There's a peculiar kind of irony baked into the current state of search. For decades, Google's entire pitch to the world was simple: we help you find things. Not we are the thing. The blue links were the point. The websites behind them were the point. Google was just the very helpful, very profitable highway on-ramp.
That deal is now dead. And it died so quietly that most people didn't notice until their favorite recipe blog, their go-to how-to site, or that one forum where actual humans answered actual questions just... stopped showing up.
The Answer Box That Ate the Internet
It started innocuously enough. Google introduced "featured snippets" years ago — those tidy little boxes at the top of search results that summarized an answer so you didn't have to click anything. Annoying for publishers, sure, but survivable. Most people still clicked through.
Then came AI Overviews. Then Gemini. Then every other search engine decided it needed its own chatbot bolted to the front page like a hood ornament on a car that no longer has an engine.
Now when you type "how long to boil an egg" or "symptoms of a sinus infection" or "what does escrow mean," you get a confident, paragraph-length answer generated by a large language model — sourced from thousands of websites that will receive exactly zero visitors for providing that information. The AI ate the content. The content creator got a participation trophy in the form of a tiny citation link that approximately nobody clicks.
This is enshittification in its purest form. The platform extracted value from its ecosystem until the ecosystem had nothing left to give, and then kept extracting anyway.
The Invisible Collapse Happening Right Now
Here's what the traffic dashboards of independent publishers look like in 2024 and 2025: a cliff. Not a gradual slope. A cliff.
Whole categories of websites — health information, legal explainers, financial definitions, DIY tutorials, travel guides — are watching organic search traffic crater by 30, 50, sometimes 70 percent year over year. The sites didn't get worse. The content didn't become less accurate. Google just decided it was more efficient to summarize the content itself and keep users on Google.
Small operations are folding. Mid-size publications are doing rounds of layoffs and calling it "restructuring." Even larger outlets that built their entire business model around SEO-friendly content are realizing that SEO-friendly content is now just... AI training data with a publishing date.
The economic logic that kept the web humming — write useful stuff, attract readers, sell ads or subscriptions, repeat — has been quietly dismantled. The web still exists. It just doesn't pay anymore.
Who Actually Benefits Here?
Not you, despite what Google's press releases suggest.
The pitch from Big Search is that AI answers are better for users. Faster. More convenient. No need to sift through ten mediocre articles stuffed with keywords. And honestly? For simple factual queries, that's sometimes true. Knowing the capital of Nebraska doesn't require a five-paragraph essay with a recipe card at the bottom.
But the web was never just a fact retrieval machine. It was also a place where genuine expertise lived. Where a dermatologist ran a blog explaining skin conditions in plain English. Where a mechanic detailed exactly which bolt to loosen on a 2008 Honda Civic. Where a bankruptcy attorney broke down what Chapter 7 actually means for regular people.
That expertise existed because there was a transaction: the expert created content, readers found it through search, the site earned enough revenue to keep the lights on. AI search has severed that transaction. The expertise still gets consumed — just by the model during training, not by the human who needed it.
And the model? It doesn't update its knowledge in real time. It doesn't have malpractice liability. It confidently tells you things that were accurate eighteen months ago, or things that were never accurate at all, with the same breezy tone it uses for everything else.
The Ouroboros Problem
Here's where it gets genuinely frightening for anyone who thinks about the long-term health of the internet.
Large language models are trained on human-generated content from the web. That content was produced because there was an economic incentive to produce it. Remove the economic incentive — which is exactly what zero-click AI search does — and the volume and quality of new content drops. The models then have less fresh, high-quality data to train on. Their outputs get worse, or at best, stale. They become an ouroboros eating their own tail, recycling older and older information with increasing confidence.
Some researchers have already started calling this "model collapse" — the theoretical point at which AI trained primarily on AI-generated content starts producing meaningfully degraded outputs. We're not there yet. But we're building the conditions for it while simultaneously dismantling the content ecosystem that could prevent it.
The internet's memory, which was already fragile (ask anyone who's tried to find a webpage from 2009), is about to get a lot shorter.
What Happens to a Web With No Incentive?
The optimists will tell you that creators will find new models. Newsletters. Paywalls. YouTube. Podcasts. Community platforms. And sure, some will. The hustle never really stops.
But those alternatives don't replace the open web — the publicly accessible, indexable, linkable, searchable infrastructure that made the internet useful to everyone, not just people who knew which Substack to subscribe to. A web that exists primarily behind paywalls and inside walled gardens isn't really a web anymore. It's a collection of toll roads where the highway used to be.
The sites most likely to survive this transition are the ones backed by institutions — universities, major media companies, government agencies — that don't depend on ad revenue to justify their existence. Everyone else is playing a game where the rules changed without notice and the referee is also your competitor.
BadrWeb's Completely Unsolicited Verdict
The enshittification of search isn't a bug in AI's rollout. It's the business model. Keep users on the platform. Reduce dependency on external content. Monetize the interface. The fact that this systematically undermines the open web is, from a quarterly earnings perspective, not really Google's problem.
It's everyone else's problem. Including, eventually, the AI systems themselves — when they run out of fresh human thought to digest and start serving you a very confident, very smooth summary of nothing in particular.
The internet is broken. We took notes. Unfortunately, the AI already summarized them.