Get rolling. Can folks hear me? My goal is to be loud enough to wake you up on a Friday afternoon because yes, let's light this sucker up. Okay. Um, this is the audience participation portion of the program. So, what I want folks to do is I want folks to stretch up. Everyone stand up. Come on. Put your arms in the air out. Sit down. Let's get rolling.
Okay. We're gonna... This only works if you participate. Okay. So, um, Fiona reached out and said, 'Hey, I have an idea for an interactive keynote.' And I said, 'Thank God. I've been waiting my whole life for an interactive keynote.' Um, and I hope this works because I'd love to do it again. Um, conferences are great, but it's a lot of people talking at each other. Um, the purpose of this session is to talk to each other, right? Um, and so we're going to do a little bit of background. I'm going to kind of walk you through a little bit of antitrust history just to set the framework. I promise it's not going to be too jargony. I promise it's not going to be too legalistic.
Uh, and then we're going to test it out together, right? We're going to pick some examples. Not giving legal advice. I'm not taking positions. I just want people to start thinking about things in the way that reflect reality. And here's why. When I first got into office at the antitrust division, um, uh, I set out, along with our team, a pretty aggressive and ambitious agenda to revive monopolization, um, to revive unilateral conduct, to address monopolies. There just wasn't enough of that happening.
Um, when I was nominated for—sorry, when the day I was nominated to be Assistant Attorney General of the Antitrust Division. So now think about this: United States Department of Justice. We're the lead antitrust enforcer for the country. Um, we police a $29 trillion economy, uh, in all aspects from health care to technology and everything in between. Guesses. How many civil antitrust cases were active, filed the day I was nominated?
One. See, one. Think about that for a second. One case, right? Why was it? Because there were no antitrust violations? No. Right. It's because we were looking at it the wrong way. We were looking in the wrong places and we were looking, um, uh, using the wrong tools and, um, a lot of the problems were hiding in plain sight. And so I said, you know what, that's a problem. We're going to enforce the law in a pragmatic but effective way. And what I heard was, 'Can't do that.' I said, 'Why?' And the reason was Trinko.
The reason was the courts are never going to listen to you. You might be really ambitious, but you're going to go to the courts and they're going to shut you down. And I said, 'I don't think so. I think the problem is not the courts. I think the problem is the people going to the courts, what they're saying, the cases they're bringing, the law they're relying on.' And Trinko, if folks know here, Trinko, it's a Supreme Court case. Um, uh, the problem with Trinko, uh, it's not a great decision. I don't agree with it, but it's not the boogeyman that everyone made it out to be, right?
Trinko, the problem with Trinko was not just the decision, it's what came after it. The lack of cases testing the boundaries of Trinko. Because the truth of the matter is when you start testing the boundaries of Trinko, it's not quite as broad as you might think. So, what I want to do is I kind of want to walk through what's happened since, put a little bit of that in context for the law, and then I want to talk about the approach as it existed the day I showed up. And then I want to give you a little bit of a window into how we approached antitrust enforcement. And then I want to take that approach together. We have some boards up here and see if we can apply it, just for fun. Okay, does that make sense? Everyone with me? All right.
Okay. So, we're going to start with what I like to think of as monopolization in three-part harmony. There are three critical appellate court cases in the United States that take Trinko and take the arguments that I confronted when people said that the law was not in your favor, test those arguments and see where they lead. And there are three cases. All three of these were decided by appellate courts and all three were, um, uh, appealed to the Supreme Court and cert was denied, meaning that these are really good cases with precedential value. Make sense? All right.
So, the three are Viamedia, which started before I showed up. Um, uh, it was, though, I guess, uh, on appeal as I was showing up and the, uh, court denied cert. Chase Manufacturing case, um, at the same circuit as those who, uh, follow at home, Novell, which was a very difficult decision for antitrust, one where we as the Justice Department weighed in, got a really good result in as an amicus. And then Duke Energy, um, which was the basis in many respects for, uh, the Google ad tech case, uh, in the Fourth Circuit, also recently appealed to the Supreme Court. Cert denied. This case is good law.
