Google, Tim Wong, Nico Newman discussed on Freakonomics Radio
G. My day job is I'm a research fellow at the Center for Security and emerging technology at Georgetown. And before that I was previously global head of public policy. For a I'm machine learning at global, long recently published a book called Subprime Attention, Crisis Advertising in the Time Bomb at the heart of the Internet. It's about how big tech monetize is our attention. When I started to do research, by very naturally started to talk to a couple friends who work at these big tech companies, and it was a little bit like talking to someone who works in national security or the intelligence community or something like that, because they would be like, Oh adds definitely work, but we can't tell you how or why, or give you any evidence for it. Google would plainly dispute that there is no evidence for whether online ads work. Tim Wong recognizes he is tilting at windmills here. Trillion dollar windmills. But in fact, he first grew skeptical about online advertising. While still working at Google. He began reading trade journals and going to conferences and I had this fascinating experience with one of these key notes at this conference was given by Nicholas Newman, who basically is a big ad critic. Nico Newman is a marketing professor at the Melbourne Business School in Australia. And he presented to really fascinating studies that his lab had done. The first one was looking into the quality of data used in the ad tech industry, basically demonstrating in many cases, it was incredibly accurate. And the second one was he took dead aim at the type cycle around a I that exists in ad tech right now, where people are saying, If you have this latest machine learn, and you have this AI, you'll be able to do targeting in a way that you never ever were able to do before. And Nico's lap, did some experiments that demonstrated that in many cases, machine learning was finding people who would have bought the product anyways. Bringing this message to an ad tech conference is a bit like bringing a safety pin to a balloon conference. I looked around being like way. So when people be angry, and it was just total dead air, no one responded. No one engaged with it. And it got me really interested in thinking about like, Is there a bubble here? Bubble like the dot com bubble or the subprime bubble or the tulip bubble, because this is exactly the kind of behavior that occur in other financial bubbles, where the red lights are flashing, But everybody in the street just refuses to take a look at the real data. Long began thinking about how bubbles happen. So the origins of every bubble come in this cap that occurs in a marketplace. On one hand, you have people who believe that an asset whether it's collateralized debt obligations or advertising inventory is extremely valuable. And then, on the other hand, what you have is declining asset value. So in the subprime mortgage crisis, we believe that mortgages were always going to just pay out regularly forever. Right when next you turned out that the package of mortgages were actually a terrible asset. They were toxic and about to go belly up. So how can one justify a parallel with digital advertising? I think the first piece is really the big question of do people ever see ads at all? So Google actually did a fascinating study not too long ago, which concluded that close to 60% of ads on the Internet are never ever even seen. The ad is delivered, but it just ends up in some dumb part of the page. Right. It's below the fold. It's along a sideline, but what about the precise targeting the digital ads are supposed to offer? 2019 study, this one done by three academic researchers addressed this question by measuring the impact of a user's cookies. Those remember are the tracking codes. Most of us allowed to roam our computers and phones in exchange for all the free information we get from companies like Google and Facebook. This study found that when a user's cookies were unavailable, ad revenues on Lee dropped by about 4%. Why would cookies be so ineffective? Tim Wong argues that people pay a lot less attention to online ads and they used to people often forget that when banner ads were first launched on the Internet there, click through rate was like 50% completely mind bending right and it's just continued to fall and fall and fall and now it's like 500.12 point. 03%. Some estimates of click through rates are higher than what Hong sites here. That said precise measurement is hard because there are so many bots clicking on adds a whole other problem with the digital ad universe. But no matter how you measure it, click through rates have fallen a lot. As the novelty wears off habituation sets in, and an ad that might have once grabbed your attention becomes invisible or worse, Annoying. People increasingly don't want ads. So ad Blocking, for example, is really, really increasing overtime. And I think these factors not being able to see ads the questions about the effectiveness of ads. And the rise of things like ad blocking, bring into question whether this thing that we think is so valuable is actually worth as much as we think it is. But if there's such a big gap between the perceived and real value of digital advertising Why are Google and Facebook worth so much money? Look at it this way. There are a couple trillion reasons why Tim Wong might be wrong. He doesn't think so. His theory is that digital advertising is grotesquely overvalued because it is still so hard to measure and one reason it's hard to measure is that the marketplace is exceedingly opaque. So there's a fascinating incident that I always think about, which is one of the last times that Mark Zuckerberg was called up to Congress and one of the questions that he got from one of the senators was. How do you guys make money? How do you sustain a business model in which used to stump tape your service and Mark's hacker? Berg was like Senator we run as and at the time. A lot of the chatter on Twitter was like, Ha ha! Look at the super Old senator, he doesn't know anything about the Internet. But it's true that even if you talk to people in the tech industry, and you're like, Okay, level with me, Joe Engineer how do ads work on the Internet? It's kind of a rumor like we know this is how the business model works, but no one can really explain how it works in detail. So when I say advertising a lot of people normally think of like mad men, right, but it really looks like what the NASDAQ looks like, which is largely automated system that moves millions and millions and billions of pieces of ad inventory on a daily basis. As Steve today, Ellis explained earlier. Most ad inventory is sold by auctions, which are run by algorithms operating at phenomenal speed. This is one contributor to the opacity of the industry..