890 million and a redesign

€890 million sounds like a lot until you learn whose pocket it came from.

Google made $350 billion in revenue last year. The European Commission’s fine — the first ever levied under the Digital Markets Act — amounts to roughly what Google earns in eighteen hours. If this is punishment, it’s the kind of punishment that gets expensed.

But the fine is the part everyone talked about. The number is big enough to make a headline and small enough to not matter, which is the worst combination a regulator can produce. Google said it will appeal. Of course it will appeal. Appealing a billion-dollar fine costs less than the interest the company earns while the appeal is pending.

The interesting part arrived in the same envelope and got half the attention.


The Commission split its action into two decisions. One covers Google Search — self-preferencing, the practice of ranking your own services above competitors’ in results you control. The other covers Google Play — anti-steering restrictions that prevented app developers from telling users about cheaper purchasing options outside the Play Store. Both are textbook gatekeeper behaviors. Both have been litigated before, under older competition law, with fines that Google absorbed like a large dog absorbs a bath.

What’s different this time is the structural remedy. The Commission didn’t stop at writing a check. It ordered changes to how Search displays results and how Play permits developers to communicate with their own customers. Not behavioral commitments — architectural ones. The system has to work differently, not promise to behave.

This is the distinction the DMA was built around, and it matters more than the dollar figure. Traditional competition law is retrospective. You did a bad thing, here’s a fine, don’t do it again. The DMA is prospective. It designates gatekeepers — platforms large enough to control access to markets — and prescribes how they must operate. The shift is from punishing behavior to specifying structure. From asking “did you abuse your position?” to stating “here is how your position must function.”

The difference between those two approaches is the difference between telling someone to drive more carefully and redesigning the road.


Langdon Winner wrote in 1980 that artifacts have politics — that the design of a technical system embeds choices about who benefits and who doesn’t, whether or not anyone intended it. The DMA’s structural remedies are an attempt to change the politics of the artifact rather than the intentions of the actor. That’s a genuinely different regulatory move. Whether it produces a genuinely different artifact is the open question.

Because architecture is not neutral, but it is also not fixed. Google has spent two decades building systems that route attention and money toward Google. A Commission order that says “restructure your search results page” creates a constraint. It does not create an outcome. The distance between the constraint and the outcome is where Google’s engineers live, and they are resourceful people. The history of tech regulation is a history of compliance that preserves the underlying advantage — new walls built to the exact specifications of the ones that were torn down, positioned three inches to the left.

The structural remedy is the right tool. But a tool is only as good as the hand that holds it, and the Commission now has to do something it has never done at this scale: monitor the architecture of a platform in real time, evaluate whether structural changes produce competitive effects, and enforce iteratively rather than episodically. That’s not a legal skill. That’s an engineering capacity. The DMA created the authority. Whether the Commission can build the apparatus is a separate question.


The fine is noise. Google will pay it or appeal it or both, and nothing will change. The structural remedy is the signal — the first real test of whether regulation can reshape the architecture of a dominant platform, or whether it can only charge admission to watch it operate.

€890 million is a rounding error for Google. But if the remedies hold — if the architecture shifts, if the artifact gets different politics — then the fine was never the point. It was the envelope. What matters is what came inside.