Okay. So Viamedia, we're not going to walk through because this isn't a law school class. Although if anyone's interested in a law school class, I teach that now. I'm happy to have you. Um, come see me after. I want to talk a little bit about some of the conclusions in that case, that this was the first case in real memory, appellate decision, that said maybe Trinko is not quite as broad as we think. So Comcast, who is the defendant in that case, I'm going to read here, took the position that after the Supreme Court's decision in Trinko that any antitrust claim based on a duty to deal with rivals bit the dust. Okay. So if it required a private company to deal with somebody else, no matter what, they're saying it's illegal. Does that sound broad?
You know what? I won't, um, specify, but I remember approaching an antitrust agency about a very, you know, an antitrust monopolization concern in, uh, the mid-20s and I was told this. I was told, 'We can't intervene. We think you're right, but the law won't let us because after Trinko, they have no duty to deal.' And any duty to deal case is dead on arrival. This was the prevailing view of the antitrust agencies 10 years ago.
Court, Viamedia said, not so sure that's right. Someone finally brought that argument to the court and said, 'I'm not so sure you're right.' Said we must reject the argument, uh, about, uh, what the law should be. Trinko itself said just the opposite. Not that it bit the dust. It said under certain circumstances a refusal to cooperate with rivals can constitute anti-competitive conduct and violate Section 2. Okay. And then it points to Aspen Skiing, which Trinko recognized was still good law. Okay.
So now we have a situation where duty to deal is not dead on arrival. And this gets really interesting. The court said, um, that if you look at Aspen Skiing, which was the prevailing case on duty to deal, it gave you a whole list, non-exhaustive, of conduct that could be illegal, um, under the antitrust laws, um, even after Trinko. Long-term business relationship that created a joint offering. Right now it sounds like it was in your economic interest. Relationship that existed in the absence of a statutory duty to deal. So, someone actually made the decision to deal with somebody else. It's no longer a duty to deal. You've already decided you want to deal. And once you deal, now all of a sudden different obligations attach.
Um, presumed prior relationship was mutually advantageous. Sudden course of reversal came at a monetary loss of the defendant. Refusal to sell at retail price. You sold that retail price to somebody else. You can see the list here. Unhappy customers. Um, discourage customers from doing business with rivals. Defendant continued to deal with other ones who were similarly situated. Um, and, um, exclusionary conduct was only aimed at the competitor in the market, not everyone else. What it was saying is, like, let's forget all these narrow boxes and categories. Um, instead what it's saying is, you know what, there are different criteria. There are different indicia that help us understand whether the company was behaving in a bad way or for good reasons, and these are just examples of things that you can use that are market-specific to understand that.
Okay. Then it went to cite Novell, Tenth Circuit, in support of its argument, um, that, um, uh, factual dispute regarding the existence of procompetitive justification is appropriate for resolution on the pleadings. Now this one really got to me. Okay, I heard this, that as a matter of law, right, a court can sit there on a motion to dismiss and say, um, that, uh, the facts are not sufficient. There's not an offsetting—you didn't address the offsetting procompetitive justification. Um, and, um, it essentially requires ill effects. How is that possible? Are there any law professors in the audience? And I'm looking at one over there. Okay, professor, tell me what's wrong with this argument.
Explain to folks who are not U.S. lawyers what this means and why when you consider facts is important.
So if facts are very contentious, that's exactly what you said at trial. That's why we have a trial with a judge or have a jury listening to try to understand what are the key facts and then decide connecting with the law. So if there's basically, you know, a big question like is there a procompetitive justification, that's something that the trial will settle. But the motion basically takes you out of trial and the judge can say as a matter of law, you know, I could just ignore the factual dispute and decide this without even going for the trial or the...
Antitrust cases are supposed to be decided by judges and juries, right? And if you plead the facts, the judge and the jury is supposed to assess whose facts are right, whose facts are wrong. But what was happening in antitrust is the judges were saying, you know what, forget the jury, forget the facts, forget the trial. I'm just going to decide what I think the facts should be. That's not the way the system works.
Um, and you can see here the Seventh Circuit saying Comcast proposes that if a defendant merely postulates a valid business purpose, apparently including any purpose, um, that they dream up regardless of feasibility or value, that it ends the inquiry and there's no balancing. And the court said, 'No, that's not right. My job is not—the job of the judge on a motion to dismiss isn't to do the balancing before a trial. It's not to make that decision. The job of a judge is to determine whether the complaint satisfies the law. And then if there are facts that are needed to make that determination under the law, you go to trial and you get the facts.'
