B (2:47)

A lot of fish T shirts in the closet. Should have worn one. I would have worn one.

A (2:52)

Had I known I would have.

B (2:53)

I knew. Yeah, like pause the episode. Episode and let's go get changed. So our audience is used to me interviewing people on this podcast that are in the ad tech industry or work in Publishing, some CMOs at brands. I have talked to a few privacy lawyers, but I haven't had an economist on this show before. So tell me a little bit about yourself. Introduce yourself to our listeners. And then also, what is the International center for Law and Economics like, what do you guys do exactly? Because I imagine you just drinking a lot of coffee, spending a lot of time with whiteboards, but get more specific.

A (3:34)

Nice of you to say coffee. It's usually whiskey. Well, I'm really an amateur economist. My degrees are in law, but my degree, my degrees are from the University of Chicago, which is sort of like getting an economics degree, you know, no matter what you take. Excuse me. So I do consider myself an economist, but I'm a lawyer by training, I guess. I started, I used to teach law at Lewis and Clark Law School here in Portland where I'm based. And I was asked while I was there by Microsoft to run this program that they were starting that was sort of building a relationship between Microsoft and the, what's called the law and economics academic community. And law and economics is a sudden field in economics and it's pretty self descriptive. It's economics and law together and that sort of got me into this, into a much more of a, excuse me, sort of policy space and much more focused on the relationship between academia and things that are actually happening in the world. Most academia focuses on things that are not unfortunately really important in the world. And I thought that was pretty useful. So after I left Microsoft, instead of going back to teaching, I started ICLE to basically facilitate the growth of law and economics as a movement within, within the Law and Economics academy, but also to facilitate this sort of relationship between the academy and business and policy world. My interest, my focus has always been antitrust and competition. But at ICLE we focus on huge range of issues. Privacy, as you mentioned, intellectual property, telecommunications, just about everything really. And we do research and writing, we're sort of a research center, but we also have about 100 or so academic affiliates here, law and economic scholars kind of around the world who we, you know, we provide support for and try to encourage and facilitate their, their ability to influence thinking in the academy and outside the academy as well.

B (6:06)

Whiskey fueled and I guess also Topo Chico fueled. Because right before we started recording we both realized we were sipping from bottles of Topo Chico. So spicy water all day, right? Take a sip.

A (6:19)

The only card. Yeah.

B (6:22)

So you brought up antitrust. And I want to spend pretty much the rest of the episode talking about the Google Ad tech antitrust trial. The Remedy phase is going on right now as we record this. But first, how does the ICLE get funded and then how do you make sure to stay like nonpartisan in cases where maybe some funding comes from a company like Google? Because the ICLE has taken money from Google in the past.

A (6:49)

We are funded by a mix of private companies, foundations, a few individuals. I mean, kind of like probably most nonprofits are. We our all of our support is general support with no strings attached. No funders have any say over what we write or when we write it, or they have no ability to, well, they have no insight into what we're doing really other than to look at it retrospectively. So that's an important part of maintaining our independence. We also have such a variety of funding sources that there's rarely anything that we do that we don't where we don't have at least one funder on sort of both sides of the issues. So we have had funding from Google before, but you know, also from, from, in this case, companies who would certainly be considered publishers, companies that are, you know, direct competitors of Google. We used to do a lot of work around net neutrality, for example. Very much opposed to Google's position on I'm against it, they were for it. So, you know, it's making sure we don't have any contractual strings attached, making sure we have a diversity of funding sources. And everyone who funds us recognizes that our value is in our independence and the quality of our work. And so, you know, they'd be sort of hurting themselves if they tried to assert it in kind of sight over what we do.

B (8:28)

Separation of church and state. I get it, it's very important. So I'm going to get us up to speed and then we're going to get very nerdy with this case. But the timeline to date is that the liability phase of the ad tech antitrust trial against Google took place last September in Judge Leonie Brinkoma's courtroom in Alexandria, Virginia. It was a three week process. I had opportunity to be there for the first week. It was really interesting experience. I brought my notepad and a pencil and I filled three, almost three full subjects of a three subject notebook with notes in one week because there were no recording devices in the courtroom. Closing arguments and thanks. It was at around Thanksgiving time. And then Judge Brinkham found Google guilty of operating monopolies in two very specific markets. It was the publisher ad server market primarily through DoubleClick for Publishers, DFP, now GAM and the ad exchange market, ADEX, open web display advertising and now we're in the midst of the remedy phase, which is when the court gets to decide what actions the defendant has to take to stop the illegal conduct and restore competition. And the big debate is whether Google should have to spin off the ad server and the ad exchange, or whether behavioral remedies will be enough, like, whatever that means. Because Google is willing to make certain behavioral changes. They don't totally align to what the DOJ is asking for. But the DOJ doesn't think that's enough. The DOJ is really pushing for forced divestiture. But my question is like, zooming out, like, looking back at the history of antitrust, do force divestitures, like, quote unquote, work? Like, do they help create more competition, or despite the best of intentions, are they like, so messy and complex and I guess, unpredictable, and they take so long that they're not necessarily worth it? What's the literature on that?

