Google's Ad Exchange Like "if Goldman or Citibank Owned the NYSE"
Jonathan Bellack, the author of the infamous “NYSE” email, finally took to the stand on the last day of the government’s case-in-chief. Expert witness Dr. Robin Lee shines for the government.
Jonathan Bellack, the author of the infamous “NYSE” email, finally took to the stand on the last day of the government’s case-in-chief.
September 19 (Day 9) began with Judge Brinkema hearing other matters before the Google trial began, which stretched a bit past our usual 9 AM start time. This had Google and government lawyers, witnesses, and press mulling about the halls together. It’s an interesting scene—some of the most powerful people in law, government, tech, and business, along with members of the media, forced to awkwardly stand together in a 7th floor corridor in the US District Court in Alexandria, VA. I guess it says something about the fleetingly egalitarian dynamics of our justice system. Of course, nobody was allowed any electronics, and opposing sides are weary of discussing anything of value in earshot of the other. So there’s a lot of smiling, nodding, and shifting around. It’s like being backstage at a play in some respects.
Once things got underway, the courtroom was significantly more packed today, as full as it has been since the trial began. The specter of the end of the government’s case, and the highly anticipated testimony of Jonathan Bellack, created something of a draw.
One reporter was discussing the stunning notion that the trial could end next week, which Karen Dunn, Google’s attorney would confirm at the end of the day.
Some housekeeping matters popped up to start. Ms. Dunn wanted to bring in 19 Google witnesses. Julia Wood for the United States contested this, disputing 6 FAA witnesses who would testify as government advertisers. The government contended these witnesses are cumulative, live too far away, and the court has already examined the relevant evidence. Ms. Dunn argued hard for bringing them in, and the importance of advertiser (not ad agency, which she distinguished) testimony. Ultimately Judge Brinkema told Ms. Dunn to pick her favorite 2 witnesses, and did not allow the full 6.
This means the government has around 15 witnesses on its witness list. Note that Google is not required to call all of them.
We then heard cross examination of Dr. Timothy Simcoe, an Econometrics and Industrial Organization Expert.
Ms. Dunn asked Mr. Simcoe if Google was “pricing in” its “superior fraud and malware detection” to the rates it charges to advertisers and publishers. In other words, a pro-competitive justification for Google’s higher prices.
Once again, this unexplained thread of “fraud” and “malware” made its obligatory two-minute appearance today.
Dr. Simcoe said he saw no evidence of this, and saw no evidence customers were responding to value in something other than Google’s scale.
Ms. Dunn then veered into the highly technical, arithmetic details of Dr. Simcoe’s analysis.
This testimony focused heavily on the “but for” elements of his findings. In other words, but for the Google Ads/AdX tie, the AdX/DFP tie, and publishers being unable to set floors—would Google’s market dominance be sustained?
We then checked off the daily bingo card free spot (as my colleague Arielle Garcia has humorously referred to this pattern) that is witnesses being pressed by Google about Facebook and Amazon as competitors, only to repeat what even the court security officers in the hallway have probably committed to memory by now, given how often this dialogue comes up—that these were not competitors in the relevant market (display ads) because they did not operate publisher ad servers.
Judge Brinkema then admonished Ms. Dunn for prefacing her questions repeatedly with lines like “…and this will be important in this case…”
Google has been pushing the notion that the market has been growing across the board in absolute terms as evidence against a destructive Google monopoly. Dr. Simcoe had what I thought was a memorable line, saying that “When Standard Oil was a monopoly, the oil industry was also still expanding.”
On redirect, the government re-established that open web display ads do not work in mobile apps (walled gardens) and that Google’s AdX take rate remained stable at 20% while other firm’s take rates fluctuated and declined dramatically over time, a symptom of anticompetitive behavior in a market.
It’s unclear what the rest of this highly technical and at times overwrought testimony on cross examination elicited. Reporters put their pens down. I saw one reading a book. This will be a challenge for Google in its case-in-chief, which begins this Friday, if it is going to rely heavily on economists, statisticians, and data analysis.
In other words, when the government is pulling out smoking gun emails where executives like Jonathan Bellack are comparing their practices with “Goldman or Citigroup owning the New York Stock Exchange,” while you are squabbling with an economist about the “endogeneity” of a dataset in their report, you can see whose case is more readily apparent.
Speaking of Mr. Bellack, his long-awaited testimony began just a few minutes after noon.
Jonathan Bellack has loomed over this case from nearly the beginning. Present as the author or a recipient on some of the most damning Google emails, but kept out of court until now, the anticipation was palpable. The way he was referred to in the gallery, as “a star witness” and “I can’t wait to hear what Bellack has to say” and “I convinced my editor to send me back down for his testimony,” added to the hype. You can’t help but to start imagining this sort of Mr. Potter character, coming through the doors of court fresh from shuttering Bailey Brothers Building & Loan after George Bailey couldn’t afford to keep kicking up a 20% take rate to Google for AdX.
Instead, in walked a sturdy, bespectacled, gray-bearded gentleman with a clean-shaven head, a black suit and a blue button down shirt with no tie, who hustled straight to the witness stand.
Mr. Bellack had a booming, raspy voice—the opposite of most witnesses we’ve had to this point in that instead of speaking too low, if anything, he spoke too loud.
He was confident, quick on his feet, and expressed himself with alacrity and poise.
