Causation Confusion: Judge Mehta Got the Standard Wrong But Does it Matter?
I kick off a four-part deep dive on Judge Mehta's landmark US v. Google decision. We start with the part of the opinion the D.C. Circuit is most likely to reject: causation. But will it matter?
Welcome back to Competition on the Merits! Did anything big happen in the antitrust world last week?
I’ve got a few topics I want to cover soon. But it seems like a good time to do a detour for a deep dive on the US v. Google decision. It is a landmark case. It is extremely important. It is a big win for the DOJ and the States (both the Trump DOJ that filed the Complaint and the Kanter DOJ that got the win – and all of the staff and States that worked on it). Judge Mehta took the time to author a very careful and clear opinion.
It’s a monster of an opinion. So we are going to try to break this into some smaller pieces. First things first – we are going to carve out as “relatively uninteresting” a few parts of the opinion.
Google has monopoly power in general search? Fine. Yawn. Not a violation of the law by itself – whether one agrees with the proposition or not (and I do) – because one needs to show exclusionary conduct. That is whether the action is and so we will spend most of our time there.
I’m also going to carve out the discussion of the Section 2 refusal deal claim involving SA360. DOJ and the States (I’m just going to say DOJ now – with all due respect to the States who played a large role here) did not get close on that claim. They should not have gotten close on that claim. And I think the prospect of a reversal on appeal is slim to none. Trinko is and remains your favorite antitrust case’s favorite antitrust case.
Enough about what I’m not going to cover. Here is the plan. But I reserve the right to change my mind on whim, fancy, or further reflection about the case!
PART I
Causation Confusion: Why Judge Mehta Got the Causation Standard Wrong and Whether it Matters
PART II (TOMORROW)
Foreclosure Fun and Games
PART III (THURSDAY)
Procompetitive Predicament: Why Does Google Pay Billions for the Default If It Doesn’t Foreclose Rivals?
PART IV (TBD)
Remedies Revisited: Does Google Want an Injunction Barring Search Default Deals? Does Apple?
In the meantime, this seems like it is as good a place as any for a general disclosure. I’ve done a good bit of work for Google over my career in a variety of roles. Google has funded some academic papers – disclosed in each instance. I’ve also previously advised them in my role as a lawyer and antitrust expert. I did not work in any capacity on this particular case at any time and they are not a current client.
Let’s get started. Today will be the longest entry – I think – because we need to do a good bit of primer work on Section 2 monopolization law to understand the decision. But we will get there together. Here we go.
Oh, and please subscribe, tell a friend to subscribe, or upgrade to a paid subscription.
FRAMING MONOPOLIZATION UNDER SECTION 2 OF THE SHERMAN ACT
The debates over proper antitrust tests for Section 2 of the Sherman Act are as old as the act itself. They have become intense in recent years – but they have been around for a very long time. There are relatively unserious corners of these debates. For example, a few still call to abolish all of Section 2 of the Sherman Act out of the incorrect belief that a monopolist cannot do anything to harm competition or out of an ideological opposition to any form of antitrust law. On the other side, and equally unserious, are those who take the position that earning monopoly power through innovation or outcompeting rivals should itself be unlawful.
Ignoring the two ends of the spectrum, the core monopolization debates have been about a fundamental, and really hard, problem. Antitrust has easy cases. Price-fixing is a good example. Why is it easy? Naked price-fixing is always or almost always anticompetitive. From a competition standpoint, there are not many good reasons to enter into an agreement with a rival to fix prices or output. Antitrust law has adapted by summarily condemning such behavior and reducing the burden of proof faced by the plaintiff. The “per se rule” in antitrust is essentially a simple blanket rule of illegality for a subset of conduct we have learned always or almost always harms competition. In those cases, requiring the plaintiff to show detailed proof of harm (via an increase in prices or reduction in output) to go through the trouble of defining markets is unnecessary because we have other information that already tells us the likely effects of the conduct are bad.
Mergers come somewhere in the middle. Some mergers harm competition. Some generate welfare benefits by reducing the costs of the merging parties or allowing them to make something together that they could not do independently. For those sorts of cases, antitrust rejects summary condemnation and instead asks the plaintiff to prove that this is one of those mergers that creates market power; and that the negative effects outweigh any of the benefits. Different people have different views of how common mergers that generate anticompetitive effects are, how sustained those effects are, and how likely consumer benefits are to arise from mergers. Some have different arguments about how those views should impact the burdens parties face when litigating mergers, when we should presume harms or benefits, and what quality of evidence should be required to establish the requisite effects. But the basic structure of antitrust law in these “medium” difficulty cases is that we acknowledge both effects can happen and force each side to bear the burden of showing the relevant effects: the plaintiff must show this transaction harms competition. If the plaintiff succeeds, the merging parties bear the burden of showing the merger will benefit consumers.