A (10:27)

Yeah, so I think it's worth just sort of separating conceptually between divestitures as part of a merger remedy or usually a settlement of a merger challenge, and divestiture as part of a conduct remedy in a conduct case, which would be like the ad tech case. It's a monopolization case. So just quickly, in the merger context, I think there's a pretty good sort of history of divestitures. A great example, I guess, is. Or not a great example, but an easy example is had the Kroger Albertsons merger actually happened, which it was spoused by the agencies, they certainly would have divested some of their stores. Right. You have, I don't know, a couple thousand retail locations around the country. You have a merger that in some markets might, might make the market go from having two competing grocery stores to now having, you know, one, well, two grocery stores, but under one banner and not really competing with each other. So you force them to divest one of those stores and it maintains competition in that, in that market. And that is sometimes not work. But. But it's very common in the merger context, and excuse me, and it facilitates. This is important. It allows a merger to happen. If you think that mergers, you know, in many cases have some value behind them but also pose some competitive risks, you know, you can really kind of split the baby by still allowing the merger and just, you know, removing the places where there's the most obvious competitive risk, like that two to one situation, it's, you know, a small tenant value horse. Okay, so that's very different than in the, you know, the sort of conduct context in which we actually can and should talk about that a little bit because of course the purchase of DoubleClick was itself kind of a question in this case. And we can come back to that. But primarily in this case the issue was around Google's conduct, how it operates, this, you know, integrated ad tech stack. And what we would say in, as economists is that in this market Google is vertically integrated. Right. That they operate at sort of different levels of the supply chain. Harder to think of it that way when it's bits. But essentially they're facilitating this exchange between buyers and sellers happening at different levels of the supply chain. And there's a kind of a, I wouldn't say a long history, but a decent history of antitrust remedies in conduct cases that have tried to break up companies or provide structural remedies across these vertical levels. One example would be the so called Paramount decrees, Paramount cases in the 50s where movie theaters, movie studios used to own movie theaters as well as the make the movies. And that was deemed to be problematic. They had a sort of lock on distribution and they would, they would prevent competing studios from being able to distribute movies in their theaters and vice versa. And excuse me. And in that case they agreed.

B (13:58)

Hey, take a sip of topo chica.

A (14:00)

Exactly.

B (14:03)

That's what it's there for.

A (14:04)

Delightful insights.

B (14:05)

Delightful. Yeah. This episode is brought to you by.

A (14:11)

So in that case they were forced to, not exactly structurally separate, but essentially structurally separate that vertically between the studio production of movies and the distribution of movies. And so actually now I answer your question 20 minutes later. That was generally viewed as a pretty unsuccessful divestiture. And in fact pretty much every antitrust divestiture of that sort that's been tried has been viewed as largely unsuccessful. So it hasn't been tried that much probably because it is such an aggressive approach to restructuring, to reimagining an industry. The courts recognize that they're not really that good at doing that. That's not, you know, that's not their expertise. Even insiders in these industries probably are that good at it that they often can do more harm than good. And, and so the, the research shows that that's largely the case. Well, shows that either it's had no appreciable effect on competition or it's been harmful. Like literally. There are cases where prices have gone up after these divestitures instead of down. Output has gone down instead of up. The expected new entry never really materialized. Or in some cases, one of the great examples of this is the AT&T breakup that everyone knows about. You had you Know, a gold rush of new entrants, virtually all of whom ultimately went bankrupt because nobody really knew exactly how to operate a, in that case, purely local telephone provider in this sort of new brave new world that the court invented, that separated the, you know, long distance from the local. And that's fundamentally the difficulty is that what might look good on paper, usually for the purpose of facilitating new entry, to facilitate new competition in a market in practice may not be possible. There may be cross subsidies that are happening. That's something we should talk about in this case where maybe you can really only profitably operate. And this isn't true because we know they're counter examples to this. But maybe at the scale of Google, you can't operate a supply side platform without cross subsidizing it from an exchange or a buy side platform or vice versa. So historically these remedies have happened. They have been pretty much uniformly, either, as I said, either harmful or unsuccessful. But it raises the great question, which is, what's the alternative? Which is sort of probably your next question.

B (17:07)

Well, yeah, I mean, so I'm just going to keep calling you an economist. But from someone who frames themselves as an economist's perspective, what would you say healthy competition in ad tech actually looks like then? Considering that some things look great on paper and we have the reality of Google being found to be a monopolist in certain markets by a court, and now we're trying to figure out what to do, what is healthy competition in ad tech?

A (17:34)

I mean, I think part of the problem is that we don't know. And that's sort of what I was getting at before and Judge Brinkerman in this case, it's very clear that she's very sensitive to the idea that she doesn't really know. She keeps saying in the remedies trial, she keeps saying, she keeps stopping the testimony. Have you seen this? To say, I think one of your colleagues reported this. This is just window dressing. Either I've already heard this.

B (18:01)

Get to the point.

A (18:03)

Yeah, I need to hear from experts who, many of whom are at Google, of course, who actually know how to sort of operate this industry or how to, you know, who are engineers on the ground who are trying to figure this stuff out. Because she doesn't know. And I think that's very commendable on her part. And the concern would be that no one really knows. You know, my sense is, my suspicion is competition in this market would look a lot like what it looks like now, no matter what you do, that this is a market that is this is a market. These are markets. This is a market. We'll talk about market. This is an industry in which there are huge returns to scale, there are network effects. There are reasons why you would expect a, some fairly substantial kind of concentration. It doesn't make sense to be operating an infinite, near infinite number of simultaneous auctions that are, that are, you know, somehow not speaking to each other. You'll never get the best price that way. You have to have, you don't have to necessarily have literally a single auction, but you do have to have some way of, of. What's the right word? Combining, you know, diversity of sources of on the buy side and the sell side that naturally lends itself to a fairly high degree of centralization. Now we can talk about, maybe I'm just talking about the ad exchange market market there. Maybe that's not true on, on either the sell side or the buy side, you know, and I think there is more diversity or more, more, less concentration, I guess, or a greater number of competitors, I suppose, on the buy and sell side platforms or at least the sell side. But again, I just think, I think my concern would be no matter what you do, you're going to end up back at a market that looks a lot like the one we have now. Maybe it's not Google sort of at the center of it, but someone's going to be. Or if it's not a single entity, they're going to have such a tight contractual relationship with the other entities that are performing the functions that Google currently performs at different layers of the stack that it's functionally the same thing. This is from someone who is not a great expert. I mean, I've spent a lot of time studying it, but I, you know, I find this market to be so difficult and so confusing, you know, so I mean, I want to sort of. Does that resonate at all with you? You have a lot more experience here than I do. I mean, I want to turn it back on you. What do you think a competitive market here would look like? You're better positioned to say to know than I am.

