Trial Update, September 24, 2025: Three experts explore all that is feasible

DOJ kept their promise on day 3 of US v. Google remedies, calling three experts to the stand. It was a relatively uneventful day, as all eyes were on DOJ’s technical expert after Judge Brinkema’s “hint” yesterday, who took the stand at the end of the day, and will continue tomorrow.

1. Robin Lee, Economic Expert

Dr. Lee spoke to the competitive effects of DOJ and Google’s proposed remedies on the ad server and ad exchange markets. He used a framework comprised of three objectives: (1) stop Google’s anti-competitive conduct, (2) resolve competitive harms from Google’s conduct, (3) create a high likelihood of preventing future monopolization through similar conduct.

He noted that while he considered what the market would have looked like but for the anticompetitive conduct, he did not specify what a but-for world would have looked like.  He explained that it was not feasible to specify - there are innumerable permutations of a potential but-for world, given the length of time over which the anticompetitive conduct occurred, and that is all compounded by scale and network effects. Judge Brinkema nodded in apparent agreement.

Several times throughout, he spoke about the importance of removing Google’s ability and incentive to crafting an effective remedy - particularly in resolving the latter, longer-term objectives. This is why divestiture is particularly effective. We are reminded that 89% of Adwords transactions went through AdX, along with 63% of DV360 transactions.  Meanwhile, for 3rd party bidding tools (e.g. DSPs), only 34% went through AdX.  AdX divestiture would change Google’s economic incentives here. Divesting AdX also reduces the incentive to engage in conduct like First Look, Last Look, and UPR, because Google can’t internalize the benefits.

We’re reminded that his auction simulations during liabilities phase found that removing AdX from an auction reduced publisher payouts by 28%, 50 x greater than the impact of removing any other ad exchange. We’re reminded of AdX’s enduring supracompetitive take rates - clearly whatever efficiencies Google experienced from vertical integration, he notes, weren’t great enough to prevent engaging in anticompetitive conduct.

Regarding contingent DfP divestiture, he says that competition may not be restored if, for example, Google creates an exclusive pipeline between AdWords and the remainder of DfP. This is the type of scenario that would render divestiture of the remainder of DfP necessary.

On behavioral remedies, he discusses that while the Court’s opinion rejected advertiser bidding tools as a relevant antitrust market, the behavioral restrictions proposed by DOJ are consistent with the Court’s findings. Google Ads is an important source of demand, and thus market power, irrespective of defining a relevant antitrust market. On Google 1st party data self-preferencing restrictions, he says this proposal would prevent re-monopolization.

Judge Brinkema asks if his analysis is based on an assumption that Google will try to re-monopolize. She asks if there is any other case where Google has been enjoined with regard to antitrust, and violated the injunction. DOJ jumps in to point to Google’s 2021 adtech antitrust fine in France, which they note that Arnaud Creput would be speaking to. The Judge says that what holds parties up in remedies is that one side doesn’t trust the other. She asks Dr. Lee that if he had confidence that Google would act in good faith, and not circumvent, could behavioral remedies be sufficient. Theoretically, he explains, but to the extent that the conduct is foreclosed up front. Anything unforeseen cannot be prevented behaviorally. My sense here isn’t that the Judge is expressing a preference for behavioral remedies, necessarily. Her expressed interest in the technical feasibility experts makes it clear that she wants to understand divestiture as an option just as much.

Isaacson badgers Dr. Lee again for several hours. The most interesting points he raised were about whether Google would have an incentive to favor non-open-web display over open-web display under the DOJ’s divestiture proposals - e.g. would Google have incentive to funnel money into YouTube? This was a weird reverse Uno move, particularly when Google in the same breath is pushing back against the 1st party data self-preferencing remedy and insisting AdWords and DV360 are wholly off the table in remedies discussions, while these are the proposed remedies that would aim to mitigate exactly this risk.

2. Paul Crisci, Commercial Feasibility Expert

Next up is Paul Crisci - an investment banking expert speaking to whether AdX and DfP would be sufficiently attractive to a motivated pool of buyers. Unsurprisingly, he found that there would be sufficient interest by potential buyers. Judge Brinkema asks if a court-ordered divestiture or need for approval would discourage buyers, and he explains that regulatory approval is common and if anything it would indicate that Google is motivated to sell. More figurative waterboarding by Isaacson follows. He points out an inconsistency in the articulation of the materials relied on by the witness in one of his reports, so a long-fought congratulations to Isaacson for finally successfully embarrassing one of the DOJ’s witnesses by catching a presumably accidental error.

‍

3. Jon Weissman, Technical Feasibility Expert

Finally, we get to the technical feasibility expert. Weissman lives and breathes cloud computing and distributed systems. He is so evidently qualified that even Google’s Jeannie Rhee laughs when the Judge asks as a formality if there are any objections. He explains that he evaluated whether it is technically feasible to copy and migrate certain technical assets of AdX and DfP, to open-source the DfP final auction logic, and to develop the APIs proposed by DOJ. He finds each of these to be technically feasible.

We walk through his evaluation, beginning with APIs. Judge Brinkema asks if programming language could be an impediment. He says that adapters can be used to “translate” from Swahili to English, to use the Judge’s analogy. Google’s technical expert, Jason Nieh, didn’t challenge the API feasibility either, he confirms.

He explains that copying and migrating AdX and DfP is feasible, and that the open-sourcing process would be similar to the migration process. He said that what would be a big lift is to change software architecture, or to do a from-scratch re-implementation - but these things aren’t necessary here because these tools have been architected in a modern way, to scale.  He says Google is written in C++ and Java, modern programming languages that a lot of developers like. He said that Google’s source code was beautifully organized, well-defined, and consistent.

Judge Brinkema asks how he can know that quality of the end result would not be degraded without knowing the new environment it is migrating to. He reiterates that it is because it is written in a common language, etc. She asks what type of buyer would be able to support AdX and DfP given Google’s size and adtech complexity. He explains that the buyer only needs to support AdX and/or DfP, not all of the other applications Google’s infrastructure is built to handle. He disagrees with Google’s expert that complete re-writing of the software would be required, based on his review of the materials like the engineering style guide, architecture diagrams, and source code. We end on a walkthrough of an architecture diagram of DfP, and will pick back up here tomorrow.

‍