Publishers Weekly recently hosted a stimulating and smart online session about AI and publishing, thanks to the organizing and moderating skills of Peter Brantley and Thad McIlroy. The day began with a presentation by former PRH CEO Markus Dohle and ended with one by thought leader Ethan Mollick of the Wharton School, which framed the day perfectly. Both of them were enthusiasts for AI. But they also presented what, to me, was the stark contrast on display for the whole day between people who think book publishing is largely the business it has always been and those who are seeing it isn’t and won’t ever be again.

Here’s the long story short. The task of establishing a new title in the marketplace has gotten progressively more difficult for more than two decades and will continue to. A business that used to be primarily focused on the current batch of titles is now increasingly attentive to the long tail backlist. This is structural, not cultural.

When I was a pup, books were mostly sold in bookstores and bookstores focused their stocking attention on what was new and likely to be hot. A plethora of book review media — led by the NY Times Book Review but complemented by reviews in newspapers and magazines and author appearances on radio and TV timed to a publication date, combined with robust and ubiquitous display of featured new books in thousands of bookstores across the country — enabled thousands of titles every year from publishers large and small to sell many thousands of copies, many doing well enough to “backlist” and become enduring sellers.

But, in those days, they didn’t all become enduring sellers that the publisher would keep available “forever”. Keeping books “in print” was not automatic, even though the penalty for not doing so was frequently losing the rights to that book to author reversion. In the 20th century, each “next printing” was a bit of a commitment, usually in the low thousands of copies. If the book was only selling dozens or hundreds a year, it made economic sense to kill it.

Two things changed that calculus. Ebooks don’t require inventory. And Ingram’s increasing competence with print-on-demand — tied, of course, to their wholesaling operation that reaches every bookstore and online seller in the world — meant there was a lower-margin but virtually investment-free alternative to ceasing publication of any title.

And, at the same time, the shape of the market was changing, and continues to. Most books consumers buy are still printed rather than electronic, but stores are not where most of them are purchased. Various online resources, starting with Amazon but also through an increasingly long tail of web sites sourcing primarily through Ingram (who will ship direct to the website’s customer, requiring no “handling” by the creator of the sale), now sell the vast majority of the print books (and, of course, ALL the ebooks.)

Although I get very little pushback from sage publishing veterans when I suggest that “trade publishing as we knew it is dead”, it isn’t at all clear that the surviving incumbents see it that way. (In fact, Dohle pretty much said these are the “best of times”.) They all know, of course, that their sales increasingly come from the thousands of titles they control from their years of publishing rather than the new titles issued this year or recently. But nobody has (yet) said publicly “publishing new titles is getting very hard” or “we’re going to publish a lot fewer new books.” (Of course, whatever the established publishers do, the world will publish many more new books; more in a month than we used to get in a year.)

But it does appear that cutbacks by the big publishers on new title publishing is the current reality. Agents are well aware that their business just gets harder and harder. Big publishers are increasingly encouraging longtime veterans of their companies to take retirement packages. In the 20th century, it was a perfectly reasonable ambition to start at an entry level position at a publisher and have it lead to a many decades-long career. Very few sentient young people would see that as a sensible plan or expectation today.

The math that explains this is compelling. In 1990, there were about 500,000 titles in print in the English language in the world, and not all of them were easily available. Today there are about 20 million titles in Ingram’s Lightning Print database and if you order one tonight, they’ll print it in the next day or two and send it to you. So the competitive set for each new title coming into the world has increased by FORTY times. And the publishers no longer have a moat around their new title launchpads. Bookstores can’t sell enough to make a book happen; the marketplace is online and incumbent publishers have a vastly diminished advantage in that world over many other players.

In fact, publishers increasingly depending on “Internet influencers” to push their books to their loyal followers are now seeing those very marketers turned into competitors by the start-up Bindery Books. So agents will now be considering Bindery as an alternative to PRH or HarperCollins.

And all of this has happened without any help from AI.

One thing AI threatens, of course, is a massive increase in the number of titles made available. One observation missing from the mostly-fabulous AI presentation from PW was the acknowledgment that a 40-fold increase in actively competitive titles has already taken place over the past two decades. There is very little doubt that a new surge of titles in the marketplace from AI will only compound the situation that has changed the landscape over the past two decades.

I’m very bullish on AI and what it will do in many ways. What the PW presentation made clear is that, across the industry, people are experimenting with it for a multitude of tasks: title creation (of course, although that didn’t get the emphasis it deserved in the PW event), marketing from creating copy to finding audiences, and in production in a multitude of ways. But also clear, as much from audience reaction as from any particular presenter, was that the image of an industry where the author writes a book (and fiction was most frequently discussed), gives it to an agent, who sells it to an editor who shepherds it through a publishing process, which with many “rinse and repeats” which result in authors making a living from selling books, is more vivid in many minds and hearts than it is reality anymore.

The day started with Dohle saying this was a great time for book publishing. Which, indeed, might be true for PRH with its nonpareil backlist and bigger-than-anybody scope. But it isn’t generally true. The view from here is that trade books will have two distributors by the end of this decade: one publisher (probably PRH) and Ingram. Books will continue to be published in increasing numbers because you no longer need to have a big organization under you to publish one, and the investment required to publish any single title (beyond acquiring the content) is less than it has ever been.

The PW AI day ended with Ethan Mollick talking about the myriad ways to use AI and making it clear that AI is only going to get better and more powerful. And that people trying different ways to use it was going to be the path to making it an increasingly important part of the workflow.

Whether new leaner publishers with Internet followings and AI “staffs” will enable more authors to have successful careers in the new publishing era is something we will just have to find out. But the view from here is that anybody who expects anything like a revival of traditional trade publishing, particularly one that provides a sizable number of authors with a real living, is going to be very disappointed.