559 If anything, and this is going to sound crazy, 00:25:41.700 --> 00:25:44.099 we might be slightly less messy than we used 00:25:44.099 --> 00:25:46.039 to be, believe it or not. Prove it. Take me back 00:25:46.039 --> 00:25:49.559 in time. Let's start in 1798. The ink is barely 00:25:49.559 --> 00:25:52.059 dry on the Constitution. The paint is still wet 00:25:52.059 --> 00:25:54.200 on the White House. And what does the U .S. Congress 00:25:54.200 --> 00:25:57.130 do? They passed the Alien and Sedition Acts. 00:25:57.250 --> 00:25:59.630 I remember this from high school history, vaguely. 00:25:59.670 --> 00:26:01.289 This was under John Adams, right? That's right. 00:26:01.410 --> 00:26:04.289 And these acts made it illegal, a federal crime 00:26:04.289 --> 00:26:07.750 punishable by prison, to publish a false, scandalous, 00:26:07.750 --> 00:26:09.750 or malicious writing against the government. 00:26:09.910 --> 00:26:12.430 So literally state -enforced media bias. You 00:26:12.430 --> 00:26:14.690 could not legally criticize the president. Right. 00:26:14.769 --> 00:26:17.269 If you think cancel culture is bad today, imagine 00:26:17.269 --> 00:26:19.390 being thrown in a federal dungeon for writing 00:26:19.390 --> 00:26:22.690 an op -ed calling the president fat or incompetent. 00:26:22.690 --> 00:26:25.200 That was the reality. Okay, point taken. That's 00:26:25.200 --> 00:26:28.220 pretty extreme. What about the Civil War? Abraham 00:26:28.220 --> 00:26:31.900 Lincoln. We think of them as Honest Abe, a saint 00:26:31.900 --> 00:26:35.019 of American democracy. But during the war, he 00:26:35.019 --> 00:26:37.500 ordered newspapers in border states to be closed 00:26:37.500 --> 00:26:39.079 by the military because they were sympathetic 00:26:39.079 --> 00:26:42.019 to the South. He just shut down the press. With 00:26:42.019 --> 00:26:44.539 soldiers. He arrested their editors. He had printing 00:26:44.539 --> 00:26:47.480 presses destroyed. He suspended the writ of habeas 00:26:47.480 --> 00:26:49.579 corpus for people accused of disloyal speech. 00:26:49.880 --> 00:26:52.599 He viewed them as a direct threat to national 00:26:52.599 --> 00:26:55.940 security, and he acted accordingly. It puts our 00:26:55.940 --> 00:26:58.779 current debates about media responsibility into 00:26:58.779 --> 00:27:00.829 a very different perspective. And then we have 00:27:00.829 --> 00:27:03.809 the Vietnam era, the nattering nabobs of negativism. 00:27:03.829 --> 00:27:06.950 1970, Vice President Spiro Agnew, he goes on 00:27:06.950 --> 00:27:09.069 national television and delivers this absolutely 00:27:09.069 --> 00:27:11.329 blistering attack on the news media, calling 00:27:11.329 --> 00:27:13.829 them these nattering nabobs of negativism. What 00:27:13.829 --> 00:27:16.130 a phrase. He accused them of undermining the 00:27:16.130 --> 00:27:18.990 war effort in Vietnam, of being out of touch 00:27:18.990 --> 00:27:21.529 coastal elites, of being biased against real 00:27:21.529 --> 00:27:24.190 Americans in the heartland. That sounds incredibly, 00:27:24.329 --> 00:27:27.170 almost word for word familiar. It's the exact 00:27:27.170 --> 00:27:29.779 same political script we hear today. Attack the 00:27:29.779 --> 00:27:33.140 messenger. Delegitimize the institution so that 00:27:33.140 --> 00:27:36.220 people will only trust the administration. It's 00:27:36.220 --> 00:27:38.380 a political tactic that is at least 50 years 00:27:38.380 --> 00:27:41.099 old in its modern form and much older in general. 