B (20:51)

Oh no, the interview subject has become the interviewer. I mean, some of the proposed behavioral remedies feel like they could introduce some measure of better compet where you're not locking people in to DFP because they want ADEX demand. So opening that up a bit I think could help. But I do think we're always going to have a few really large entities that have gravitational pull. It seems like the market lends itself to that. I Mean, there's definitely more and more money going to walled gardens. There's also, there's the argument that the industry, the market, because of people's tastes and where eyeballs are going. It's changing so greatly that we're trying to figure out remedies for a market that's on the edge of being passe, which I want to talk about too. But I think interoperability, more data sharing. If I was a betting person, I would say that Judge Brinkoma isn't going to order for a forced divestiture. I just don't, I don't think that's going to happen. But I would hope that some of these behavioral remedies, more so the ones that DOJ is calling for than the slightly watered down ones that Google is calling for would hopefully create more competition. There's already some competition amongst the smaller players, but Google sucks a lot of the air out of the oxygen out of the room.

A (22:19)

Yeah, and it's a great point. You asked that question and I was immediately thinking sort of more structurally and I think we agree structurally there's, there's a lot of reason to expect, you know, big, big players are always going to dominate, maybe not be the sole providers that dominate in this industry. But you're absolutely right that that competition could look, you could still affect competition with behavioral remedies. And I agree that. I mean, obviously, I mean even Google agrees, right. They're proposing, I mean they are proposing behavioral remedies also. You know, I think it's something else we'll talk about. But you know, some of those remedies are implementing changes they've already implemented that you know, wouldn't be a sort of a substantial change from how things look now, but they would just sort of ensure that it never goes back to how it was before.

B (23:16)

I guess. So it feels a little weak to me though. Right.

A (23:19)

Because they already exist. Like what are you really offering here? But you know, but it does go to your point about how you're inherently sort of litigating history in these and that is actually as a, as a practical matter that's always how this goes with antitrust. Your except perhaps maybe in merger cases. But in, in conduct cases you're always looking at the past and then when you're imposing a remedy, the remedy is supposed to be forward looking. You're not. Antitrust is not about punishing the offender. The ADHEs remedies are not really about. In government cases we can get to, private cases are not about punishing the offender. They're about, you know, trying to, to correct whatever was, was distorting the market to ensure that we have competition going forward. And that means, like you just said, you know, you, you're, you're looking backward at it to, to, to define what the harm was and then you're looking forward to define, you know, what the remedy is. There's always going to be a disconnect. There going to be a disconnect.

B (24:28)

By the same token, though, I feel like you could run out the clock as a monopolist and then not only do you get away with bad behavior, but then you potentially get to use behavior that is subsequently found to be illegal as a springboard into a next big market, like AI Search, for example. And I know that's not this case, that's jumping over to the, to the search case, but so obviously prior behavior is fair game, but it doesn't feel fair. That delay could just be like a legal barrier.

A (24:59)

Yeah, it's, it's, you know, it's a, it's a, it's a longstanding kind of existential debate in antitrust. You're always litigating the past and it always takes a long time. And by the time you get around to remedies, things have changed. And, you know, and, and it's why in, in other parts of the world, Europe most notably, it's at least how they defend sort of, you know, complementing their regime or even sort of switching their regime over to what's called an ex anti regulatory sort of environment. The DMA is meant to be much faster and much more agile than traditional antitrust litigation, which has this inherent sort of time problem. I don't know that it's actually happened that way yet. I mean, it's still new, you'll see what happens. But there's a widespread recognition of this problem. But in a world in which you have due process and in which you have a, you know, somewhat unique body of law, excuse me, that is, that is defined by its economic effects. Meaning, you know, unlike, I don't know, a murder case or something, the harm is not really the action itself. It's the consequence of action, the downstream effects. Yeah, exactly. And, and that means litigation is very complicated and very tenuous and very long and difficult. And if you want to sort of preserve the accuracy and the efficiency and the desirability of antitrust enforcement, and of course preserve due process, you're sort of stuck with this inherent trade off. And in a way you say there's not much you can do about it. I mean, there are things you can do obviously at the, at the agents. And one of them would be what you just mentioned. And you know, we can talk about the legality or the, the, you know, the history of this, but it's not, there's a question whether it's appropriate as part of a remedy in an antitrust case that looked at, say, you know, let's talk about search for a second. Looked at search and the, the sort of, the way search looked in the past. There's an interesting question about whether it would be appropriate to impose a remedy there on a market like AI search and chat and GPTs and other things. There's a lot of this question about whether it be appropriate to impose a restriction that affected a market that wasn't even at issue in the case at all for precisely the reason you're suggesting. So I guess what I'm saying is you've hit the nail on the head and there isn't an answer to the question. I mean, I have thoughts about it and we can, you know, we can talk about it if you'd like. But, but it is a, it's an important question because again, inherently you have this problem in heavy trust. And that's why some people thought the remedies in the search case were insufficient because they didn't sufficient, among other things, didn't sufficiently sort of constrain Google from leveraging its apparent dominance in one market into what is emerging as the next big market where they could become dominant.