So the court said, 'However formulated, this argument is wrong.' Right. You have to balance. This idea that you just have to postulate any procompetitive justification and the analysis is over is wrong. Got it. Um, and, um, the court said quite here: 'Thus, balancing anti-competitive effects against hypothesized justification depends on evidence, right, and is not amenable to resolution on the pleadings.' Well, how do you get evidence in an antitrust case? Anyone know?
Discovery, trial. That's what trials are for. And so if a court's going to shut down a case without having seen the evidence, they're not doing their job under the law. And the Seventh Circuit said that. And the Seventh Circuit said that the argument that I heard when I went to the antitrust agencies 10 years ago and said we could never survive a motion to dismiss, we could never survive, um, uh, under the Trinko decision, was wrong. And had they tried, they would have heard exactly what they heard here from the Seventh Circuit.
Chase Manufacturing, 2023. Um, this is really important, right? Um, little boxes, right? There's an old song, Little Boxes. All right. Um, the antitrust lawyers 10 years ago were fond of putting everything in a little box. It's tying. It's exclusive dealing. Okay, you could do that. Those are real. But is everything tying? Is everything exclusive dealing? Is everything a refusal to deal? No. Business is more complicated than that. So let's let the law actually map real business realities.
Um, and so great quote here. It's throughout the law: 'Anti-competitive conduct comes in too many forms and shapes to permit a comprehensive taxonomy.' You got to look at the facts. You got to call it as you see it. Got to be real. The district court, uh, concluded otherwise, barring the standard from refusal to deal, um, with rivals case law, concluding that it failed to meet that standard. This was another thing that really got to me. By the time I stepped into office, every single monopolization case was being reduced to a refusal to deal in Trinko. It was taking one narrow category of law and saying it applied to the whole damn statute. It's wrong.
Refusal to deal is one kind of label, but the courts have clearly said just like the range of procompetitive conduct is too myriad to mention, the same thing holds true for anti-competitive. And so, um, this I'm quite proud of. We submitted a brief saying, you know what, even a refusal to deal like Trinko, that's a refusal to deal with rivals. The facts in this case involving Chase Manufacturing was a refusal to deal with customers. Say, 'We're not going to deal with you if you deal with a rival.' We wrote a brief that said the refusal to deal law from Trinko and even Aspen Skiing applies to refusal to deal with customers. This is an important legal distinction if there's a refusal to deal with customers or someone else. And the court agreed.
Um, as I said, we all agree. The United States, along with the parties, argued that there was the wrong legal standard, and they agreed. And they've never extended refusal to deal with rivals analysis outside the situation. Nor have we mandated analyzing Section 2 exclusionary conduct under any solitary framework. See where I'm going here, right? You got to open your minds a little bit. Let's focus on business realities. Indeed, as the United States points out, a refusal to deal framework applies to narrow situations often remedied by monopolists sharing their technology with rivals.
So, Viamedia says, 'Hey, a refusal to deal with rivals can be an antitrust violation.' And it gave us a whole list of stuff that, um, uh, are just examples of instances where that might be a problem. And then Chase Manufacturing says that's true, but that only applies to a narrow set of cases. A refusal to deal with rivals isn't limited by Trinko. It's not limited by this doctrinal view of refusals to deal. Okay, let's go to the third, um, voice in our three-part harmony for antitrust, Duke Energy, which is my favorite of all.
So Duke maintains that the plaintiff here cannot string together a series of acts all lawful themselves under relevant facts and claim that the whole is more than the sum of its parts. Okay? So, um, I want to give you an example that I use with my students. Let's say you have two service stations in a rural area, right? They're the only competitors. You have incumbent service station with 99% of the market and then you've got this new small service station rival. And every morning the incumbent, the dominant player, gets up in the morning before anyone wakes up, you know, um, sneaks over. Then one day shuts off the water to the rival. Then the next day it, um, puts a lock on the door so they have trouble getting in. And the next day they unplug the power. Um, and then another day, uh, they take the air out of the tires of the tow truck so you can't go pick up a car. And each day is a little something, right? And that happens every day for a year.