Monopolization is inherently more difficult when it comes to line drawing for a few reasons.
The first is that we know with certainty that much of the conduct at issue in monopolization cases benefits consumers. Yes, it has the potential to harm competition as well. We know that a monopolist using, for example, an exclusive dealing contract can harm competition. Economic theory teaches it is possible. And we do have a handful of empirical examples. But we know with certainty that contractual provisions like exclusive dealing and tying, or vertical mergers, or a firm’s decision to refuse to deal with rivals, or to lower prices often generate benefits for consumers and are a result of competition rather than a distortion of it. We know this because we see small firms — firms without a single iota of monopoly power — enter into the same agreements. That tells us they often have other purposes. Of course, we also have tons and tons of economic research that tells us the same.
The second is that it is really hard to distinguish the “good” for competition examples of, say, exclusive dealing, from the bad ones. The contracts themselves look the same. This is not like the easy cases. We know a cartel agreement is very likely to have bad effects. But if I put an exclusive dealing or tying agreement in front of you we do not have much of an idea about its likely competitive effects.
The third is that society has to be really sensitive about errors in the monopolization context. If we punish a few too many naked price fixing cartels and get one wrong, i.e. punish something that looks like a cartel but is not anticompetitive, what is the cost? Not a ton. The so-called “false positive” or “Type I” error costs are small. What do consumers lose out on? But that isn’t the case when we are talking about Type I errors in the monopolization context. If we screw up and punish a discount that is actually procompetitive (like most of them) or an exclusive dealing contract that benefits competition – we deter conduct that helps consumers. Mistakes here are costly. Of course, they are costly in the other direction as well!
The point here is that line drawing is not only much harder in the monopolization context, it is also much more important because errors are more likely (it is hard to distinguish anticompetitive contracts from procompetitive ones), and they are more costly. This area of antitrust law – monopolization – draws the most debate and the most intensely held views for a bunch of reasons. But at least one of them is that when cases are hard to figure out, people naturally rely upon ideology and priors for evidence. Another, of course, is that a lot of people use products whose sellers have monopoly power. And this is certainly true in the modern monopolization cases: Google, Apple, Qualcomm, Meta, and so forth.
But what does antitrust law do in the hard cases? An anticompetitive presumption such as in the price-fixing cases does not make economic or legal sense. The probability that the underlying conduct is anticompetitive is much lower. Anticompetitive price-fixing is the rule, not the exception. That is flipped on its head in monopolization cases. Most exclusive dealing or tying or discounts are not just competitively neutral, they are good for competition and consumers. Instead, courts in Section 2 Sherman Act cases have developed a burden-shifting framework that – consistent with economic learning and evidence – has two features: (1) it requires plaintiffs to prove THIS exclusive dealing contract or tying arrangement is the exceptional one that harms competition in order to distinguish it from the competitively benign ones; and (2) it is open to procompetitive explanations from the companies deploying those contracts, though they bear the burden of demonstrating their effects.
The most well-known description of this burden shifting framework comes from the DOJ’s turn-of-the-century monopolization case against Microsoft, which plays a large role in Judge Mehta’s analysis. Of course, Microsoft is also from the D.C. Circuit – but it is also widely regarded as a state of the art approach to Section 2. Let’s take a look at how they lay out the court’s approach to monopolization claims so we can follow Judge Mehta’s analysis in the Google case.
It is pretty sensible stuff. And as you’ll see, the fundamental challenges of monopolization cases described above play a large role in how courts approach these cases. Here’s Microsoft:
Whether any particular act of a monopolist is exclusionary, rather than merely a form of vigorous competition, can be difficult to discern: the means of illicit exclusion, like the means of legitimate competition, are myriad. The challenge for an antitrust court lies in stating a general rule for distinguishing between exclusionary acts, which reduce social welfare, and competitive acts, which increase it.
From a century of case law on monopolization under § 2, however, several principles do emerge. First, to be condemned as exclusionary, a monopolist's act must have an “anticompetitive effect.” That is, it must harm the competitive process and thereby harm consumers. In contrast, harm to one or more competitors will not suffice.