B (28:24)

I do want to talk more about that, but I think this is a good place to stop. Take a sip of the old topo chico and quick moment of silence. That is delicious. And when we're back, we'll talk about the connection, if any, between the search case and the ad tech case and a little bit about Judge Brinkoma's vibe and a few, few other things. So stick with us. All right, we're back. And we were talking a little bit before the break about new markets like AI. We veered a bit away from the ad tech case and started focusing on the search case and I guess rather naively was saying that it's sort of unfair to be able to parlay your existing monopoly in a market that's starting to change and as a springboard to dominance in a new market. But you're not dominant yet. It almost seems like just the best of every world for you if you play your cards right and the timing works for you as a monopolist.

A (29:45)

I think that gives a little bit more credit to the, the, the monopolist than they deserve. I don't think they are quite as. I'm sure they try to be that strategic. But I, you know, I.

B (29:57)

A little bit of luck. But still, it's a little bit, it's worked out well.

A (30:00)

No, it does. And in many cases, in fact there's, there are those who have argued in the past, again because of, because of this inherent indeterminacy and the problems with, with, you know, like you said, do you really want a body of law that finds it, it's, finds it illegal to be a new entrant in a new market just because you were powerful in another market? You actually might see, you know, Google entering a new market as bringing very important competition to that market that might otherwise become dominated by, by someone else. It's a little bit like Minority Report, you know, are you gonna, are you going to find, judge them to be behaving illegally before they even have the ability to behave illegally? But you could totally understand why that there's some logic to that, despite the fact that there are also some inherent problems with it. And so you have seen some scholars who suggest things like if you have, if you're a big monopolist, especially a big tech monopolist, you get 10 years. This is, Tim Wu says this, you get 10 years. And Tim Wu's maybe your listeners, I don't know.

B (31:09)

But yes, I read his book and now I'm blanking on the name. What was his famous book? I'm going to Google it while, while you're talking. But it was a great slim little. The Curse of Bigness. Yeah, my editor bought it for me a bunch of years and I read it on an Amtrak down to D.C. but yes, he's a very smart man and used to work for the White House briefly. Right. I think under the past administration competition fallacy under Biden.

A (31:35)

Anyway, he proposed in his other book, I think the name of which is also escaping me that these big tech monopolies, they get 10 years and then they're basically broken up. I can't remember exactly what he suggested now. I think there are big problems with that, but again, I understand the impetus for it. It's sort of a, if it, you know, if it goes on too long, you're, you are going to be able to potentially parlay your dominance into so many other areas and it's, you know, potentially really problematic. And maybe we think that, you know, 10 years is enough to provide a sufficient return on your investment that you, you know, we still want to, we want to encourage companies to try to become monopolies, we just want them to sort of fail, but we really want them to think they can and you know, to, to expect the returns of monopolization because that induces them to make huge investments and innovations and things like that. So you know, it's not an insane idea, but it, you know, you can see the, there's a, there's an artificiality to it, but again, there's sort of an artificiality to any approach to antitrust. Anything we do is going to be based on suppositions about the future that we don't know that much about. It's going to be based on harms that happened in the past that may not be particularly relevant today. And so where there's always, as I said, a disconnect or almost always a disconnect between the behavior that caused the harm and what we do about it going forward. And you just have to sort of pick on what dimension we want this artificiality to exist.

B (33:07)

Well, I don't want to call the market definition in the Google Ad tech case artificial, but it is a little bit because I mean, yeah, if we want to talk about the market definition that the DOJ used, it's like a little awkward and I think it might come back to bite them now during the remedy phase because it's super critical. Market definition. I know you've written about this like in an antitrust case because you really need to show exactly what a company in question is monopolizing. There needs to be a clear delineation. And in this case the DOJ limited the relevant market to open web display advertising. And that's arguably a pretty narrow definition because it's only talking about ads on websites sold through exchanges and not direct deals. It excludes mobile apps like MO ads and mobile apps ads on ctv. So yeah, I mean, could this definition end up being a problem for the DOJ if they're seeking these stronger remedies because by narrowing the market so much they were able to prove their case. But then I feel like it makes it a little harder to justify these broad sweeping remedies that they're looking for.

A (34:17)

It's, it, it's, I mean, it's, it's a great question and a great follow up to what we were just talking about because it's exactly the same. Maybe not exactly the same. It's essentially the same issue, the same, the same sort of problem. Market definition is always backward looking and, and remedies again are always, always forward looking. In, in this case, I think you know, you hit upon something that's really very interesting. Well, so they, this happened in the search case. This actually happened in the famous Microsoft case. This happens in a lot of antitrust cases, but not all of them. And I, I would say it's fair to say this is happening more than it used to. And that is the separation of the liability and remedy phases of the, of the, of the trial. And there are lots of reasons to do that because, and you don't have to go into them. Let's just, you can imagine reasons why that's a beneficial way of running the trial, but it exacerbates exactly the problem that you're describing. It gives a strong incentive to the plaintiffs, of course, to define markets and define harm and various other things in such a way that they're sort of assured of winning at the liability phase. That's the most important thing. But then those definitions, those decisions that they've made carry over into the remedies phase and may limit the ability of getting sort of a proper remedy. So in this case, you know, there is, as I said a minute ago, or lilacord, probably, I've spoken so much, it was probably an hour.

