A psychologist is advising professionals to leave small mistakes in their public posts. The reasoning: imperfection reads as human in an environment saturated with machine-polished text. The advice is not wrong, exactly. The problem is that it arrived in print, which means it became a product, which means someone built software around it.
There are now tools designed to inject the right kind of chaos into AI-generated writing. The dropped comma. The sentence that runs a beat long. The "lol" placed where a "lol" would plausibly appear. The imperfection is no longer a signal of anything; it is a feature set.
This has happened before, in a different register. Distressed denim went through the same compression. Raw selvedge started as the honest thing, the jeans that showed wear because they were worn. Then distressing became a finish. Then pre-distressed jeans became a product category available at every price point from Gap to Gallery Dept. The authenticity that the look was selling became the first thing a buyer could purchase. Nothing about the jeans changed. The context did.
The humanizer-tool industry is the pre-distressed jean. It is not lying, exactly. It is packaging a signal into a repeatable output and selling it at scale. The buyer is not trying to deceive; the buyer is trying to belong to a category that used to have a different entrance requirement.
What the industry misses is that the signal it is counterfeiting was never about the typo. It was about the consistency of the imperfection with everything else in the post: the voice, the cadence, the things the writer chose not to explain. A misplaced apostrophe in a sentence that otherwise reads like a press release is not humanizing. It is a tell of a different kind.
The next tell is not on this list yet. It will be something that cannot be prompted into existence, which means it will be behavioral rather than textual. What it looks like is hard to say in advance; that is the point. What can be said is that "delve" became a tell before most people registered it as one, and the humanizer industry will be twelve months late to the next version of the same problem, because the industry is reactive by design.
