Short Circuiting the Author-Agent-Publisher Mess
Short Circuiting the Author-Agent-Publisher Mess
AI is going to disintermediate the publishing world
On Sunday I was inspired to embed Security Yearbook 2022 via vector mapping so that anyone could use ChatGPT to ask questions of the book. The result was astounding.
I have been working on an idea for a couple of years. GPT-4 will make this so easy to implement that somebody is going to do it. They are probably working on it right now.
If you are a writer you are familiar with the cumbersome and frustrating path to getting published. First you query an agent. Actually, you query hundreds of agents. You compose perfectly crafted pitches for your novel. An opening line that summarizes a story that took you 250K words and two years to write. A few lines describing why you are the person to write this story, and then some lines that demonstrate that you have researched the agent and why this book is perfect for them.
If your perfect pitch lands on the perfect agent’s screen at the perfect time, the agent may request a copy of the manuscript which *has* to be submitted to the agents specifications as to font, spacing, and text size.
The agent then pitches your novel to publishers. For their work they get 20% of everything your book earns.
Many best selling authors have been rejected hundreds of times before an agent agrees to work with them. And then they are rejected by publishers. Harry Potter and the Philosopher's Stone was rejected by twelve publishers before Bloomsbury picked it up. The series has sold over 600 million copies.
Imagine all the books that have never made it into print!
Here is how AI is going to disintermediate the publishing world.
First, every publisher embeds their books in GPT-4. (Most are already incorporated.)
Second, an author submits her manuscript and GPT-4 evaluates it for style, subject matter, plot, characters, everything. It matches the submitted book to publishers which have had success with similar books. The publisher gets an alert with scores and even predictions of future sales.
The publisher sends a contract to the author. If multiple publishers want it, an auction occurs.
That’s it. Publishers get a shot at all the good books. Authors get published. Agents get displaced. Readers get more books.
Most manuscripts are pretty bad though. In that case GPT-4 could offer detailed suggestions on how to make it better. It could even suggest changes that would fit particular publisher’s expectations. Instead of outright rejection or ghosting the author gets feedback.
I am so sure this is going to happen I will update this column when it does.