Second, the plaintiff, on whom the burden of proof of course rests, must demonstrate that the monopolist's conduct indeed has the requisite anticompetitive effect. In a case brought by a private plaintiff, the plaintiff must show that its injury is “of the type that the statute was intended to forestall,” no less in a case brought by the Government, it must demonstrate that the monopolist's conduct harmed competition, not just a competitor.
Third, if a plaintiff successfully establishes a prima facie case under § 2 by demonstrating anticompetitive effect, then the monopolist may proffer a “procompetitive justification” for its conduct. If the monopolist asserts a procompetitive justification — a nonpretextual claim that its conduct is indeed a form of competition on the merits because it involves, for example, greater efficiency or enhanced consumer appeal — then the burden shifts back to the plaintiff to rebut that claim.
Fourth, if the monopolist's pro-competitive justification stands unrebutted, then the plaintiff must demonstrate that the anticompetitive harm of the conduct outweighs the procompetitive benefit.
That framework is a useful one for walking through the Google decision. But a theme in the Microsoft framework, and a theme in my own discussion of Judge Mehta’s analysis, is that monopolization case law squarely places upon the plaintiff the burden of distinguishing allegedly unlawful acts as anticompetitive as opposed to procompetitive. The difficulty of doing so is well-known. But the plaintiff bears that burden of showing that the conduct harms more than a competitor, as all forms of legitimate competition do, but also competition and consumers. Uncertainty in monopolization cases is generally read in favor of the defendant because the plaintiff bears this burden, unlike in price-fixing cases where uncertainty is read against the defendant because we generally know the direction of the economic effects.
FRAMING US v. GOOGLE AS A MONOPOLIZATION CASE
At its most simple level, the Google search case was always going to be about whether Google could provide a compelling answer to the question: “Why do you pay $20 billion dollars a year to Apple for the default search position if not to exclude rivals?”
DOJ’s answer is a simple one: the payment is for the exclusion of rival search engines. In more economic terms, the payment to Apple (and Android OEMs) is to maintain Google’s monopoly position in search for a longer period of time by depriving rivals of the opportunity to compete for minimum efficient scale.
Google’s answer is not so simple. Google responds that search defaults are not sticky enough to keep eyeballs on Google search rather than a rival’s product and thus DOJ’s anticompetitive answer cannot be correct. But then why pay? Apple has created a really valuable ecosystem and, like a large retailer, its shelf space is incredibly valuable. Because many consumers do not care enough about which search engine they use to switch from the default, Apple can “sell” access to those searches to the highest bidder. Those are very valuable for Google – and so long as Apple can credibly threaten to deliver them to another bidder, Google should be willing to pay for them.
The answer to why Google would pay Apple even when it has the best search engine is the same answer that Coca-Cola gives in cases challenging its slotting fees (payments for premium shelf space). Some consumers will find Coca-Cola whether it is in the eye level shelf space, the bottom of the shelf, or hidden in the produce aisle. Some who prefer Pepsi – which is obviously an inferior drink (joke) – would do the same. But the “marginal” consumers will buy what is at the eye level or on promotion. The supermarket really does not care whether Coke or Pepsi makes those sales. Its margin is the same no matter who wins. But Coke and Pepsi care a great deal and compete for those marginal sales just like Google and Bing compete for marginal searches created by the default. I cut my teeth in industrial organization economics working on understanding the economics of shelf space arrangements and their competitive effects, including the conditions under which they can harm competition.
The point is this is not a new question in antitrust. Competition for distribution is common. Shelf space payments for eye level shelf space, or end caps to induce impulse sales, or even exclusive distribution in some retail settings are very common. And despite the “high tech” atmospherics around the case, DOJ and Google’s dueling answers to the question are really not all that different from those in the shelf space or other exclusive dealing cases.
WHY DO WE CARE ABOUT FORECLOSURE AND WHAT IS IT, ANYWAY?
