The machine did not arrive with the manuscript generator.
Amazon has publicly described relevance algorithms in product search, documented decades of recommender-system development, and continued adding machine-learning systems to book discovery. Retail search, recommendations, sales rank, ads, metadata, categories, and customer behavior all affect which books become visible.
Creators and publishers respond. They study covers, titles, categories, release timing, series read-through, advertising costs, and the signals a platform rewards. Sometimes that produces useful matching between readers and books. Sometimes it produces imitation, manipulation, trend-chasing, and a great many covers that seem to have attended the same branding seminar.
What changed with generative AI?
Generative tools can change the cost and speed of producing text, images, translations, metadata, and ads. That deserves analysis. But the economic pressure to make work legible to ranking systems is not new, and blaming every market distortion on generation technology conveniently lets retailers, ad markets, recommendation systems, and business models wander out the side door.
A serious account follows the entire pipeline: production, packaging, distribution, ranking, recommendation, advertising, purchase, and enforcement. “The book exists” is not the end of the system. It is barely where the interesting part starts.