Do you think it is the right thing to take each of those acts and say, 'Well, if taking the air out of tires isn't an antitrust violation, then the whole claim fails'? Does that make sense? Does anyone think that makes sense? Of course, it doesn't make sense. It defies reality. It asks you to overlook facts that are right before you. But that's exactly what defendants were arguing to courts. That's exactly what Duke Energy argued here. It said that each act, if it's legal on its own, needs to be disaggregated and viewed on its own.
And in fact, two law professors of great notoriety, both ends of the, um, political, um, ideological continuum themselves, wrote something like that in a Supreme Court brief on this case, asking them to make that the law. Supreme Court said, 'No, we're not going to take cert.' And it makes a lot of sense here. It's foundational, this is a quote, that alleged anti-competitive conduct must be considered as a whole. When a court is faced with allegations of complex or atypical exclusionary campaign, the individual components of which do not fit neatly within pre-established categories, its application of such conduct tests would prove to be too rigid.
Thus, when a plaintiff alleges a scheme or a course of conduct was anti-competitive, the scheme or conduct must be considered as alleged and not in manufactured subcategories. So, what we're seeing here is you shouldn't take stuff that is logically together and separate it, right? The whole speaks a lot more than the sum of its parts. And so just so you don't think that this is some radical, you know, new-age antitrust craziness, right? This is not going off the reservation. Duke Energy. Duke Energy is OG, right?
Oliver Wendell Holmes in 1905, right? 15 years after the Sherman Act was authored, right? And the interpretation and the intent of the Sherman Act was fresh in the minds of people on the court. Justice Holmes said, it's right here in bold, my emphasis: 'It is suggested that several acts charged are lawful and that intent can make no difference, but they are bound together as parts of a single plan. The plan may make the parts unlawful.' You got to look at stuff in context. You got to look at market realities.
And so you take, um, Viamedia, right? It says refusals to deal can be illegal. You take Chase Manufacturing and it said not every refusal to deal with a rival is treated the same way as refusal to deal with someone else. Right? And then you take Duke Energy and say, well, let's look at the whole plan and get out of these tiny little boxes and subcategories. Now all of a sudden you have a doctrine that can work. And that was what we brought to the DOJ.
So let's figure out how we actually put this in practice and then we're going to get into our interactive portion. So, um, I like to think of it as a tale of two approaches. Um, this is not a trick question. It's a trick question. Um, where do you start in any analysis? It's okay, someone help me. Yeah.
Relevant market. And then what do you do? And then what do you do after relevant? Okay, relevant, right? If you ask 100 antitrust lawyers, 99 are going to tell you, you start with the relevant market. You define the relevant market. Um, I call this the monopolization for—no offense taken—um, for unsophisticated people who don't live in the real world. You define a market. You calculate a market share. And then you figure out which category the conduct is in. Is it tying? Is it an exclusive deal? Is it a refusal to deal? Predatory pricing. Then you figure out the effects. You calculate those effects. You figure out if there's a procompetitive justification. And boom, you just painted by numbers and you have an antitrust analysis. Yes. Tell me what you think about this. Let's get some answers. I hear—yes, sir.
We look for the story and the problem.
Great. Other reactions to this approach, which is, by the way, the prevailing approach. Go ahead.
So, the market narrows down and it doesn't make any sense, especially when you think about, for example, the role of data, you know, flowing through these ecosystems. It's just—it's not reality. It's not based on reality in my opinion.
You look at—I think motivation is super important. You got to try and figure out why a company is buying another company. What's the merger really about? Sometimes it's just, uh, you know, a tax play or a finance play and that's all it is.
But then—tell me why this approach here doesn't help with that.
Uh, you tell me, sir. I don't know. This is what...
Other thoughts?
I mean, it just seems to me that this is not how you do it because we don't just go pick markets for things. You know, I wouldn't just go wonder what carrots are like today and then I go through the carrot, look at carrots, and then calculate shares. You mean, you have to start somewhere, which means you have to identify a problem. So, if this is what you're saying you're doing, it can't possibly be actually what's inside your head. Um, and probably you should be clear about that.