The shelf space analogy is very useful for understanding both DOJ and Google’s positions about foreclosure. Foreclosure of rivals is harder in the shelf space context since, presumably, a product manufacturer like Coke has to foreclose a lot of shelf space in a lot of different supermarkets in order to deprive Pepsi from having access to enough of it to prevent it from achieving minimum efficient scale. The entire anticompetitive theory rests not just on affecting the rival’s scale, but preventing it from reaching minimum efficient scale. Every time Coke competes to steal a sale from Pepsi, it affects its scale. So we must mean more than that lest we equate competition with foreclosure. But if Coca Cola can find a way to prevent Pepsi from competing for minimum efficient scale (the point at which Pepsi’s average costs are minimized), Coke can raise Pepsi’s costs and make it a less effective competitor moving forward. That is the whole idea. It is the idea in the DOJ’s complaint. So what we are really talking about when we say foreclosure in the antitrust context is whether Google has tied up distribution in a way that prevents the rival search engines from competing at scale. Not because we care about their scale for the fun of it. But because if the rivals’ costs are raised, they may be less effective competitors and thus allow the monopolist to maintain its position for longer.
Note again, foreclosure is necessary but not sufficient for an antitrust violation. Foreclosure generally means the rival is harmed. But Coke can harm Pepsi in all kinds of ways that help consumers (make a better product, reduce the price and steal sales, etc.). Same with Google. So, in general, antitrust law requires a plaintiff to bear the burden of showing both that the rival or rivals were foreclosed and that the foreclosure created an anticompetitive effect on the marketplace. Again, this requirement is to distinguish the myriad cases where a rival is harmed but consumers are made better off – which is part and parcel of market competition – from the anticompetitive case where both are made worse off. The plaintiff must always bear that burden.
In the search context, one difference is that foreclosure of rivals might only require securing contracts with a single distributor, Apple. On the other hand, the smaller number of distributors makes that story a bit trickier for the DOJ to tell its anticompetitive tale. Why would Apple create a monopolist to deal with as an input provider? Isn’t Apple creating the auction for exclusives in the first place? And if whoever wins the Apple auction is foreclosing rivals – what the heck does foreclosure mean? If Bing and Google compete and Bing won the auction, would Bing be foreclosing Google?
Does foreclosure mean “winning the competition for distribution?” Is every Coca-Cola sale a unit foreclosed from Pepsi? Isn’t that just competition? Or does it mean something more than that? Antitrust law embraces competition to get on the shelf — so-called “competition for the contract.” But sometimes it calls the outcome of that competition foreclosure. The line between the two is hard to draw. The cases try hard to make sensible distinctions. Judge Mehta goes to great lengths to try to do so as well. But these are hard issues and the cases, frankly, are not a model of consistency in how they treat these issues.
My second Competition on the Merits entry is going to focus on Judge Mehta’s foreclosure analysis – which is really the core analytical section of the opinion and where a lot of the action is. We will talk about what he gets right, what he gets wrong, and — of course — whether any of it should give Google any hope on appeal. I’ll save most of the foreclosure section for my next entry — but because it is an overarching theme of the case, it is useful to frame some of that now.
CAUSATION CONFUSION
I want to lead off my discussion of US v. Google with causation rather than delving into the weeds of the foreclosure analysis in Judge Mehta’s opinion. That comes tomorrow. Of course, the two are interrelated. The DOJ bears the burden of showing that the allegedly anticompetitive conduct caused anticompetitive effect. And the showing of substantial foreclosure is indirect evidence of competitive harm.
First, let’s start with understanding the DOJ’s theory of harm and build to understanding the causation standard and what the court found and did not.
The basic theory of harm is easy enough to understand. The DOJ alleges Google’s distribution contracts with Apple and Android OEMs lock enough of a critical input (distribution) that search rivals cannot achieve minimum efficient scale. The theory — as it goes — is that rivals like Bing and others have their costs raised, are rendered less effective as competitors, and Google is able to maintain its monopoly position for longer than it would in the absence of the agreements.
There are a few key ingredients to these so-called “raising rivals’ costs” theories of harm. Of course, the defendant must possess monopoly power. And while Google contested that at trial — the conclusion that Google possesses monopoly power in general search is not surprising and is probably correct. Here’s how DOJ articulated its foreclosure theory of harm in the Complaint:
That’s a fairly standard formulation. The substantive requirements of DOJ’s theory of harm at issue at trial were:
Did the agreements actually foreclose rival search engines from the ability to compete for minimum efficient scale?
This is where a lot of the action is. And we will save that discussion for tomorrow.
The second key ingredient is demonstrating the causal nexus between the allegedly unlawful conduct and the demonstrated competitive harm. We are going to focus on that for now because I think it is the place where Google has the strongest and most interesting argument that Judge Mehta was just not just wrong on the facts, or his analysis, but is wrong on the law — and legal errors are the sort of thing appellate courts are most interested in.