B (35:49)

You're good.

A (35:51)

As I said, there's an open question about how much, how appropriate it is to impose a remedy that constrains a market that was not the market at issue or markets at issue in the case where harm was found. And the narrower the market, the more difficult it is to impose a remedy that doesn't expand into other markets outside of.

B (36:16)

Feels like a catch 22. Right. Win your case and then it makes the remedies that much harder to get.

A (36:22)

Right. But it's, it's sort of, well, it doesn't, it doesn't make, in some ways it makes remedies easier. It makes it harder to get the more aggressive remedies.

B (36:31)

Oh, right. What you want, what you want for Christmas.

A (36:33)

Exactly. Exactly. Right. I mean, I guess, I think that, that maybe, maybe it's easier to see, it's easier for me to conceptualize if we talk about another aspect of the market definition that was contested, which is. Right. So, so clearly dividing the buy side from the sell side. Right. Google wanted to argue that, that this was a, you know, sort of integrated transactional market and you couldn't really separate one from the other. And the court said, and you know, appropriately, that there's obviously no interchangeability, there's no substitutability between supply side tools and demand side tools. They're not competing with each other. And that's usually how we define markets in antitrust. The scope of products that are potential substitutes for each other. But that artificiality does miss something really important in this market. There's something about what Google is talking about here that makes a lot of sense. It is hard to. You could artificially sort of sever off the portions of this, of this market. But I think it's important to recognize that there is an inherent kind of artificiality to doing that. The two sides are very importantly related. One side may be subsidizing the other side. We talked about this before. Again, I don't know whether this is true or not, but I could certainly imagine that it might not be profitable to participate in one side of the market without also participating in another side of it. The need, as I mentioned before, to sort of centralize and combine competing bids and asks is essential here. And while the tools that are deployed are very buy and sell specific, they're all aimed at the same thing, which is maximizing revenue. I mean, I guess this is something we have to talk about, but, but let's say they're all aimed at maximizing the number of profitable transactions that can happen in this industry. So what happens when you say, yeah, okay, we understand they're all interrelated, but for purposes of either defining harm and then subsequently thinking about the remedy, we're going to sort of lop them off into different pieces? I think that, I think that is going to complicate things for the doj. I think that that finding by the judge is going to. She obviously sees it as a constraint on her, on her ability to impose remedies here. For example, I think she's made it pretty clear that maybe divestiture of DFP makes sense, but I don't. Am I correct that in earlier iterations of the DOJ's request for remedies, they also proposed divestiture of AddX or I'm sorry, not of the bicep.

B (39:43)

Yeah, the network business. Yeah, they didn't get that one right.

A (39:47)

But that's, I think in part because of the way she sort of, you know, determine the case. It goes to your point that the, the, by separating the pieces of this market so much and, and not find. And in this case finding that there wasn't monopolization of the buy side market, it meant that it's going to be really hard to get a remedy that, that really clearly constrains Google's ability to compete on that side of the market. So, I mean, she's obviously grappling with that. And then to my earlier point, on the other hand, I don't think it's really possible to so clearly carve them up that. I mean, the reality is that anything you do in any part of this vertical stack is going to affect the other parts. Right. I mean, one of the things that I find really interesting about this case is it's entirely litigated from the perspective of the publishers. Right. There was some testimony from marketers.

B (40:58)

But.

A (40:58)

The reality is that at some level what's good for publishers is bad for buyers and vice versa.

B (41:07)

There's a tension in the market for sure. And the hope is that publishers can sell their ad space so that they can survive and maybe even thrive, but at this point point just survive. And yeah, advertisers want a good price and they want outcomes.

A (41:20)

The publishers want that. I mean the, if you're Judge Brinkenma and she did talk about this and you know, she definitely created space for this to happen. She wants those, the, the idea that publishers should get more revenue in her mind and in the, in, in proper antitrust analysis that should happen at the expense of Google's take rate. Google's profit in the middle. It shouldn't really happen. I mean again, it's inevitable. But it would be a lot harder to sort of countenance both the finding of liability and the remedy in this case if the sum total of the effect was we're just transferring wealth from advertisers and giving it to publishers. That's not what antitrust is supposed to do. Now that may accompany a increase in efficiency, competition, output, lots of other things that antitrust is concerned with. But that's going to be, that's a concern, I think. And, and it's interesting to me how, you know, how I didn't, I mean, I guess Google tried to, to, you know, sort of put focus on the advertisers as much as possible. I think I would have done more of that because I think I would have tried to, to set up a, a dynamic in which whatever the publishers get, whatever they think is a harm is actually Google's effort to, in the first instance, and this is what Google did say to you know, sort of balance these two sides of the market. But you could even say it's, look, in some cases Google is making decisions to benefit advertisers at the expense of publishers. In some cases they're doing exactly the opposite. That's not inherently problematic in any way. In fact, that's just a complication of being in the middle of this kind of market. And, and no matter what you do, no matter what remedy you impose here, whoever's in the middle is going to have that exact conundrum in the future.

B (43:22)

Well, let's talk about the woman who has to make that decision. I think Judge Brinkham is great. She's so sharp. I think she understands ad tech better than some actual practitioners of the dark arts of online advertising. She's also very quotable. I mean, you, you referenced a quote that was cited by my colleague James Hercher, who was there for the first three days of the, of the remedy phase. But she's been quoted on this by, by others, too. She was just getting really annoyed at people rehashing all the stuff she already knows. And so at some point on day two, she just said, like, this is to some degree window dressing. Let's get to the mannequins. She was just so, so colorful. But like.

