Day 3: The Experts Arrive

This is coverage of the remedies trial in US v. Google. I’ll be in Virginia writing every day on the courtroom activities.

Day 3: The Experts Arrive

We closed yesterday’s (extremely boring) day of testimony with the judge admonishing the DOJ to move beyond the “window dressing” and get to the heart of the matter, with expert witnesses. The DOJ heard the call, and today we got experts in economics, banking, and software.

Witness #7: Professor Robin Lee, Harvard

Professor Lee of Harvard testified in the Liability phase and presents himself as deeply knowledgable and unflappable witness. He was called by the DOJ to validate whether their proposed remedies would restore competition in the relevant markets.

AdX Divesture: Professor Lee testified that divesting AdX would encourage AdWords to bid more broadly on inventory than now, when only 10% of demand goes to other exchanges. His analysis showed that as of 2022, 89% of AdWords display dollars went to AdX, versus 63% of DV360 and only 34% of other external bidding tools.

An AdX divestiture would, Lee stated, discourage cheating within DFP as there would be no incentive to implement features like “First look”, etc.

Finally, the expectation is that once AdWords bidding was more spread out, there would be more competition among exchanges, which would lower exchange take rates. To back up this point, Lee did an analysis that found that 60% of AdX winning auctions had zero competition, and more than half of these had AdWords as the only bidder. This reinforces the point I’ve made many times, that AdX essentially only exists because it is subsidized by AdWords.

DFP Divestiture: Professor Lee indicated that the phased DFP divestiture would lower switching costs for customers, lower barriers to entry for competitors, and prevent the ad server from making the ad exchange market non competitive.

Behavioral Remedies: The behavioral remedies proposed by the DOJ are meant to prevent Google from doing an end-run around AdX (post divestiture) by, for example, bidding directly into DFP. To illustrate how powerful AdWords demand is, a chart was shown indicating AdWords represented 46% of all indirect open web display ads, while DV360 represented another 21%. It’s Google’s internet, we just live in it.

Judge Brinkema’s Comments and What They Mean

Towards the end of Lee’s testimony, while Lee was focused on the various behavioral restrictions being proposed, Judge Brinkema piped up to ask a question about trust. This is not a jury trial, so the future of this whole issue is really, 100% up to Brinkema. When she asks a question, it should be taken note of very seriously.

The question she asked, more or less, was “is there any evidence Google would violate an injunction and try to re-monopolize?” She followed up saying “Lack of trust has a huge impact on potential remedies.”

What did she mean!? In my opinion, this statement was a way of saying that she would prefer to have behavioral remedies, and not structural remedies, because if we trust Google, then that would be sufficient. If you recall from Day 1, witnesses like Andrew Casale were explicit about not trusting Google, bringing up the idea of “third look” or “fourth look” as new schemes to advantage the company. I asked other court observers, and there was a disparity of opinion as to how to interpret these remarks.

Scary Questions on Cross

In the cross examination of Professor Lee we got some glimpses of a scary future post spin-out. Counsel asked theoretical questions like:

• If AdX was shut down or degraded, what would happen to the 60% of auctions where it was the only demand?

• If Google sold AdX but remained in non-display, could or would they favor non-display with their demand?

• Would anything prevent Google post-spin from putting all of its demand to YouTube?

• Is there anything in the remedies that says Google must participate in the open web?

If you’re scared the “open web” is in decline already, there’s a real risk that too harsh a penalty in this case could make it worse!

Competitive Intel on DV360

We’ve never really known how much spend goes through DV360. In the 2020 financial document posted in the liabilities trial, the media spend through DV360 was $6.7 billion, including fees. Today, a chart was put into evidence giving a lot more detail about that spend, including an astronomical amount of YouTube growth. Easier to just link to my thread on Twitter than reproduce here:

DFP Free Accounts

I brought this up in yesterday’s newsletter, but the problem of DFP Small Business remains in any spin out scenario. There are tens of thousands of small publishers who get DFP for free, subsidized by AdX revenue. Without the tie between them there’s no reason to support these.

Professor Lee was asked about this on cross, and didn’t really have an answer. He said an acquirer of DFP “might” increase prices, or would have to offer a complementary product. I predict this is going to be a key part of Google’s argument against a spin.

Witness #8: Paul Crisci

This expert is a technology investment banker. His testimony amounted to “yes, there would be people interested in buying AdX and DFP. Duh!

This was a waste of time, and the only notable thing was his assertion that “open web display” was a growing market, which was totally demolished by Google’s counsel on cross. The witness has seen “display” in the IAB annual report and forgot that the number includes social. Whoops.

Witness #9: John Weissman, PhD

Professor Weissman is a deeply experience software academic with a focus on distributed systems. He testified about the technical feasibility of APIs, spin-outs, and open sourcing the auction logic. For someone who has worked in tech, his testimony was painfully simplistic and obvious. For the 82-year old judge, however, it was pretty important to go over some rudimentary stuff, like whether an API can connect two systems written in different software languages.

Honorable mentions

• None today, it was dull