Yann LeCun’s prestigious career and well-deserved reputation as a cutting-edge researcher in deep learning and artificial intelligence led him to be hired by Meta (then Facebook) in 2013 to establish the company’s AI research division.
In 2018 he stepped down from management to take a more active research role at what is now called the Fundamental AI Research Lab (FAIR). LeCun also teaches at the Courant Institute of Mathematical Sciences at NYU, where he’s the Silver Professor of Computer Science.
LeCun, who has over 600,000 followers on X, is also a frequent commenter on public debates about AI technology. He’s been particularly active in pushing back on the idea that AI poses any sort of significant or extinction-level dangers to humanity.
But on Monday, LeCun kicked off the year by sparking a firestorm of controversy when he tried his hand at a prognosis for the publishing business. LeCun’s suggestion was simple. Given that many authors make barely any money from publishing, authors should instead give their books away for free.
“Only a small number of book authors make significant money from book sales,” LeCun wrote in a post on X. “This seems to suggest that most books should be freely available for download. The lost revenue for authors would be small, and the benefits to society large by comparison.”
“2024’s first garbage take,” said Reid Southen in a viral post reacting to LeCun’s comment. “You should work for free because hey, you’re barely making any money anyway!”
LeCun’s post was a reaction to an article in Publisher’s Weekly about an Author’s Guild survey which found that the median gross pre-tax income for authors for their books was $2,000 in 2022. But users pointed out that even though that is the case, authors who publish a book know this and are taking a risk with the potential of bigger rewards.
The take also comes in the wake of the New York Times massive new lawsuit against Microsoft and OpenAI for copyright infringement, as well as MidJourney revealing what artists it was trained on as part of a lawsuit.
Nicolas Nassim Taleb, a statistician and author of popular books on risk, claimed that LeCun’s model didn’t understand the possibility of big rewards authors could be deprived of if they followed LeCun’s suggestions.
“People don’t easily get skewness,” Taleb wrote, referencing a statistical concept—which can be either positive or negative—which demonstrates that data on a bell curve isn’t usually distributed symmetrically on both sides of a median. “$2K is the MEDIAN not the MEAN. This logic implies selling tail options because they ‘only a small number’ pay off, or never insuring your house because…”
Meaning LeCun was ignoring the potential for a big payoff if a book an author writes gets very popular.
Other also posters pushed back on LeCun’s comments, claiming he didn’t understand that even a small payoff from publishing a book could make a difference in some author’s lives.
“$2,000 may be small for you but for the average people it can be a material amount that they can’t just give up without a 2nd thought,” wrote @larry0x.
And others pointed out that there already was a free model where readers can get a copy of an author’s books: the library.
“Has Yann heard of a library,” asked @MellieDM.
“It’s wild that the takeaway here is basically, ‘Oh, you’re struggling to make money and provide for yourself and family? How about even less money? You know, for humanity?’” commented Southen in a follow-up post.
The pushback led to LeCun posting a longer post clarifying his comments where he claimed that he critics had completely missed his point. LeCun said there were many reasons for adopting the free model of publishing, including the fact that many authors are motivated more by intellectual production than income, that giving away free downloads might increase sales of physical books, and that a lot of scholarly and academic writing doesn’t generate income for it authors, who get their income from other sources.
“Computer scientists are also familiar with the concept,” LeCun wrote. “It’s called open-source software. You give away your software for free. Sometimes, your employer pays you to do so. Sometimes, you just want to make a name for yourself by contributing to an important project.”
The open-source idea resonated with some, particularly programmers where the model is often the norm.
“[That] was one of the motivators for me to do open source :-)” wrote @migueldeicaza. “My products were not making a lot of money at the time. I read the GNU manifesto and realized ‘this is a better way.’”
The GNU Manifesto is a foundational text in the open-source software movement that stakes a claim to making software both free in the sense that you don’t pay for it and free in the sense that software should be completely owned by its user, who can customize it however they want.
But others pushed back on the idea that the books and software are comparable.
“There is some merit in the idea that knowledge should be free,” wrote @aathanor. “But there is no parallel whatsoever between open source and free books and you know it. Price is a measure of value, sometimes flawed but generally right.”
LeCun’s idea that giving away free copies of a book generates more sales is common in ebook marketing how-tos, but isn’t supported by much research. One review of a 2011 study tracking 41 books found a suggestion of a positive relationship between free digital books and short-term print sales. “However, the availability of free digital books did not always lead to increased print sales,” the study concluded.
LeCun didn’t respond to questions about what he was basing that belief on.
And other posters pointed out that despite LeCun’s suggestion, his own 2019 book on artificial intelligence is currently on sale on Amazon for between $15-$20.
“Walk the talk — why don’t you start with your books?” asked @mlopes.
