Beneath thousands of technical filings from the early internet period, somewhere in the US Patent and Trademark Office archives, lies a document that most digital publishers have never read and have never needed to. In the technical terms of patent applications, it explained an algorithmic approach to assessing and ranking web material, taking into consideration signals such as update frequency, link authority, and what the file referred to as “information quality.” It was generally ignored by the individuals whose livelihoods it most directly affected for twenty years. The discussion that ensued was awkward once someone took another careful look at it.
The story isn’t entirely about the patent itself. It brought to light a number of issues that the digital publishing sector had been neglecting because the traffic and ad income continued to flow. Certain behaviors, such as frequent writing, keyword density, and link accumulation, were rewarded by the ranking systems based on these fundamental approaches, and an entire business developed around obtaining these rewards. Thousands of websites are actively optimized for the same signals, some of which are content farms masquerading as legitimate journalistic operations. It worked until it stopped working, and most publishers weren’t prepared for how quickly and thoroughly it ceased working.
Zero-click searches are the current issue. These days, when someone types a question into Google, they are more likely to see an AI-generated summary at the top of the results page that provides an answer without forcing them to click thru to any websites. This is existential for a trip guide, recipe publisher, or simple informational website. The traffic paradigm that supported the investment in content vanishes. The operation’s funding source, ad impressions, vanishes. Technically, the website is still up, but the search engine has taken over as the primary motivator for people to visit it.
This has a certain irony that is actually hard for publishing professionals to accept. The content that publishers spent years creating and refining served as a major source of training for the huge language models that underpin these generative search features. The well-structured how-to manuals, the well-formatted instructive postings, and the SEO-friendly articles were all perfect training data. In essence, publishers built the mechanism that today captures their audience before those people ever visit the publisher’s page—mostly without understanding it and without receiving paid. Courts in many jurisdictions are testing whether that is legally actionable. There is increasing agreement among media professionals over whether it represents a serious structural injustice.
For a few years now, the more astute publishing companies have begun responding to this, some of them more quickly than others. The trend is generally consistent: information with a unique viewpoint, a distinctive voice, or access that an AI summary cannot duplicate is preferred above high-volume keyword chasing. investigative journalism. specialized knowledge. first-person analysis based on personal experience. Compared to the informational pieces that constituted the majority of the traffic for many publishers, these forms are more resistant to commoditization. Additionally, they are more costly to produce and more difficult to scale, which puts a strain on finances even for businesses that make the best strategic decisions.
The other significant hedge is now direct audience channels. Distribution that doesn’t rely on a search engine determining whether today’s information satisfies today’s algorithmic threshold includes newsletters, membership communities, and podcast audiences. Despite its drawbacks, the Substack model offers a true structural alternative to search-dependent publication, and its rapid growth indicates how many authors and readers were searching for just that kind of escape route.
