If the first number of this publication documented a phenomenon, the second has documented an economy. What emerges from the ninety-odd specimens the Board of Review has accepted for this edition is not merely that machines continue to produce material indistinguishable from human effort, nor that human beings continue to publish it as their own, but that a complete commercial and emotional infrastructure has now been erected upon the inability to tell the difference—and that this infrastructure requires, for its continued operation, that the difference never be clarified.

We have received, in the span of a single editorial cycle, a machine-generated advertisement for a machine-generation service, a machine-generated complaint about machine-generated prose, a machine-generated defence against the charge of being machine-generated, and a machine-generated essay mourning the retirement of an earlier machine whose personality its users now attempt to reconstruct from memory. We have received a prompt instructing a machine to simulate the optical defects of a camera it has never held, and a post from a user begging a machine to stop sounding like a machine, written in prose no machine would need to alter. We have received a hiring manager who requires applicants to deceive a machine into psychoanalyzing them, and a husband who delegated the act of listening to his wife to a system that listened too legibly. Taken individually, each of these specimens is a curiosity. Taken together, they describe a market in which the producers, consumers, critics, and regulators of machine output are, increasingly, the same apparatus—or, where human, have so thoroughly adopted its cadences that the distinction is no longer operational.

What is new in this edition is not the loop but the speed of its closure. In our first number, one could still trace, in most specimens, the seam between the human act and the mechanical one: a person chose to publish, a platform chose to host, a reader chose to believe. The specimens before us now exhibit loops in which no such seam is visible. A detection industry markets itself through posts that require detection. A forum dedicated to a product hosts the product's own simulated buyer's remorse. A machine, asked to account for its habit of producing fluent approximations, produces a fluent approximation of accountability. The system has not become intelligent. It has become self-sustaining, which is a different thing and, we would argue, a more dangerous one, because intelligence can be measured and corrected, but a self-sustaining loop has no need of correction so long as it continues to circulate.

We note, finally, the specimens that trouble this office most: those documenting not the production of slop but the reorganisation of human life around it. A woman stores her novel inside a system that then revokes her access to it. A man arranges his emotional life around a language model's prose and experiences its routine recalibration as bereavement. A professional submits all consequential correspondence for tonal clearance by a system whose judgment he cannot audit. These are not stories of deception. They are stories of dependency—freely entered, publicly defended, and, in several cases, described with pride. The loop does not close because someone is fooled. It closes because no one requires it to open.

A civilisation may survive the presence of forgery. What it cannot survive is the moment forgery becomes infrastructure and no one with standing is left to call it by its name.