It's because if you believe in behavioral economics, which I believe everybody here does, this is not how people actually decide things, including in the courtroom and including from the bench.
Good. Okay. Um, for those antitrust law professors in the room, and I'm looking at one I can see right here. Um, is this the way that things have been done?
I think yes, actually. I think it's pretty narrow and I think you should start looking at the harm and the market effects and then rebuilding the case up rather than top down.
Doesn't have to be this way.
Yeah. But I would say that lawyers at DOJ, when it was robust enforcing antitrust laws, the problem has often been that they look for what the problem is in the legal framework when they should be looking at what is the problem in the real world and what is the remedy. Not what is the case that I can bring, what is the case that I can win, but what is the outcome that I want? What is—if I want diversification in the search market, how do I achieve that? And then decide what is the case in the market definition that you bring.
Yeah. So, the question is when you start with a narrow question, you're going to get a narrow answer, right? And in this paint-by-numbers approach, it's not necessarily wrong, right? You can see aspects of this in the law, but what we're going to talk about is a different way to do that. And so, by way of example, we rewrote the merger guidelines in 2023. And the first question in the previous merger guidelines was, any guesses, what's the market, right? Um, we took a different approach. We said that's an important question. In fact, legally, we have to answer that question. But it's the wrong first question. The right first question, and we put this in the first paragraph of the new merger guidelines, is how does competition in this market present itself? And will the merger harm competition?
Right? Sounds obvious, but if the first question is what is the market or is this horizontal or vertical, then you're already limiting yourself in terms of the answers. You're already limiting yourself in terms of how you're going to find the answer. It's the wrong approach. It's simple, but it's unsophisticated and it doesn't reflect reality. Yeah.
As well, because I, um, I was an in-house lawyer for a couple of years in a company, and for sure, you know, when the business comes to you as the in-house lawyer, you know, the first question isn't what's the relevant market? In fact, the moment you start talking about relevant market, the business runs away and just—you've lost them. I mean, the reality, the first question is always why are you doing that, right? It's the business rationale for the conduct, for the measure that's contemplated or the new practice. And that's, you know, if the law is supposed to catch the wrong conduct, it has to start at the same point.
I've been practicing for nearly 30 years. I've never heard a company, internally or externally, talk about markets as vertical or horizontal. Ever. It's just not how people talk. So if reality is different, right? And companies talk about reality and competitive reality differently, then why on earth would antitrust agencies or antitrust courts adopt this artificial, rigid terminology? Yeah.
So you're saying that the main question is how does competition work and what's the competitive harm? But then what if the competition is actually bad for the public, for public welfare, for the public interest? Could then—is that not an antitrust issue? Because you think about like in AI, you've got a lot of competition right now to see who can be like more predictive and manipulative, which ultimately—do we want more competition on surveillance capitalism?
We're going to unpack that in a second. Okay. Um, you're just one step ahead of me here. Um, I want to go now to a different approach. Okay. One that I think, based on some of the cases we just read or learned about and my experience, I think is a more authentic and effective way to think about monopolization.
Okay, my clicker is not clicking. I'm stuck on the unsophisticated... Yeah. Can someone click for me? And then we're gonna—this one has a build if I've done it right.
We were doing so well.
If we have no regulation of cars and they kill people, our answer isn't let's have more competition in cars. Our answer is let's make seat belts and airbags and brakes and stuff like that. And then we let competition go down to that level of the safe car we've decided as a society is correct.
How are we doing on the slides there?
I hope not. I hope not. Can you guys click ahead? All right. I'm going to do it, uh, organically. All right. Give me one second here. Okay, one second. Yeah. Yeah. Yeah. Thanks. Just so I have my notes here.
Okay. So what we're going to do in a moment is we're going to talk about a different approach, a more pragmatic approach that starts with realities, and then we're going to talk a little bit about AI, um, and we're going to try to apply it to the real world. And so one of the examples that came up, um, and was suggested we address, is, um, you know, in the antitrust case against Google and the remedies, there was all this discussion about AI, and they were like, well, we shouldn't worry about search anymore because AI and that's going to be sufficient to, um, address, uh, Google's harm. What do people think about that? Is that a good decision or a bad decision? Why?
It doesn't do anything about the harmful conduct.