A (44:07)

Yeah, because I. You hear that phrase window dressing all the time. It never even occurred to me that there, you know, like, there was a. Another layer to that, that she found.

B (44:16)

It, that she found the next layer. And, and yeah, like, she's really ready to get to the meat. And that is what's happening now. Like, day three, the Economist started coming out, and that's who she wanted to hear from. But how much does the approach of a specific judge influence the outcome of this phase of the trial? Judge Mehta is a very different person. He was the judge in the search trial. I think a lot of people were dissatisfied with that outcome. Of course, it depends on which side you're on. But I think a lot of people saw it as a little bit like, weak or whatever. But, yeah, I mean, how much does a judge's personality and just Judge Brinkham has no nonsense approach influence how she'll make her decision? Because obviously, obviously, she also has to adhere to the law.

A (45:06)

Right, Right. Well, I, I think a couple of things. One thing to note that I'm sure your listeners have heard, I don't know if they know what it means that, that the, the court that Judge Brinkham is in, the Eastern District. Eastern District of Virginia is so called rocket docket. Oh, yeah, right.

B (45:26)

I love that. Which is she was moving it along.

A (45:31)

The poster child for the rocket docket. And, you know, the one consequence of that. So I don't know how much of this is attributable to that and how much is attributable to her per se, but as I said, it's clear that she really internalizes that rocket knocket idea. And one of the consequences of that is, you know, is. Is cutting out a lot of what amounts to rhetorical fluff. And I'm sorry that that's not the right way to. I shouldn't say that, because my whole point is it's not always fluff. It's. It's, you know, sometimes that the. The rhetoric is important to help the decision maker in this case, Judge Brinkoma, understand the market and understand the consequences of the decision she's. She's making. And I think it. And. And it tends to be defendants who are in a position to need more of that than plaintiffs are. Plaintiffs have a. A number of advantages when the government is the plaintiff in any trust case. And, you know, defendants are much more in a position of, look, let us. If you give us three days, we can explain to you exactly why we made this decision and why it's not what you think it is. And, you know, and a court that, you know, and then judge who says, I got the gist of it. Let's move on, is, you know, it's perhaps sort of putting the defendants at a. At a disadvantage. Now, on the other hand, Judge Pinkham really is sharp. She probably doesn't need that shit to. To actually understand stuff. In fact, as someone who's been an outsider to this industry, I have to tell you, and I've written a couple of papers on it, I have spent a lot of time trying to understand this industry. I feel like I understand next to nothing. And then I read her opinion, honestly, so. And I'm like, this is actually, you know, one of the best sort of simple, accessible summaries of how this, as far as I can tell, I mean, there are some things in there that I think were errors and. And, you know, again, in my. In my. In expert opinion, but for the most part, she really got it, and she's clearly really very sharp. And so, okay, all of this is a setup to saying it matters, because sometimes that lengthy discussion that goes beyond the sound bite is necessary for the judge to understand what's really going on sometimes. And this is where Judge Bacon obviously thinks it goes, it's about confusing the judge, and she doesn't want that to, you know, to both delay things that are happening, but also to, you know, literally confuse things. All right. You know, I think that ultimately she understands things pretty well. Her, you know, cutting off discussion. It happens on both sides. She's obviously not thrilled with the DOJ either, presenting a lot of evidence that she thinks is irrelevant. I think it ultimately helps sort of helps the plaintiff, potentially hurts Google. But in this case, she seems so sharp that. That I'm not sure, you know, I'm not sure that curtailing that sort of those rhetorical. Rhetorical flourishes is going to deny her access to important information to help her decide the case. I mean, look, she actually wants to hear from, like, engineers. I mean, she obviously thinks that she can understand what the engineers are saying, and I believe that she can. And it's very impressive. And it's not that common among judges, especially in antitrust. So in this case, there are two big sort of areas that judges often have issues with. One is the technicalities of the industry itself, and then there's the economics. And both of them are complicated at their core, they're very mathematical and very specialized. And. And I think she's doing a very good job of sort of understanding both. That's not the. That's not the norm. I don't want to disparage Judge Madder. He actually, his opinion was great, too. Frankly, I didn't agree with it, but. But it was one of the best written opinions I've seen in a long time. And again, he clearly got a handle, I think, on. On how that market worked. I think he. In his case, I think he made some mistakes in the law, but that's impressive. And I don't know, I guess then. Now what I'm saying is, well, you know, it can work both ways. You know, you can be. They're both, you know, at the top of their game. That is not the norm in antitrust opinions that I've read.

B (50:11)

I do have to ask, though, if you remember, what did Judge Brinkama get wrong in her decision, do you think?

A (50:17)

Oh, yeah. In particular, what I was thinking of when I said that was. Oh, I. I think the whole characterization of first and last look is it's not that it's wrong. It's that the. The assessment that it was that the purpose of it was to advantage Google at the expense of publishers goes back to what I was saying before. It's. To me, it's just as if not more likely that it was providing a thumb on the scale in favor of marketers over publishers, but not necessarily Google itself. And that's why. So I find the clear characterization of first and last look as mechanisms of monopoly maintenance to be a relic of having separated the market so distinctly that if you were looking at this from the perspective of someone who's trying to balance the interests of advertisers and publishers, it would not be so easy to Characterize those mechanisms that way.

B (51:27)

Okay, interesting. So it was more of a framing thing. I do think, though, that if you talk to any of the publishers that were on the stand and pretty much any Exchange or any company that tried to run an ad server, they would agree with that characterization. But, I mean, it was like their lived experience. Yeah, yeah.