00:27:41.279 --> 00:27:43.200 But there's one historical example you brought 00:27:43.200 --> 00:27:45.460 up that isn't really about politics. And it's 00:27:45.460 --> 00:27:48.279 the one that really chills me to the bone. The 00:27:48.279 --> 00:27:51.599 satanic panic of the 1980s. This is a critical 00:27:51.599 --> 00:27:53.890 case study. Because it shows what happens when 00:27:53.890 --> 00:27:56.049 the media gets completely swept up in a moral 00:27:56.049 --> 00:27:58.410 panic. It's not about left versus right. It's 00:27:58.410 --> 00:28:00.589 about something much deeper. For the younger 00:28:00.589 --> 00:28:02.750 listeners who might not know, set the scene. 00:28:02.829 --> 00:28:05.650 What was this? The 1980s. It starts with a case 00:28:05.650 --> 00:28:07.549 called the McMartin Preschool Trial in California. 00:28:08.180 --> 00:28:11.839 And from there, it explodes into a sudden nationwide 00:28:11.839 --> 00:28:14.640 hysteria that there was a secret underground 00:28:14.640 --> 00:28:17.720 network of Satanists abusing children in daycares 00:28:17.720 --> 00:28:19.900 across America. And we aren't talking about one 00:28:19.900 --> 00:28:22.019 or two bad teachers. We are talking about allegations 00:28:22.019 --> 00:28:24.819 of... We are talking about allegations of underground 00:28:24.819 --> 00:28:28.839 tunnels beneath schools, of ritual animal and 00:28:28.839 --> 00:28:31.519 human sacrifices, of witches flying, of children 00:28:31.519 --> 00:28:34.019 being forced to drink blood. I mean, truly absurd, 00:28:34.339 --> 00:28:37.720 cartoonishly evil stuff. And the media. They 00:28:37.720 --> 00:28:40.420 reported this as if it were real. They devoured 00:28:40.420 --> 00:28:43.539 it. Geraldo Rivera ran a primetime special called 00:28:43.539 --> 00:28:46.160 Devil Worship, exposing Satan's underground. 00:28:46.579 --> 00:28:49.480 Talk shows like Oprah, local news stations across 00:28:49.480 --> 00:28:51.380 the country. They reported these accusations 00:28:51.380 --> 00:28:53.559 as if they were credible. They put children on 00:28:53.559 --> 00:28:55.440 camera who had been coached by therapists to 00:28:55.440 --> 00:28:57.380 tell these wild stories. And it was all fake. 00:28:57.640 --> 00:28:59.559 Completely, totally fabricated. There were no 00:28:59.559 --> 00:29:02.519 tunnels. There was no cult. It was a mass delusion, 00:29:02.619 --> 00:29:05.039 fueled by anxieties about working mothers and 00:29:05.039 --> 00:29:07.859 social change. But lives were completely ruined. 00:29:08.000 --> 00:29:09.920 People went to prison for decades. Schools were 00:29:09.920 --> 00:29:12.039 shut down. The emotional damage was immense. 00:29:12.380 --> 00:29:15.180 So why did the media fail so catastrophically 00:29:15.180 --> 00:29:17.420 there? That's not left versus right. What bias 00:29:17.420 --> 00:29:19.900 was at play? It was a perfect storm of biases. 00:29:20.079 --> 00:29:22.619 It was sensationalism, for one. But it was also 00:29:22.619 --> 00:29:24.799 a kind of reverse normalcy bias. Let's call it 00:29:24.799 --> 00:29:28.240 credulity. The accusation was so horrific, harming 00:29:28.240 --> 00:29:31.700 children, that journalists suspended all of their 00:29:31.700 --> 00:29:33.779 professional skepticism. They didn't want to 00:29:33.779 --> 00:29:35.779 be the one to seem like they were defending child 00:29:35.779 --> 00:29:37.640 abusers. Exactly. They didn't want to be the 00:29:37.640 --> 00:29:39.539 one to say, hang on, I think this four -year 00:29:39.539 --> 00:29:41.599 -old story about flying witches might not be 00:29:41.599 --> 00:29:44.500 true. So they became amplifiers of the panic 00:29:44.500 --> 00:29:47.700 instead of investigators of the facts. And the 00:29:47.700 --> 00:29:49.819 expert insight here in the research is about 00:29:49.819 --> 00:29:53.500 the feedback loop of fear. Yes. The media reports 00:29:53.500 --> 00:29:56.059 the fear. The police see the reports and feel 00:29:56.059 --> 00:29:58.920 pressure to make more arrests. The public sees 00:29:58.920 --> 00:30:01.839 the arrests and panics even