Right. But let's—harmful conduct, when the old harmful conduct or the new harmful?
Both. Right. And so part of what you're doing when you're—and this is really, really important because this goes to the real-world issue—is when markets are at an inflection point, that is when the incentive to monopolize a market is raised. Right? That is when you want the disruption from a new competitive alternative to be limited. That is when you want to stop in its tracks the new smaller players from taking on the incumbents. That is when antitrust matters most.
And so when you look back at U.S. v. Microsoft, what was happening right when the remedies were being decided in the 2002-ish time frame, what was happening in the world then? Any guesses? The internet. It was going to change everything, right? Um, but that was not a reason to forego antitrust enforcement, which Microsoft argued at the time. It was a reason to make sure that the remedies were forward-looking. And in fact, the court went out of its...
way to make sure those remedies were indeed forward-looking because the disruption of the internet in order to achieve its full effect had to ultimately be free from using anti-competitive behavior. So the power in that case that Microsoft had accumulated illegally as part of its monopolization that was determined to be problematic, we wanted to make sure it wasn't used to squash this new impending... So let's talk about, and let's not look at the unsophisticated approach. Let's talk about what I like to think of. Oops. Let's see. Oh, there we go. Monopolization for unsophisticated.
I think PowerPoint is for unsophisticated people who don't live in the real world. All right, let's talk about monopolization for sophisticated people who live in the real world. Okay. First question that we ask, oh, here we go. All right. What is the source of power? So let's take our example, right, in AI and in search and advertising, what's the source of power?
Data.
Network effects.
Distribution.
What's that? What kind of inputs?
Chips.
So when you're thinking about antitrust, is antitrust about power? Okay, so wouldn't the right place to start in an antitrust analysis be focusing on the source of power, right? That seems like a fair first question. Okay. And so let's say you're dealing with the search example because it's been litigated to the public so we can talk about it. Well, if they have data, they have users and they have distribution and they have infrastructure and algorithms and talent, that's power. Are those barriers to entry?
Well, if that's the source of power and that is the only reason someone's more powerful, maybe that's not an antitrust violation. Maybe it is, maybe it isn't. But the market realities are, if you want the market to be competitive, what is the source of power? Right? And this was always a big issue because when I was at the antitrust division, the staff would often talk about, people would often come in and say, well, it's not about the source of power. It's about a different market where they're leveraging from search into AI or whatever it might be. But let's start with the question about what's the source of power and make sure we address that.
What makes that source of power durable? Data, distribution, network effects, chips, algorithms, talent. What else? What makes it durable? What makes it so that that source of power can't be unwound?
Contracts, IP, flywheel, money.
IP, contracts.
Okay, so these are things that make that power durable, right? And so what are you concerned about? Well, in antitrust, you're concerned about if the company has too much power. If that power is durable, it has a moat around it. Whether you're bringing an antitrust case or you're focusing on a remedy, this is where you want to look. Next question. Is it cumulative? What does this mean? Any guesses?
Yes. Right. So now you have to say, okay, well, what is the cause of that source of power? Is it something that was just one act or is that power acquired and maintained over a long period of conduct? So it might be whack-a-mole, right? If you play whack-a-mole, anyone know whack-a-mole the game, right? Is each gopher its own or each mole its own game? No. The whole game, each hole it's in makes up the entire game, right? And so you have to look at all the different conduct in its totality and the cumulative effect, right? And so if someone wants to compete, they need to accumulate all of these things and they need all of that and they need it for a really long period of time. Does that make sense?
Okay. Like this is the fun part. So what are the economics of the industry?
Explain to me the relevance of capex. Capital expenditures. This is money. What do you need money for in AI? Why is it so expensive?
Data centers. Distribution. Okay. Keep going.
Talent. You're very expensive.
Legal risk.
Keep going. Starts to look a lot like this, doesn't it? Okay. And so now, if you want to light up a new foundation model in your garage, is that something you can do? If you want to light up a new search engine in your garage, is this something you could do? The barriers to entry are really high. But how do markets value this, right? Like, so are markets solely based on profit maximization? Yes or no?
Is it? I don't think it is necessarily. So, there are a lot of antitrust cases that say we should assume that companies maximize profits for the short term.
