Amazon's Kindle Direct Publishing (KDP) platform is, at this exact moment in October 2025, an absolute, unmitigated, raging dumpster fire.
It is choked with garbage.
Thanks to the AI e-book boom, the platform has been flooded by millions of 30-page "books" that are nothing but raw, unedited, AI-generated slop. The "authors" didn't even read them. They just generated, uploaded, and prayed for a quick buck.
They created a digital landfill.
And this is the best possible news for you.
Why the Dumpster Fire Is an Opportunity
The lazy prospectors did what they always do: they ruined the easy money. But in doing so, they created a massive, hungry, desperate market for one thing: quality.
Readers aren't stupid. They're leaving scathing one-star reviews on all that AI slop. Amazon itself is cracking down, purging accounts that publish obvious garbage.
The noise has become so deafening that real, human-curated signal stands out like a lighthouse.
The barrier to entry didn't vanish. It just moved.
- Old Barrier: Can you find the time to write a 30,000-word book?
- New Barrier: Can you create quality in an ocean of automated trash?
The people who fail at this are the ones who think AI is the "author." The people who win are the ones who understand AI is just an intern.
You Are Not an Author. You Are a Publisher.
Stop being romantic about this. You are not writing the next great novel. You are a publisher building a library of niche, non-fiction assets.
One book is a lottery ticket. 50 books is a business.
AI hasn't replaced the author; it has just made the publisher 10x more efficient.
Here is the real workflow.
- Niche Research (The Real Job): You don't guess. You use Amazon's search bar to find what people are already looking for. You find niches with high demand but terrible, low-quality books. "Weight loss" is a terrible niche. "Intermittent Fasting for Men Over 50" is a great one.
- Outline (AI-Assisted): You use AI to create a comprehensive, logical outline for your book.
- Drafting (AI-Intern): You feed the AI one chapter at a time, using a detailed prompt to get a 2,000-word first draft.
- Editing (The 50%): This is where you earn your money. You take that 80% draft and you attack it. You remove the repetitive AI phrasing. You add your own voice, personality, and real-world stories.
- Verification (The Non-Negotiable): You fact-check everything. AI hallucinates statistics, dates, and "facts" with terrifying confidence. If you publish a health book with fake medical advice, you are liable. This is the step the "slop" publishers skip.
- Packaging (The Gatekeeper): You hire a real cover designer on Fiverr for $100. A terrible, homemade cover is an instant signal of a terrible, homemade book. It is the #1 reason you will fail.
The lazy publisher tries to turn a 3-week process into a 3-hour one. They fail.
The smart publisher uses AI to turn a 3-month process into a 3-week one.
That's the entire game. You're not racing to the bottom. You're using AI to build a brand of quality, faster than any human ever could before. While everyone else is getting 1-star reviews for their "slop," you'll be building a back-catalog of assets that pay you for the next decade.
Be the signal.
August 2026 update: Amazon moved, partially
Editor's note. Written October 2025. The platform rules described here have since hardened. Original text unchanged above.
The essay bet that Amazon would crack down and that quality would become the moat. Amazon did crack down — less dramatically than "purging accounts" implies, but concretely:
- A hard upload cap. Standard KDP accounts can publish up to three new titles per day. Introduced in late 2024, still in force in 2026. This is the single change that most damaged the pure-volume slop model.
- Mandatory AI disclosure. If a book's text, images or translation were AI-generated, you have to tell Amazon at upload. If AI merely helped you refine work you created, you don't. The distinction matters and people get it wrong constantly.
Two things worth being precise about, because they're widely misreported. The disclosure is internal — readers never see it on your product page, and disclosing does not by itself hurt a book's availability. And the cap is on uploads per day, not on how many books you may own. Nothing here stops a patient publisher from building the 50-book catalogue the essay describes.
What the essay did not anticipate: the cap plus disclosure mostly punished speed, not laziness. A slop publisher who slows to three a day is still a slop publisher. Reader reviews and the store's own ranking, not policy, are still doing most of the actual filtering. "Be the signal" remains the right advice for uncynical reasons — it's just not being enforced from above as much as we hoped.