more. The media then 00:30:01.839 --> 00:30:03.940 reports on the growing panic. It becomes this 00:30:03.940 --> 00:30:06.380 terrible, self -sustaining engine of destruction 00:30:06.380 --> 00:30:09.039 that's almost impossible to turn off. It reminds 00:30:09.039 --> 00:30:11.440 me of how media often treats minority religions 00:30:11.440 --> 00:30:13.740 in general, the whole cult narrative. Stuart 00:30:13.740 --> 00:30:16.440 A. Wright wrote about this phenomenon. He pointed 00:30:16.440 --> 00:30:18.619 out that journalists usually lack the resources, 00:30:18.920 --> 00:30:22.019 the time or the theological background to understand 00:30:22.019 --> 00:30:26.819 a new or minority religion. So if a weird group 00:30:26.819 --> 00:30:28.799 gets accused of something, they just run with 00:30:28.799 --> 00:30:31.740 the cult accused of X, Y, Z headline. It's what 00:30:31.740 --> 00:30:34.190 he calls front end reporting. They report the 00:30:34.190 --> 00:30:36.710 dramatic police raid, but they never report the 00:30:36.710 --> 00:30:38.769 exoneration two years later when all the charges 00:30:38.769 --> 00:30:41.069 are dropped. So the damage is done on page one 00:30:41.069 --> 00:30:43.589 and the correction is buried on page 40. If it 00:30:43.589 --> 00:30:45.849 runs at all. So we've established that trust 00:30:45.849 --> 00:30:48.650 has been eroded by literally centuries of this 00:30:48.650 --> 00:30:51.430 stuff. It's not a new problem. But the modern 00:30:51.430 --> 00:30:55.000 numbers are just stark. They really are. In 1985, 00:30:55.339 --> 00:30:58.079 a Gallup poll found that 55 % of Americans trusted 00:30:58.079 --> 00:31:00.660 the media to get the facts straight. By 2022, 00:31:01.000 --> 00:31:02.759 a different poll found that half of Americans 00:31:02.759 --> 00:31:05.660 believe news organizations deliberately try to 00:31:05.660 --> 00:31:08.299 mislead them. That's a fundamental shift. We've 00:31:08.299 --> 00:31:10.579 gone from they sometimes make honest mistakes 00:31:10.579 --> 00:31:13.740 to they are the enemy and they are lying to me 00:31:13.740 --> 00:31:16.119 on purpose. And this brings us to the final section, 00:31:16.319 --> 00:31:19.299 the psychology of trust. Why do we believe the 00:31:19.299 --> 00:31:21.859 media is biased? I mean, because they are biased. 00:31:21.880 --> 00:31:24.000 We just spent the last 50 minutes proving it 00:31:24.000 --> 00:31:26.660 with dozens of examples. True. But there's a 00:31:26.660 --> 00:31:28.339 metal layer to it that's really fascinating. 00:31:28.880 --> 00:31:31.160 Research by a political scientist named Jonathan 00:31:31.160 --> 00:31:34.740 Ladd suggests that the primary cause of the belief 00:31:34.740 --> 00:31:38.599 in media bias is being told over and over that 00:31:38.599 --> 00:31:41.079 the media is biased. It's a self -fulfilling 00:31:41.079 --> 00:31:44.190 prophecy. Yes. It's a political strategy. When 00:31:44.190 --> 00:31:46.690 political elites spend 30 or 40 years telling 00:31:46.690 --> 00:31:49.150 their base, the news is fake, don't trust the 00:31:49.150 --> 00:31:51.869 journalists, they are lying to you, eventually 00:31:51.869 --> 00:31:54.170 the base believes it, regardless of the actual 00:31:54.170 --> 00:31:56.089 quality of the reporting. It becomes a tribal 00:31:56.089 --> 00:31:58.930 marker, a badge of identity. Exactly. To be a 00:31:58.930 --> 00:32:01.589 good conservative or good liberal in today's 00:32:01.589 --> 00:32:04.369 environment, you must performatively distrust 00:32:04.369 --> 00:32:06.710 the other side's media outlets. So we are stuck 00:32:06.710 --> 00:32:09.740 in a trap. The media has genuine, deep structural 00:32:09.740 --> 00:32:12.940 flaws, but our perception of those flaws is then 00:32:12.940 --> 00:32:15.440 exaggerated and weaponized by political rhetoric. 