A hardcover book printed in 1925 is almost indistinguishable from one printed yesterday. It’s easy to think not much has changed.

But book publishing isn’t about printing, and it’s a useful metaphor for the systems changes we’re seeing all around us.

The book publishing system was based on scarcity.

A successful bookstore was perfect. It had exactly the right number of books — more wouldn’t fit, and fewer wouldn’t pay the rent. The only way for a book publisher to get a new book into the stores was to get the bookseller to take an old book out.

As a result of this chokepoint, distribution became the focus. Publishers came to see bookstores, not readers, as their customers—which is why there are few ads for books, or toll-free numbers to call. There were plenty of authors, so publishers selected which ones got a distribution investment. And their timing and launch strategies all revolved around the bookstores.

Bookstores have to make smart choices. Months in advance, they choose which new books to take on (and which to leave behind.) If they were wrong, if a new book they don’t carry has an audience, then they lose sales because readers go elsewhere.

The small change? Get rid of the scarcity of shelf space. Amazon never removes a book to make room for a new book. They have all the books.

The publishers’ existing strategies make little sense when the scarcity of shelf space goes away.

One industry term is the “lay down” which describes how many books a major publisher needs to print and distribute to get good nationwide coverage at launch. For books that hope to be bestsellers, that number was 25,000 copies or so… a book from a well-known author would have that many copies in the world before a single copy was sold.

Today, for many books like this, the laydown is 250. 1% of what it used to be.

This is why the industry is shifting so much attention to pre-orders. The online world not only eradicated space (you can buy things from anywhere, so shelves don’t matter), it also shifted time. You can indicate interest by buying things long before they’re distributed.

Bookstores don’t stock a new book unless they see it’s already been selling online.

Another example: Pop music.

Through a happy accident, the typical record store was exactly big enough to hold all the music that the typical listener might ever hear on the radio. The radio as a sampling medium was about the same size as the physical distribution medium of the store. You didn’t hear hula music on the radio and you couldn’t buy it at Tower Records.

First, we blew up the store. The internet meant that any song you wanted, you could download for free if you cared enough, or listen to it on YouTube (if you only cared a little.)

Then, we blew up the radio station. The internet meant that the sampling medium went from DJ-curated to streaming-on-demand. And we demanded.

Change the distribution, change the medium.

There are still hits, but they’re not driven by A&R teams, record-store distribution deals or payola. The sampling medium and the revenue medium have become the same.

And one more shift, one that’s changed both industries:

The cost of making a book or a song has plummeted. Thanks to AI, autotune and other tools, combined with the roll-your-own distribution of ebooks and social media, anyone can create and self-publish. So, anyone will.

Scarcity of creation and scarcity of distribution have been replaced by a surplus of both.

What doesn’t scale? Trust, attention and belonging.

AI is making relatively small changes to very big systems, everywhere we look. But if those systems are built on the desires of humans, we will need to earn trust, attention and belonging more than ever before.