The Moat That Compliance Doesn't Reach

Series: The Contestability Wager | Monograph 2

What the Coverage Got Wrong

Most of the commentary on the Commission's April 2026 action against Google has treated it as a news event. Google faces new data-sharing rules. Challengers will benefit. Competition in search may finally be restored. That reading is not wrong about what the Commission intends. It is almost entirely silent about whether the instrument the Commission built can produce the outcome the Commission intends — and silence on that question, at this moment, is not neutral. It costs something.

On April 15, 2026, the Commission sent preliminary findings to Alphabet under DMA Case 100209, initiating a specification proceeding under Article 6(11) of the Digital Markets Act — the provision requiring gatekeepers operating search engines to share anonymized ranking, query, click, and view data with third-party search services on fair, reasonable, and non-discriminatory terms. A public consultation is open through July 27, 2026, the statutory deadline for a binding decision. The proceeding is not a penalty. It is an attempt to redesign the conditions under which search competition is possible — to convert a privately accumulated behavioral corpus into something closer to shared infrastructure.

What I want to do here is read the instrument rather than the verdict. Not provision by provision as a legal exercise, but dimension by dimension as a competitive one — asking, for each component, which variable in the competitive structure it moves, which it leaves intact, and where the gap between formal mandate and operational contestability opens. The Moat That Compounds established that Access and Effort are the two dimensions that govern this competitive structure: Access as the threshold condition the mandate directly addresses, Effort as the asymmetric cost of converting mandated data into competitive relevance, which it does not. That framing is the analytical lens this reading applies.

Six Provisions, One Dimension

A competition instrument is not evaluated by what it intends. It is evaluated by what it moves.

The Commission's proposed measures contain six operative provisions. I want to read each one not as a legal obligation but as a dimensional intervention — asking, for each component, which variable in the competitive structure it addresses, which it leaves intact, and where the gap begins to appear.

The eligibility criteria remove the most fundamental barrier in the competitive structure: formal exclusion from the data asset entirely. Before Case 100209, Google had no legal obligation to share its behavioral corpus with any third party under any terms. The eligibility provision converts that discretion into an obligation. This is the Access dimension at its most basic — not a solution to the competitive problem, but a necessary precondition for any other dimension to be addressable at all. It also does something structurally novel: it extends eligibility to AI chatbots with search functionality, making this the first competition proceeding to classify AI products as eligible search competitors. The significance of that extension deserves its own section.

The data scope provision specifies what must be shared: click, query, view, and ranking signals. This addresses Access by defining the entitlement precisely enough that compliance cannot be satisfied by sharing an irrelevant subset. The inclusion of ranking signals alongside behavioral signals has an Effort dimension too — ranking signals tell a challenger something about how Google has already interpreted behavioral data, marginally reducing the interpretive work required to convert raw inputs into relevance outputs. But the scope provision is silent on granularity. Whether click data is delivered at the session level, the query level, or the aggregate category level determines whether it is useful for tail-query training or only for head-query calibration. Scope without granularity addresses Access — what is shared — without touching the Effort variable that depends on how it is shared.

The frequency and delivery requirement prevents the strategic delay that would allow Google to provide nominally compliant data at a cadence that renders it competitively irrelevant — sharing yesterday's behavioral signals in a market where the feedback loop processes today's. Near-real-time delivery is contemplated for certain signal types. This is a well-specified Access provision. What it does not touch is Effort: receiving data in near-real-time does not reduce the engineering, tuning, and signal-interpretation work required to convert that data into a competitive ranking model. Timeliness and usability are different variables.

The anonymization standard is where the instrument's internal tension becomes visible.