IS THE CAUSATION STANDARD FOR MONOPOLIZATION CASES REASONABLY CAPABLE OF CONTRIBUTING OR BUT-FOR CAUSATION?
This is where the rubber hits the road, as they say. Judge Mehta purports to adopt the Microsoft burden-shifting framework described above, including, Judge Mehta says – its approach to causation. And let me say here as I will a few other times this week: kudos to Judge Mehta on clear, well-written, and thorough opinion. I’m going to say a few parts are wrong. I’m going to say a few parts are right (e.g., Google has monopoly power). But sometimes antitrust opinion reading requires tea-leaf reading and Ouija boards and the like to understand what the court was trying to do, much less whether or not it is correct as a matter of law, or persuasive on the facts. Judge Mehta wrote a careful and clear opinion in a remarkably important case.
That said, I’m going to start where I think the most clear error of law lives in Judge Mehta’s opinion. He gets the causation standard wrong. It’s going to take a bit of work to explain why. But that’s why you’re here. So let’s go.
Here’s what Judge Mehta says:
So far so good. Pay attention to the “reasonably appears capable of making a significant contribution to maintaining monopoly power” standard. Let’s call it the “reasonably capable standard” or “RCS” for short. Judge Mehta says we can apply the rather low bar of simply asking whether the conduct at issue is reasonably capable of contributing to monopoly power to infer causation.
He cites Microsoft for the “causal link” and a well-known treatise for the RCS standard. In the next paragraph, Judge Mehta rejects a “but for” standard for causation. Here it is:
Let’s talk about causation in the D.C. Circuit’s Microsoft analysis Judge Mehta purports to adopt. Recall that the DOJ’s theory of harm in Microsoft was a foreclosure-based theory, but it worked a bit differently. In Microsoft, the DOJ alleged that Microsoft’s exclusionary contracts excluded Java as a middleware threat to help its monopoly position in the operating systems market. The key point for now is that the DOJ theory of harm was that Microsoft’s exclusion of a firm NOT in the relevant market today (middleware), but that might be in the relevant market tomorrow depending upon the evolution of technology, harmed competition in the relevant market.
Microsoft argued that the DOJ made an inappropriate move, i.e. conduct in the middleware market could not plausibly cause harm in the operating systems market. The Microsoft court had to find a way to deal with that unique theory of monopolization. And it did. It used the term “nascent competition” to describe the situation. Nascent was not just a colloquial term. The court used “nascent” as a term of art to describe the situation of middleware – not a substitute capable of constraining Microsoft’s monopoly power today, but a potential substitute at some point in the future. The court carved out a special rule for causation involving Section 2 monopolization claims involving the exclusion of nascent competition.
Well, let’s just read it together.
What is required for a plaintiff seeking equitable relief in a monopolization case to avail itself of the “reasonably capable of contributing” standard? Causation may be inferred in all Section 2 cases. But when “the underlying proof problem” that arises – that is, when it is difficult to reconstruct what the counterfactual world looks like, the court is comfortable inferring causation and the proposition that “the defendant is made to suffer the uncertain consequences of its own undesirable conduct.”
This is a general principle – if the defendant’s conduct made the counterfactual world impossible to reconstruct (in Microsoft’s case, because it had snuffed out nascent competition before it ever had a chance to constrain it) the court will hold the uncertainty against the defendant under certain conditions. In particular, the Microsoft court emphasized that it was willing to infer causation because of two unique conditions at play: (1) the DOJ proved exclusion of a nascent competitor, not just an existing substitute; and (2) the DOJ had already proved anticompetitive effects to the satisfaction of the court.
The Microsoft court articulates the RCS standard is an exception to the general rule that applies in Section 2 cases. The exception was about the difficulty of establishing the counterfactual when the anticompetitive effects were certain and completed, and thus the evolution of the nascent competitor hard to trace out.
If Microsoft is the exception, what is the causation rule? In 2008, the D.C. Circuit decided FTC v. Rambus and the court held the agency to a “but-for” causation standard. In that case, the FTC alleged that Rambus’s conduct in the standard setting process allowed it to be adopted by the standard and earn monopoly power. Rambus argued that with or without the alleged deception, its technology was so good that it would have been adopted either way. Here’s a piece I wrote with Bruce Kobayashi on Rambus and another one on causation in Rambus specifically.