A (51:45)

But again, that they might, they might say differently, but they don't really care if they gain at the expense of Google or advertisers. They, they want to gain. And you know, it's to their benefit to say this comes at Google's expense, not advertisers expense. But I don't think that that's clearly true. And I don't. And if they say that, well, there are probably some who, you know, who, I don't know, who are really upstanding people for the most part. I would assume that of course they'll say that, but. But at the end of the day, they'd be happy to. To decimate advertisers. I mean, not in the long run, obviously. They depend on them in the long run, but, you know, to, to take more of their wealth, whether that's, you know, good for the market as a whole or not. So.

B (52:30)

Well, what. One other question about what Judge Brinkoma has to keep in mind or what she maybe can't keep in mind when she's making her decision. How outcome of the search case impact the ad tech case, if at all. Like, does she get to take the search case into consideration when she makes her decision? Or is that not kosher?

A (52:49)

I mean, it's a, it's kosher. It's a different court. So it doesn't have, excuse me, it doesn't have sort of mandatory precedential effect. In other words, the, the holdings in that case do not bind her decisions or any decisions made in the Eastern District of Virginia. But when and if, you know the, in the search case, if that goes back to the district court, whether it's Judge Mehta or someone else in the, the D.C. circuit or in the D.C. district Court, it would be binding on. So. So on the one hand, she's not required to take account of it. On the other hand, especially in antitrust, where there aren't always that many cases out there, you see this all the time. Courts are looking at what happens in other circuits and other courts to inform what they're doing, and she's clearly going to look at what Meta did and she's going to want to align with him as much as possible, I think. But she doesn't have to, but it's total. It would be totally fine if she did. I mean, I, I can't imagine her, you know, writing an opinion that somehow improperly made reference to what Meta did. Wrote. But I'm sure she'll reference it and, you know, and that. So it's not binding, but it's informative, I guess. Yeah. I don't think she's gonna. She's going to ignore it.

B (54:24)

Also, Meta got to make the decision. She gets to make the decision because there were no juries. Do you think the outcome of either one of these cases would have been different if there was a jury and also poor jury, particularly in the ad tech case. Can you even imagine being on that jury?

A (54:43)

I. That's why I could easily imagine the outcome being different. And there's no way a jury, there's no way we would be saying about a jury what we're saying about Judith Brinkton Law in terms of her ability to understand and, and process this incredibly complicated market. And it doesn't mean that a jury would get it wrong. But, but whatever you think, whatever knowledge you think judges lack, you know, juries have that in. In spades.

B (55:09)

It's literally my job to write about this, and I still find myself googling crucial terms on the regular, so.

A (55:15)

Glad to hear that. So I do it all the time. I actually have a gloss. I've created a glossary, like a running glossary of terms. And, and even that. That in what, you know, it'll come up in some other context. It'll turn out that my sort of definition of, you know, DFP or something actually has to be amended with, you know, five different nuances to make it relevant to whatever. Yeah. So it's. I think it's, you know, I don't know. I mean, actually, I'd like to ask you this question. I. To me, it's a uniquely complicated market. I mean, I guess it's not as complicated as I've been spending a lot of time. The semiconductor manufacturing, also incredibly complicated.

B (55:56)

Good times.

A (55:58)

In a kind of logical way, the ad tech is more. It's like a cludge. It's like everything sort of kludged together how it's evolved. Right. But do you think it's. It's. Am I right? I mean, is it really more complicated or more confusing than other markets, or is it just unfamiliar?

B (56:22)

It might be partially that it's unfamiliar, but I do think this is my pet theory. From having interviewed many people, I think they try to make it more complicated than it has to be in their explanations of it, although it is quite technical. There are so many middlemen and there are so many fees and hidden fees. There are margins, so many different players that take a cut of a dollar spent by an advertiser before it gets to a publisher that shows up. There's like 10 cents of it left. And all of that stuff that's happening in there is a little bit of a cluster.

A (57:04)

And is that a function of the way it's evolved? That's my sense, is that if you were designing it from scratch, you asked this question, but let's ask it in a slightly different way. If you were designing it the sort of ad tech stack from scratch, would it look anything like what we have now?

B (57:19)

I don't think it would at all. I think there would be far, far, far fewer players. I don't know why it has to be so complicated. I also. This is going to sound absolutely sacrilegious, but I don't think that the promise of personalization and data driven advertising has fulfilled itself. And I don't know that it's even as necessary as everybody in this industry says it is.

A (57:44)

I think personalization, open display.

B (57:47)

Well, just in general.

A (57:49)

Well, so I think more so in say search advertising. But.

B (57:51)

But yeah, well, the search is so intent based, but I, I don't really need to be identified. You don't have to know that much about me to know my buying preferences or to broadly dump me into a bucket and for me to make a decision and to buy the thing. Measurement is probably the most complicated part of the entire online advertising industry. But the actual targeting of a person who might buy a thing, I don't know. It doesn't have to be as complicated as. As it is. And I also don't think it has to be as precise as people are trying to make it in order to.

A (58:30)

Provide a return to the provider.

B (58:32)

Return, yeah, you can still sell toilet paper and pasta and all of that stuff.

