Writing for algorithms has become, for many writers, the primary frame through which they approach their work. Not the only frame — most writers would insist they are writing for readers — but the organising frame. The question of whether a piece will perform in search shapes decisions about structure, length, title, and even the level of complexity the argument is allowed to reach.
This is understandable. Search visibility is real, and for writers who depend on traffic, ignoring it is not a neutral act. But there is something worth examining in what algorithms actually measure — and what, by definition, they cannot.
What Algorithms Measure
Search algorithms are sophisticated and growing more so. They assess relevance, authority, and what Google calls E-E-A-T — experience, expertise, authoritativeness, and trustworthiness. They measure engagement signals: how long readers stay, whether they click through, whether they return. They evaluate structure, readability, and the presence of semantically related terms. They reward content that answers questions clearly and completely.
This is not nothing. These are reasonable proxies for useful content, and a piece that scores well on most of these measures is probably, in many cases, a reasonably good piece.
But they are proxies. They measure indicators of quality, not quality itself. And the gap between the indicator and the thing it indicates is where the most interesting writing lives — and where writing for algorithms, taken too far, begins to cost something real.
What Algorithms Cannot Measure
An algorithm cannot tell whether a sentence is surprising. It cannot detect the moment when a piece of writing shifts from competent to alive — when a word choice is so precise that it changes the quality of attention the reader brings to the next sentence. It cannot measure the pleasure of an argument that arrives somewhere unexpected, or the particular satisfaction of a closing line that earns its finality.
These are not decorative qualities. They are the qualities that make writing memorable rather than merely useful — that make a reader return to a site not because they have a question but because they want to spend time with the thinking. They are also, not coincidentally, the qualities that build genuine authority over time. Not the algorithmic simulation of authority, but the real thing — the sense a reader develops that this writer sees clearly and thinks carefully.
An algorithm cannot measure trust in that sense. It can measure proxies for trust. But a reader who returns to a piece because it changed how they think about something is not responding to a proxy. They are responding to the writing itself.
What Gets Lost When Algorithms Lead
When writing for algorithms becomes the primary orientation, certain choices become systematically harder to make.
The digression that turns out to matter — the moment when a piece follows an unexpected thread and discovers something the planned argument didn’t anticipate — is hard to justify in a piece shaped by keyword density and structural requirements. The slower, more considered pace that some arguments require is difficult to maintain when readability scores penalise long sentences. The conclusion that doesn’t resolve neatly, that leaves the reader with a question rather than an answer, is almost impossible to write when the implicit requirement is that every piece deliver clear, extractable value.
None of these sacrifices is dramatic on its own. But cumulatively, across thousands of pieces and millions of writers all optimising in the same direction, they produce a particular kind of writing — competent, clear, useful, and remarkably similar. Writing that answers questions well but rarely asks them. Writing that performs helpfulness without always achieving it.
The algorithm rewards this writing. It does not know what it is missing.
None of this is an argument against SEO, or against writing clearly, or against giving readers what they came for. These things matter, and a writer who ignores them entirely is making a different kind of mistake.
But the writers whose work endures — whose pieces are still being read and shared long after the search rankings have shifted — are almost never the ones who optimised most effectively. They are the ones who had something to say and found the precise language to say it. Who followed the argument where it led rather than where the brief required. Who trusted the reader to stay with complexity rather than smoothing it away.
The algorithm will catch up, eventually, to some of what good writing does. It already rewards originality and depth more than it once did. But it will always be measuring from the outside — tracking signals, inferring quality, rewarding proxies.