But suffice it to say, the D.C. Circuit – the very court that will review Judge Mehta’s Google decision – applied a but-for causation standard in Rambus. The allegations involved exclusion of an emerging technology that Rambus allegedly excluded from the standard setting process. It was not a case about established technologies and widgets. The D.C. Circuit was absolutely clear that the but-for standard applied to the government’s Section 2 allegations. The D.C. Circuit found no conflict in so ruling – which is consistent with the view of Microsoft as a causation exception rather than the rule.
Interestingly, Judge Mehta has plenty to say about Rambus. He rejects the argument that Rambus requires a “but for” analysis – that is, what would have happened in the world without the allegedly unlawful conduct – when it comes to the plaintiff’s evidence of substantial foreclosure. But he does not really address it at all when it comes to causation. This is confusing because Rambus is about causation, and not foreclosure. So while Judge Mehta takes the view that “Rambus does not establish a categorical rule that the anticompetitive effects of an exclusive agreement must be measured against a but-for world,” that is correct but irrelevant. The argument is about causation. It is not about foreclosure. Judge Mehta is right courts do not REQUIRE but-for foreclosure. Some adopt it because it is useful. But Judge Mehta seems to miss the boat on Rambus and causation.
Let’s take stock. The D.C. Circuit establishes Rambus as the default causation rule for Section 2 and requires plaintiffs to show “but-for” causation. Microsoft carves out an exception when: (1) the plaintiff has proven anticompetitive effects; and (2) nascent competition has been excluded thus creating the problem of proof in reconstructing the but-for world. There is really no other plausible way of reading Microsoft and Rambus together.
You could take my word for it – but how about this instead?
That view accords with my description above: Rambus is the default but-for standard, Microsoft as the “nascent and proven” exception.
Why is that passage important? Because it is written by Judge Douglas Ginsburg (with Koren Wong Ervin) – who is Senior Judge on the D.C. Circuit, the foremost antitrust expert on the D.C. Circuit, and sometimes rumored to have been the author of the D.C. Circuit per curiam opinion (a fact I can certainly neither confirm nor deny). But suffice it to say Judge Ginsburg’s own view on what Microsoft means when it comes to causation is likely to be pretty influential within the Circuit.
So Judge Mehta made a mistake. But is it an important one? A but-for causation standard would certainly require a tighter nexus that he required in establishing anticompetitive effects. But recall the two conditions required to access the RCS standard: proof of anticompetitive effects and exclusion of a nascent competitor. If those conditions are satisfied on the existing record, perhaps this is just harmless error. Let’s explore.
DOES US v. GOOGLE INVOLVE EXCLUSION OF NASCENT COMPETITION?
The short answer is no, not really. But it is a bit messy. Google is not like Microsoft when it comes to allegations of nascent competition. The DOJ’s allegations in Microsoft were entirely about nascent competition – recall that is defined by technology that is not a substitute for the monopolist’s product today but might be in the relevant market tomorrow. It is a potential substitute that might develop into an important constraint on the monopolist – just as middleware posed a potential (certainly, perceived) existential threat to the operating system.
By contrast, DOJ’s Google allegations are almost entirely about the exclusion of existing general search engines like Bing. Bing is not a nascent competitor. Nor is Mozilla. Both are existing substitutes within the well established market. The allegations are not about stopping those search engines from developing into substitutes. They already are.
That certainly reads so far like Rambus, and not Microsoft, should apply because DOJ’s case does not turn on the exclusion of nascent competition. But they do at least mention it. The Complaint talks about exclusion of new and developing search access points. The DOJ certainly ARGUED that nascent competitors were excluded. Judge Mehta’s opinion rejects at least one of those arguments. When it comes to the exclusion of Branch – who DOJ argued might develop into a potential substitute – Judge Mehta concludes: “while there is some evidence that Branch aspired to compete with Google in general search, the nascent-threat evidence here is far weaker than in Microsoft. The trial court there “made ample findings that both Navigator and Java showed potential” as nascent threats. 253 F.3d at 879. This court cannot do the same about Branch.”
The very best the DOJ and Judge Mehta muster when it comes to exclusion of a nascent competitor is with respect to Neeva, who the court argues was unable to develop as a substitute because of Google’s conduct:
But when it comes to the DOJ’s overall demonstration of anticompetitive effects, its arguments concerning foreclosure are almost entirely about Bing. In Microsoft, the crux of the DOJ’s foreclosure arguments were about the distribution of JAVA through Netscape. That is, the foreclosure arose from exclusion of the nascent competitor. It made sense to apply the nascent competitor-based RCS standard for causation. In Google, almost all of the proof involves the exclusion of existing search engines like Bing. Other than the throwaway line about Neeva, the case is entirely about existing general search engines rather than nascent rivals.