A (58:35)

There is some economic literature on this that you may have seen that. My recollection is, not surprisingly for economics literature, it is sort of all over the map. There does seem to be some value to targeting. I think Eric Brynjolsson did a paper, he's a really good economist who looks at these kinds of markets. It's always hard to, you know, getting data and understanding data and parsing it is always very hard. But I think he and some others have found that there is some value, you know, but that doesn't dispute your point, which is how much value is there? Like, we could all agree it's. It's better to target than not, but is it enough better that it's worth what you're sort of paying for? And then, and, and as other economists have written, Catherine Tucker in particular has a great paper about the saying exactly what you just said, that measurement is the big issue. And I don't think she's. She says this exactly. But my stuff is, my suspicion has always been, or since I've been looking at these kind of markets, is that if measurement were better, these markets would be a lot smaller than they are now. That there's a lot of.

B (59:49)

You would know who proves. Yeah, you could, if you could prove value, you would know who gives that value. And then you would cut the ones that don't, you know, help you in what you're trying to achieve. And that would be. That would be a much smaller market.

A (60:02)

Exactly. I think that there is something to learn from the search taste with respect to this. One of, you know, one of the big drivers of the remedy in the search case for Judge Meadow was, excuse me, the effect of the remedies that were being asked for on other participants in the market. Mozilla Firefox would be the third most obvious example. In other words, it definitely mattered to him what was happening in markets that were not at issue in the case. That's important because he very explicitly said that. Exact same question came up at the liability trial, and he said that's not relevant to finding liability. But then he says it is very relevant to how we impose a remedy here. And again, you know, Mozilla Firefox is not in the market that was at issue in the search case, but obviously tangentially related. So in that if Judge Brinkerma does the same sort of thing, and while it's hard to say it's sort of legally required, it's kind of logically required. She'll do that. She'll take account of the effect on others. You know, I think that it will be interesting to see how much the idea that everything is sort of moving on to other markets resonates with her and whether she therefore says, well, okay, if we're, you know, potentially harming this particular market that we're actually trying to, to protect, maybe that's okay as long as. And this is a point you raised before, as long as the remedies do something to prevent the same thing from happening in the markets that are becoming more important in this, in this space. You know, I don't, I don't know how well that exactly translates into to what we've heard so far. But I would think that that's exactly the way she'll be thinking about system. It is here. Yeah. I don't. Now, having said that, I think given the way she decided the liability ruling, she's not going to accept the idea that Open Display is a dying business and it doesn't really matter what she does there. And what she has to focus on is what's happening elsewhere. She is very much going to focus on that market, but I don't think she's going to ignore other markets as well.

B (62:31)

Interesting. Well, last question. Are you a betting man?

A (62:37)

Not really, but I mean, I'm happy where there is no literal money at stake. I'm happy to bet.

B (62:43)

Well, what's the outcome going to be here as we record this? We're not even officially, I think halfway through the remedy phase because the DOJ is still making its case. Or maybe they're wrapping today. Today is Thursday and then Google gets its turn. But yeah. What does your gut say?

A (62:58)

One thing my gut says is that although she allowed for two weeks for the remedies phase, I don't think it lasts the full two weeks.

B (63:04)

Oh, it ends before Yom Kippur for sure.

A (63:07)

Yeah, absolutely. I agree. I would not take what happened in the search case as much of a signal at all about what's going to happen in this case with respect to the possibility of a structural remedy. That said, structural remedies are really undispamored for reasons we talked about, they're rarely successful. Google is absolutely right that the effect, the initial effect of a structural remedy here is decreasing competition in one of the markets. I wish I had a good brink and the like analogy here. What happens when a flood is flooded out? I mean, you're going to have a flood and they're all going to go bankrupt and it's going to be.

B (63:47)

You can see all the mannequins.

A (63:49)

Yeah, yeah, exactly. That's it. There'll be a bunch of. You'll have a flood of mannequins floating through the. The ocean. I think she's attuned to all of that. And so, I mean, you know, this is an easy. This is like a. This is how, you know, I'm sort of an economist because it's always on the one hand. On the other hand, it's a lot of.

B (64:09)

That said. Well, that said.

A (64:10)

Yeah, that said, let me tell you this. So I guess the important thing to note is structural remedy much more likely here than it ever was. It was never a possibility in the search case. It's very much a possibility here. And if it happened, would never be. Wouldn't be overturned on appeal. In the search case, if it had happened, almost certainly would have been overturned on appeal. And that has to matter to her. But on the other hand, there's this concern that, that doing so would be inherently decreasing, at least in the short run, competition. And that's not what she's supposed to be doing. So again, that's going to inform her decision, too. If it's really hard, if it's easy, like the grocery store cases you mentioned, the mergers, grocery store cases, spinoffs are really easy. These are completely separate stores. You just say, okay, sell that store and there it goes. I mean, there is no difficulty there. I think there is a little bit more difficulty here. But it's also possible, as I said, that a behavioral remedy is more technically difficult. Even though spill off. All right, my betting is. Yeah, my betting is.

B (65:19)

I don't know here, I'll sweeten the deal. I will buy you a topo Chico if you're right.

A (65:28)

All right, my bet's going to be no scale off of conduct remedies that are. That go beyond what Google's asking for in a couple of ways, but are mostly reflective of that. I mean, look, the Google conduct proposals are directionally correct, like you were saying earlier. They may not go far enough to sort of ensure the kind of, you know, realignment of incentives. So there are probably a couple of additional things that she could do, maybe taking from the DOJ remedy. But I don't think it'll be too far off from people's behavior. What do you think?

B (66:10)

I don't think there's going to be a divestiture. I would be extremely surprised. But I guess stranger things have happened. And like you said, it's more likely in this case than it ever was in the search case. That was just like a pie in the sky. So I guess, Jeff.

A (66:34)

We will have to do this again after that decision comes out.

B (66:38)

Our hair will be grayer. I will check in with you then.