Unlike nearly every other Google witness, Mr. Bellack did not claim to be unable to remember anything, nor was he evasive. He gave solid, reasonable, and at times strong answers to the questions he was asked.
He made his case well, and did not fall into familiar defense pitfalls.
In one instance, he was questioned about marking email with attorney/client privilege, a bogus practice at Google designed to hide documents related to sensitive competition issues. Other Google witnesses have admitted to writing this on emails without any lawyers copied, when they were not seeking legal advice.
Mr. Bellack pointed to people on his emails who were attorneys, as well as language where he wrote that he was copying these individuals for a “legal POV” on “pricing discussions” and “not just to include them on the email.”
Mr. Bellack attested that mobile platforms at the time of key events in question, did not have the bandwidth or the processing power for header bidding, one non-anticompetitive rationale for not supporting the technology. He put distance between the Google executives we’ve heard from by alleging the decision to stay away from header bidding was a Google leadership decision, and not a product decision, and that the government should question them instead.
In one government exhibit, Mr. Bellack is seen trying to mend a relationship with Hearst, a large publisher, after Hearst disabled AdX, and sought to respond to their concerns.
When the infamous NYSE email came up, Mr. Bellack explained he was saying this with respect to his concerns that the reason that large companies like Verizon and Amazon were seeking to compete with Google is because they did not trust the company, seeing “Google’s structure” as “unacceptable.”
Jonathan Bellack was impressive. He did not fall on his face, he did not have amnesia like other Google witnesses, he did not evade questions—he took everything head on, and gave strong and reasonable answers.
Ms. Dunn did not cross examine him, and just under an hour after his testimony began, it ended.
Next up were a series of extremely brief deposition read-ins. By brief I mean only a handful of minutes each. The only value here could have been for the government to affirm certain facts it felt it needed a little bit more on, as well as (and more importantly) to enter exhibits.
One notable line was from the deposition read-in of Michael Shaughnessy, COO of Kargo, an ad agency, who said “It would create pain in the publisher’s business to switch away from Google”—something we’ve heard from nearly all publishers, referring to the inability to forgo Google’s scale and inventory, and how difficult it was to change systems.
To end the day, we had the direct examination of Robin Lee, a Professor of Economics at Harvard, a government expert witness, and the government’s final witness. While the government finished its direct examination around 5:25 (it started around 3 PM), Judge Brinkema decided to recess early for the day to allow Google lawyer Bill Isaacson to start his cross examination fresh in the morning, as opposed to in the last half hour or so of the day.
Dr. Lee, in a word, is brilliant. His direct examination lasted two and a half hours, and he had the content memorized. No fact, no matter how obscure or lost in the sea of binders the government handed out, seemed unfamiliar to him. He knew his analysis, the market, and Google’s conduct, backwards and forwards, and is truly a master of his craft. Mr. Isaacson, who’s specialty in this trial appears to be cross examining economic experts, has his work cut out for him.
Dr. Lee testified that he saw publisher ad servers as the relevant markets from an antitrust perspective, and said that Google possesses “substantial and sustained market power.”
He attested that he performed regression analysis, auction simulations, and reviewed internal Google simulations and experiments as part of his methodology. In total, he performed approximately 100 experiments, and used both Google and third-party data.
I could be wrong, but it seems like the government may have learned from Mr. Isaacson’s habit of trying to rough up expert witnesses for any perceived shortcoming in their analysis (i.e., they did not do enough quantitative analysis, that they have not previously specialized in this exact market, that they reviewed the wrong data, that the data was printed in the wrong font on the wrong cardstock paper, etc.). The government seemed to cover a lot of these bases on direct.
Dr. Lee discussed the hypothetical monopolist test (HMT) – where a product in question (in this case Google’s DFP) is evaluated in terms of if, hypothetically, one firm were to control it, would they be able to significantly raise prices? In other words, by definition, a monopolist, hypothetically, must be able to raise its prices. In this case, Google was able to raise prices, and even lower quality in terms of features, without the expected decrease in demand.
While some of this evidence seemed cumulative, Dr. Lee was tying (no pun intended) together the government’s case, as its final witness.
After discussing the cellophane fallacy (an economic theory related to monopolists’ ability to raise prices), Dr. Lee went through more of his report.
Some of the most important testimony concerned Dr. Lee’s conclusions as to Google’s five forms of anticompetitive conduct. He argues they are:
1.) Tying Google Ads to AdX
2.) Conditioning real time bids access to DFP
3.) Exclusive advantages built in to DFP
4.) Unified pricing rules
5.) Google’s acquisition of Admeld
Finally, Dr. Lee testified that Google’s conduct harmed publishers and advertisers in three ways:
1.) Higher sustained prices and lower payouts
2.) Google’s conduct reduced advertiser options and efficiency
3.) Google hurt innovation, as evidenced by competitors like Verizon and Facebook exiting the market.
Bill Isaacson’s cross examination of Dr. Lee will begin tomorrow morning. Then the government will rest its case.
Ms. Dunn told Judge Brinkema that Google expects to rest its case this coming Wednesday or Thursday, just two and a half weeks after the trial began. For reference, Google’s other antitrust trial (Google search) last fall took ten weeks.
The rocket docket is for real.
Excellent. Succinct. Thanks.
You are really good at this. Thank you.