Interestingly, and somewhat confusingly, Judge Mehta seems to acknowledge here that his choice of the lighter RCS causation standard REQUIRES exclusion of a nascent rival.
Indeed. What is confusing is this excerpt seems to give away the game. Here it reads not as if Judge Mehta believes RSC is the default causation standard but rather that he accepts it as an exception and argues the conditions necessary to apply it are satisfied. Perhaps. But exclusion of Neeva alone is not the showing upon which the DOJ demonstrated market-wide anticompetitive effects. And it is quite a thin-reed upon which to establish the RSC causation exception in a case otherwise proven entirely about the exclusion of existing search rivals like Bing. At a minimum, the allegations and proofs look far weaker than those present in Microsoft when it comes to the centrality of nascent competition as a constraint upon the monopolist.
For what it is worth, the DOJ appeared to argue (incorrectly) that Microsoft allows the government to avoid ANY causation showing. Here’s DOJ’s lawyer Ken Ditzer at trial:
Well, that is just not correct when it comes to causation. Nor is it what the Microsoft court said or how Rambus treated the causation question years later.
One more fun fact. DOJ itself appeared to adopt the position that its case was mostly about existing competition rather than nascent competition. Here is DOJ’s Dintzer again:
Perhaps the concession occurs because he does not understand the Microsoft RCS causation standard is an exception that requires exclusion of nascent competition. Indeed, his assertion that Microsoft both excuses any showing of causation and that the DOJ’s Google case is “the easier one,” turns the Microsoft logic on its head! The entire reason the Microsoft court watered down causation in the context of cases with proven exclusion of nascent competition (again: potential technologies that are NOT in the relevant market today but might develop into competitors in the future) is because constructing the counterfactual is especially hard. Here, Dintzer seems to not only get Microsoft’s causation standard wrong, but also to concede the government’s case is not about establishing the requirements necessary to invoke it.
Whether Dintzer’s mistaken argument is to blame for the court’s confusion on this point is – well, we cannot reconstruct that counterfactual. But the bottom line is that Judge Mehta seems to have botched the causation standard. The but-for causation standard of Rambus should have applied rather than the Microsoft RCS standard. That is a serious mistake of law. And it is an error that we know at least some on the D.C. Circuit will see and understand.
The real $64,000 question is whether that error is harmless? Or what would happen if the D.C. Circuit reverses and remands in line with a ruling that the but-for causation standard should apply. It is hard to tell. And the record certainly does contain some expert and other evidence regarding what might happen with and without Google’s various default contracts. For example, some record evidence suggests Apple or Android OEMs might have picked Google anyway. In other places, it might be very difficult for DOJ to establish beyond a preponderance that Bing or Mozilla would have constrained Google but for its conduct given Judge Mehta’s findings about Google having superior quality. It is very difficult for me to imagine the D.C. Circuit agreeing that the district court made an error of law on something as critical as the causation standard and concluding it is harmless.
To me, the causation confusion is really the star of the show when it comes to appellate arguments because it is the most clear error of law rather than battle over the facts. Antitrust is a tricky area when it comes to the fact-law distinction because so many of the legal questions (“does this merger substantially lessen competition”?) are inextricably intertwined with fact questions (“will the merger cause prices to go up or output to go down?”). But an error over the causation standard is pretty clearly one that resonates in law. I expect the D.C. Circuit will want to clean up the relationship between Microsoft and Rambus when it comes to causation in any event. And the difference between but-for causation and the “edentulous” causation standard the court applied in Microsoft and again in Google is more than enough to be the difference between winning and losing. I do not have access to the entire record obviously, but from Judge Mehta’s own opinion, the DOJ’s own arguments, and my own reading of the cases, I think the causation confusion is one obvious and likely successful path for appeal for Google.
But I suppose you do not have to believe me. But you should probably believe the D.C. Circuit judge who was part of the Microsoft opinion and has already rejected the argument that the RCS is the default for Section 2 claims.
We mostly spotted the DOJ and the Court their arguments and conclusions about foreclosure food the purposes of this causation analysis. Tomorrow: Foreclosure Fun and Games!