Sean Hollister

I thought we were done — but we came back to decide a few last things.

First, Google tried to argue Epic didn’t have enough evidence for a jury to win. Judge Donato dismissed that, saying he saw “more than enough evidence for the jury to find for plaintiff on each of their claims.”

Second, Judge Donato says the whole jury verdict will follow the rule of reason standard — no per se, not even for the Activision Blizzard Project Hug deal.

Third and perhaps most intriguingly, Judge Donato says he has has been “forced” to investigate Google, on his own, outside of this trial, for conducting “a frontal assault on the fair administration of justice” by intentionally deciding not to preserve chats. But he also says he will let the jury decide to infer whether Google destroyed evidence in this case — he will not issue a mandatory inference instruction in this trial, he says.

He has decided “the best course of action is for the jury itself to decide whether it will make an inference. I am not going to constrain the jury’s discretion by making that inference for them,” he says.

We’re still hearing more argument over jury instructions — I’ll update this post if there’s more that feels notable.

After the ends of Epic v. Google day 13 and day 14, where it became painfully obvious I am not a legal expert, I figured I should probably talk to one!

My biggest question was how the heck we’re going to get from a bunch of competing theories about what market definition should be to the actual questions in front of the jury. Does the judge pick market definition, like one did in Epic v. Apple? Do we have to go with the ones Epic named? Does the jury get to make it up?

Here’s Dan McCuaig, a partner at Cohen Milstein, who spent over a decade in the DOJ’s antitrust division. Not only did he answer my question, he also gave us an elegant summary of the how the process works.

Market definition is a question of fact rather than law. So, in a jury trial, the jury decides what the relevant market is. (The judge instructs the jury how to make that determination.) The jury then determines whether the challenged restraint/activity generated anticompetitive harm in that market. (If not, the defendant wins.) If so, the jury determines whether that same restraint/activity also generated procompetitive benefits in that same relevant market. (If not, the plaintiff wins.) If so, the jury then determines whether there was some less restrictive alternative that could have achieved the same (or virtually the same) procompetitive benefits with no (or substantially less) anticompetitive harm, and, ultimately, the jury balances the anticompetitive effects against the procompetitive benefits to determine, on balance, whether the challenged restraint/activity harmed or benefited the competitive process.

The jury need not find the same relevant market as the plaintiff has proposed in order for the plaintiff to win but, as a practical matter, the jury will always or almost always come out for the defendant if it rejects the plaintiff’s proposed relevant market.

He adds:

Epic v. Apple was a bench trial, so the judge served as finder of fact — and thus made the call on relevant market.

If you take a look at the near-final jury instructions (pdf) for this case, you’ll see the flow sounds like what McCuaig is describing. You’ll also see that a jury seems inclined to consider Epic’s original proposed markets: “Android app distribution” and “in-app billing services on Android devices.”

Bernheim’s penultimate word was to point out there’s a well-known concept that explains that competition naturally happens even in monopolies:

The cellophane fallacy says that even a monopolist will raise its price to the point people will switch to something else, and then business people say we’re competing, but you’re competing at a very high price, way above a competitive level and the antitrust laws are supposed to stop that.

Epic lead attorney Gary Bornstein announced that Epic rests its case. The jury has been dismissed for a week. We’ll be back on December 11th with closing arguments and jury instructions.

I’m working on one more quick post for you before I leave for the day, though — refresh our StoryStream in a couple minutes for that.

Epic v. Google won’t be over for a bit yet, as we’ll come back December 11th for closing arguments and jury instructions before the verdict. But the evidence is all in. Both Epic and Google have rested their cases. See you the week after next!

Epic has made some hay out of the idea that Google took 18 months to match Apple’s subscription price cut, but Google’s lawyer just called out Bernheim for that — because while the Epic Games Store launched with a 12 percent service fee in December 2018, Microsoft didn’t match that with its PC games store until mid 2021.

(Epic has since come back to point out that the Epic Games Store launched with no traction, and would have taken a while to gain some — and that Microsoft did move from 70/30 to 85/15 in 2019. So maybe it influenced Microsoft right away?)

Google hasn’t gotten a lot of points on Bernheim IMO, but it did just seem to nail him for comparing what he calculated as Google Play’s average service fee (26 percent) to ranges of fees (such as 13 percent to 23 percent for the Galaxy Store) in his slide.

Google suggested he should have pointed out that Google’s range was 4 percent to 30 percent, which I have to admit seems right! Bernheim wore it well, saying the comparison was appropriate because Play’s average fee was higher than the top of the other ranges, but he seemed to know he got got on that one.

➡ Tucker compared Google’s fee to fees of app stores on other platforms. “When game console platforms charge 30 percent, their margins are much lower.”

➡ Tucker “uses list prices for commissions,” suggesting that Amazon charges 20 to 30 percent when it actually charges more like 10 percent and the Galaxy Store charges as low as 13 percent, suggests Bernheim.

Bernheim says Google’s economist Tucker didn’t appropriately consider switching costs between Android and iPhone — and not necessarily because it’s hard to switch.

Rather, because a 50 cent increase in prices to the end user (“that’s about 50 cents for the average smartphone user over the life of their phone,” he says) wouldn’t be enough to overcome the switching costs that do exist, given people only switch phones every 2.7 years on average.

Bernheim shows us the chart Google used to downplay RSA 3.0 was showing active Android devices — including devices that had already been sold.

He also says that graphs that show no change after RSA 3.0 are flawed: “Of course you don’t see any change, the competitors are pretty insignificant; it might become significant, Google starts its conduct, and then nothing happens, you stay at this very low level.”

As far as the Macy’s example, he says Macy’s isn’t dominant in its market, and we treat dominant companies differently.

On rev-share: “If you are sharing profits with a competitor you are disincentivizing competition and that’s what the antitrust laws are designed to prevent.”

And he claims that Professor Tucker’s proposed market definition is far beyond what makes sense. “You can’t put transactions that satisfy different needs for different buyers in the same market,” he tells Judge Donato, additionally saying he’s not sure what to make of Tucker’s idea that a licensing arrangement between Google and an OEM is not a transaction.

Epic’s expert Bernheim argues that Google’s expert Gentzkow “ignores four critical aspects of Google’s conduct,” including:

1. Google impairs competition without preventing it entirely

2. Google’s conduct targets comeptition as it emerges

3. Google is dominant

4. Google shares its Play profits with its competitors

“When push came to shove, he talked about whether competition is prevented” rather than impaired, says Bernheim.

The upshot of that: Bernheim believes Epic doesn’t need to prove Google actually blocked competition entirely. In his opinion (for Epic), Epic only needs to show there were no good alternatives to Google Play and Google Play Billing. It doesn’t need to show there were no alternatives at all.

For example, says Bernheim, Gentzkow presented a chart titled “Was Fortnite Blocked?” showing that revenue tanked on Google Play after the app was kicked off the store, but didn’t tank for Android phones that got Fortnite a different way.

But “If off-Google Play was a good substitute for Google Play, you’d see when one drops, the other goes up commensurably.” That didn’t happen: demand stayed stable outside of Play, according to the bar graph we just saw. “There’s no indication that any of the people here are substituting to off-Google Play.”