What I'm doing is simple to state and strange to do. I have started publishing books before they reach the level of human quotient that most people would be comfortable with.
Yesterday I published a book. It is at HQ 6.
If you have not come across the scale yet, here is the short version. Human Quotient is a 0-to-10 measure of how much human effort, intention, and craft went into a piece of work. It is not an AI detector and it is not a quality score. It is a self-assessment — an attestation by the creator about how the work was made. Sometimes the important thing to know about what you're reading is whether it was produced 100% by human or 0% by human. I put the scale together, and you can read about it at humanquotient.ai.
HQ 6 means something specific. It means AI generated the portions. I have reviewed every word of it. What it does not have yet is my tone and my voice — the book says what I want it to say, but it does not always say it the way I would say it. An HQ 6 book is accurate and unfinished. It is a real book with a real ISBN that is not yet the book I intend it to be.
So here is the experiment. With this book — it's called Unplacement — I am publishing first at HQ 6 and editing chapters up to HQ 8 over time. Every day you will see the book change. Different sections will shift, some passages will be rewritten, the voice will move closer to mine. The edition you buy on Tuesday will not be the edition that exists on Friday. The book is live in a way books have never been live before, and I am the one editing it in public.
I want to be honest about what this will cause. It will cause problems. A reader who buys the Tuesday edition and quotes it on Friday may find the passage has changed. A reviewer may feel they reviewed a moving target. A bookstore that prints a description will have a description of a book that no longer quite exists. Some of these problems are serious and some of them are just new, and part of what this experiment is for is finding out which is which. That is also why I am not publicizing the book beyond this post. People don't stumble upon books these days anyway. I am not going out of my way to talk about it, and I am not doing any publicity until I am comfortable with marketing it. What I am doing right now is the experiment itself.
The question underneath all of this is simple: can a book be published first and edited gradually over time? Software ships this way all the time. Version 1.0 goes out, updates follow, users live inside a thing that is never finished. Books have never worked that way. A book was a finished object or it was nothing — the printer's deadline decided what the words were, and after that the words were fixed forever. The economics of print made the book immovable, and we built our whole idea of authorship on top of that immovability.
AI removed the economics. When a book can be produced in days, the expensive part is no longer the making. The expensive part is the judgment — deciding what the words should be, which voice they should carry, whether they are right. And judgment improves over time, the way any craft improves. Which raises the question this experiment is really about: if the making is cheap and the judgment takes time, why should the release wait for the judgment?
There is an honest answer: because readers deserve finished work. And there is an honest reply: readers have always been able to buy first editions with mistakes in them, and nobody demanded the publisher recall the print run. The difference is that now the first edition can be corrected by tonight. The book does not wait for the next printing. It changes under its own roof.
So that is where things stand. A new book as of yesterday. It is changing and it is going to keep changing, quite often, out in the open, at a level of human quotient I would not have been comfortable showing anyone a year ago. Whether this turns out to be a sensible way to publish or a strange way to embarrass myself in public is exactly the kind of thing you only learn by doing it.
We'll see how it works. I will keep writing here as it goes — the corrections, the problems, and what the book turns into on the way from HQ 6 to HQ 8. If you want the thing I am measuring the work against, the scale and the reasoning behind it are here: humanquotient.ai.