Power Consolidates Quietly

Katherine Mangu-Ward wrote an essay in the New York Times this week about executive overreach. Her argument was structural, not partisan. Power consolidates incrementally, she said, through a series of individually reasonable steps, and by the time anyone looks up, the architecture is already in place.

She was talking about government. But the pattern she described is playing out right now in how businesses reach their customers, and nobody is treating it with the urgency it deserves.

Google’s Decade Long Strategy

Consider how Google spent the last decade. DeepMind, Looker, Mandiant, Fitbit: each acquisition made sense in isolation, and each quietly expanded the surface area of control over how information is organised, interpreted, and delivered. The industry watched it happen the way you watch someone rearranging furniture in a room you share. Each move is small, each move is reasonable, and then one morning you realise the room has been entirely redesigned around someone else’s preferences.

Then came Agent2Agent, which is not a product but a protocol. Google is positioning itself as the entity that defines how AI agents communicate with each other. If A2A becomes the standard, and it is well on its way, every agent-based interaction runs through Google’s logic, Google’s definitions, Google’s architecture. This is not a feature announcement. It is the equivalent of writing the rules of engagement for a system everyone else must operate within.

The relentless release cadence is the normalisation phase. Gemini in Search, Gemini in Workspace, Gemini in Chrome, Gemini in Android. Each release makes the previous consolidation feel like settled ground rather than contested territory, and people stop questioning whether Google should control the discovery layer because they are too busy reacting to the latest capability drop.

Captivity, not cowardice

Mangu-Ward blamed “cowardly congressmen afraid to do their jobs” for the political version of this problem. The content industry has its own version, but it is not cowardice. It is captivity.

Brands optimise for Google’s rules because the alternative is losing the organic traffic their business depends on. A mid-size company that stops playing Google’s game does not make a principled stand; it loses 60 to 70 per cent of its discovery overnight. Google has structured the system so that compliance is survival. The cost of resistance is immediate and concrete; the cost of compliance is abstract and deferred, so everyone complies. And now, as AI restructures the discovery layer entirely, that dependency is being transferred to a system with even less transparency and zero recourse.

Here is where it gets worse than the political version.

Executive overreach has a constitution, a Supreme Court, and structural checks even when they are weakened. Separation of powers, appellate processes, and public accountability mechanisms. They are imperfect, but they exist. The AI discovery layer has none of this. No constitutional framework, no separation of powers, no appellate process when your brand gets misrepresented in an AI-generated answer, no regulatory body overseeing how your twenty years of expertise gets synthesised, flattened, or ignored by a model trained on your content but under no obligation to credit you, represent you accurately, or send anyone your way.

The consolidation is happening in a regulatory vacuum, and the businesses most affected are the ones least equipped to see it because they are still measuring success by the old metrics. Rankings, traffic, click-through rates. They are optimising for a system that is being quietly replaced beneath their feet.

The concrete is setting

I have spent nearly three decades in search and digital strategy. I watched featured snippets arrive and thought that was interesting, watched knowledge panels expand and thought that was worth watching, watched People Also Ask boxes multiply and started to feel something structural shifting. When AI Overviews launched, I knew the concrete was setting. Each step shaved a little more agency away from the publisher and the user and concentrated it in the platform. Each step was individually defensible. Collectively, they built an architecture where Google answers queries without sending traffic, where AI platforms synthesise answers from content they did not create, for users who never see the source. The entire discovery layer now sits in the hands of three or four entities, and unlike the political system the libertarian essayist was describing, there is no mechanism for getting it back.

Change can still happen

But there is a difference that matters.

The libertarian piece reads like a eulogy, describing damage already done, power already consolidated, checks already dismantled. The political architecture she wants to reform has been decades in the making and is now deeply entrenched. The AI discovery architecture is still being built. The protocols are not yet universal, the models are not yet the sole point of entry, and the formation layer, that invisible space where customers form opinions and build shortlists before they ever reach a traditional search channel, is still taking shape.

Which means there is actually something to do about it. Not optimising harder for the existing system, which is the equivalent of lobbying the overpowered executive for a better deal. The brands that will retain leverage are the ones building authority that is not dependent on a single platform’s architecture. Direct relationships, proprietary insight, the kind of expertise that an AI model has to reference, because nothing else carries the same weight.

The argument is not political; it is structural. When power consolidates incrementally, and nobody builds countermeasures, you wake up one morning with no leverage and no recourse.

The only question is whether you do something about it while the concrete is still wet.

Harton Works offers an AI Readiness Audit that helps you identify specific gaps, prioritise hight-impact areas, and build a clear AI implementation roadmap.

AI Readiness Assessment

Martin Jeffrey

Martin Jeffrey is the founder and strategic lead of Harton Works, a SEO and AI Search agency focused on Retrieval-First™ Marketing and AI-era visibility. With over 25 years of experience in digital strategy, he helps businesses adapt to the new rules of search, aligning SEO, content, and AI readiness to drive sustainable growth.