00:32:15.680 --> 00:32:17.839 How do we get out of this mess? Are there any 00:32:17.839 --> 00:32:20.259 solutions? There are attempts. People are trying 00:32:20.259 --> 00:32:22.039 things. Don't tell me the solution is just to 00:32:22.039 --> 00:32:24.980 put both sides on TV more? No, absolutely not. 00:32:25.059 --> 00:32:28.759 As we discussed, both sides. Hedonism often leads 00:32:28.759 --> 00:32:31.539 directly to false balance. But there are some 00:32:31.539 --> 00:32:33.880 interesting algorithmic solutions being tested. 00:32:33.960 --> 00:32:35.500 Look at Taiwan, for example. What's happening 00:32:35.500 --> 00:32:38.099 in Taiwan? They use a system in their civic processes 00:32:38.099 --> 00:32:43.259 called Pol .is. P -O -L dot I -S. That's the 00:32:43.259 --> 00:32:45.880 one. It's a digital platform used for public 00:32:45.880 --> 00:32:49.119 deliberation. But instead of optimizing for engagement, 00:32:49.319 --> 00:32:51.200 which, as we know, usually just means fighting, 00:32:51.420 --> 00:32:54.140 it's designed to optimize for consensus. How 00:32:54.140 --> 00:32:56.380 on earth does it do that? It maps out everyone's 00:32:56.380 --> 00:32:59.569 opinion on a topic on a visual graph. And then 00:32:59.569 --> 00:33:02.069 the algorithm specifically hunts for statements 00:33:02.069 --> 00:33:04.750 or ideas that large groups of people, even people 00:33:04.750 --> 00:33:06.910 who usually disagree on everything else, can 00:33:06.910 --> 00:33:08.990 both agree on. So instead of throwing red meat 00:33:08.990 --> 00:33:11.450 to the most extreme bases, it actively searches 00:33:11.450 --> 00:33:13.529 for the common ground that everyone's standing 00:33:13.529 --> 00:33:16.410 on but nobody's noticed. Exactly. It helps identify 00:33:16.410 --> 00:33:18.650 shared values and points of agreement that are 00:33:18.650 --> 00:33:20.269 usually drowned out by all the shouting. It's 00:33:20.269 --> 00:33:22.609 a tool designed for listening, not for broadcasting. 00:33:22.890 --> 00:33:25.460 I love that. Using technology to fix the problem 00:33:25.460 --> 00:33:27.640 that technology helped create. What else is there? 00:33:27.759 --> 00:33:30.539 A more old -fashioned idea. Radical transparency. 00:33:30.839 --> 00:33:33.859 The concept of the ombudsman or public editor. 00:33:34.039 --> 00:33:36.619 The person inside the newsroom whose job it is 00:33:36.619 --> 00:33:39.180 to criticize the newsroom. Right. NPR tried this 00:33:39.180 --> 00:33:41.259 for a while. Every time they cited a think tank 00:33:41.259 --> 00:33:44.140 in a story, they would explicitly state its political 00:33:44.140 --> 00:33:47.000 leaning. The Heritage Foundation. A conservative 00:33:47.000 --> 00:33:50.079 think tank or the Center for American Progress, 00:33:50.359 --> 00:33:52.619 a liberal think tank. Just putting all the cards 00:33:52.619 --> 00:33:54.460 on the table for the audience. It treats the 00:33:54.460 --> 00:33:56.480 listener like an adult. It says, here is the 00:33:56.480 --> 00:33:59.119 source. Here is their potential bias. Now you 00:33:59.119 --> 00:34:01.559 have the context to decide for yourself. That 00:34:01.559 --> 00:34:04.079 seems like a really good place to land. We have 00:34:04.079 --> 00:34:06.799 covered a massive amount of ground today from 00:34:06.799 --> 00:34:10.019 the. initial definitions to the invisible biases 00:34:10.019 --> 00:34:12.360 of concision and negativity that shape everything. 00:34:12.760 --> 00:34:15.239 We looked at the ski resorts and the 1890s baseball 00:34:15.239 --> 00:34:17.739 fans, the eternal dance of supply and demand. 00:34:18.019 --> 00:34:20.780 We saw how the algorithm seems to amplify the 00:34:20.780 --> 00:34:23.460 right while the right claims censorship. We traveled 00:34:23.460 --> 00:34:25.780 back in time to the sedition acts and the satanic 00:34:25.780 --> 00:34:28.880 panic, and we ended on the psychology of distrust. 00:34:29.000 --> 00:34:31.349 It's a heavy lift. There's a lot there. So if 00:34:31.349 --> 00:34:33.550 the listener is going to walk away with one thing, 00:34:33.610 --> 00:34:36.130 one tool for their mental toolbox to navigate 00:34:36.130 --> 00:34:39.849 this, what should it be? I think we need to stop 00:34:39.849 --> 00:34:43.929 viewing bias as a bug in the system that we can 00:34:43.929 --> 00:34:47.329 fix by just firing the bad journalists. Bias 00:34:47.329 --> 00:34:49.829 is a feature. A feature, not a bug. It's a feature 00:34:49.829 --> 00:34:53.070 of our biology. We are hardwired to crave bad 00:34:53.070 --> 00:34:55.949 news. It's a feature of our economy. We pay for 00:34:55.949 --> 00:34:58.269 confirmation. And it's a feature of the modern 00:34:58.269 --> 00:35:01.019 media environment. Brevity kills nuance. It's 00:35:01.019 --> 00:35:03.400 the water we swim in, not a shark that we can 00:35:03.400 --> 00:35:06.219 hunt down and kill. Exactly. So the only defense 00:35:06.219 --> 00:35:08.800 is to become a more conscious, active swimmer. 00:35:08.820 --> 00:35:11.000 It's about critical thinking. When you see a 00:35:11.000 --> 00:35:12.900 story, don't just ask yourself, is this true? 00:35:13.360 --> 00:35:16.119 Ask, why is this story being presented to me 00:35:16.119 --> 00:35:18.599 in this specific way at this specific time? Is 00:35:18.599 --> 00:35:21.400 it supply? Is the owner of the outlet pushing 00:35:21.400 --> 00:35:24.449 an agenda? Is it demand? Am I clicking on the 00:35:24.449 --> 00:35:26.530 story because it makes me feel good and validates 00:35:26.530 --> 00:35:28.630 my worldview? Or is it structural? Is this just 00:35:28.630 --> 00:35:31.070 a short, simple, sensational story that was easy 00:35:31.070 --> 00:35:33.590 to produce? Precisely. You have to diagnose the 00:35:33.590 --> 00:35:35.650 type of bias before you can account for it. And 00:35:35.650 --> 00:35:38.190 that leads us to the final thought, the provocative 00:35:38.190 --> 00:35:40.389 one you wanted to leave everyone with. We love 00:35:40.389 --> 00:35:43.329 to blame the media. It's the perfect faceless 00:35:43.329 --> 00:35:46.590 villain. But the research, especially all that 00:35:46.590 --> 00:35:49.949 demand side stuff. It points the finger right 00:35:49.949 --> 00:35:54.289 back at us. We say we want unbiased news. We 00:35:54.289 --> 00:35:57.809 say we want the complex, nuanced truth. But our 00:35:57.809 --> 00:36:00.210 clicks, our subscriptions, our shares, and our 00:36:00.210 --> 00:36:02.570 viewing habits tell a very, very different story. 00:36:02.949 --> 00:36:06.210 We vote with our attention. We do. And right 00:36:06.210 --> 00:36:08.769 now, we are overwhelmingly voting for comfortable 00:36:08.769 --> 00:36:11.300 lies and rage bait. The market isn't broken. 00:36:11.519 --> 00:36:13.619 It's just giving us exactly what we ordered. 00:36:13.900 --> 00:36:16.300 So the real question isn't how do we fix the 00:36:16.300 --> 00:36:18.280 media? The question we have to ask ourselves 00:36:18.280 --> 00:36:21.280 is how do we fix our own appetite? Wow. That 00:36:21.280 --> 00:36:23.059 is the challenge. The next time you click on 00:36:23.059 --> 00:36:25.320 that angry headline, ask yourself, am I ordering 00:36:25.320 --> 00:36:27.360 the junk food again? It's the only way to ever 00:36:27.360 --> 00:36:29.420 change the menu. Thank you for guiding us through 00:36:29.420 --> 00:36:31.300 the mechanics of the machine. Always a pleasure 00:36:31.300 --> 00:36:33.500 to take things apart. And thank you all for listening. 00:36:33.659 --> 00:36:35.360 We'll see you in the deep end next time.