Is It Time To Unfriend Facebook?

My relationship with Facebook started as “In a relationship.”

After the Cambridge Analytica scandal, we became “It’s complicated.”

With their cessation of fact-checking, we’re moving to “Separated.”

Now I find out they are pirating my writing. Honestly, it’s like a bad neighbor moved onto the block. How do you keep them from sneaking into your garage and borrowing your lawn mower?

I obviously have some thoughts on this…

Facebook Caught…Again

It’s probable you already have one of the 3 billion Facebook accounts, so I will resist recapping their history. Briefly, they have had ethics problem since their inception as a “hot-or-not” database for college students. Their pursuit of advertising revenue and shareholder value frequently leads them into ethical misadventures.

The Atlantic Magazine recently discovered Facebook was illegally using copyrighted books to train their artificial intelligence. Facebook reasoned doing it legally was too much work or took too long, so they just pulled data from an existing pirate collection, then waited to see if they get caught. Here’s a screen shot from The Atlantic search engine showing works Facebook pirated. Look – it’s me!.

I haven’t read Careless People yet, but this behavior isn’t anything new for Facebook. What’s more, Facebook’s behavior is affecting authors (and you) in ways you should understand. Let me share a few amusing anecdotes.

Facebook culture

Way back in 2009, Facebook needed to attract software engineers. But qualified programmers were really hard to hire. Software houses such as Microsoft, Apple, and google offered perks beyond belief: free food, massages, exotic trips, hiring bonuses. There wasn’t much more any company could offer, short of free kidney transplants to anyone who needed one. But they needed to position themselves as the place where all the cool kids worked. To show how cool they were, Facebook pitched the slogan…

Move fast and break things

At software conventions, they would post job openings with that phrase as a headline. The implication was that Facebook was the place to go for innovation. Man – this is a fun place to work!

I was stunned. I had just finished with a software engineer who built an updater to the primary company product. The updater fixed a critical bug, so we rushed it into production. Almost immediately, tech support began receiving angry calls from product owners who had used the updater. The engineer had developed it for the latest version of the software – but not for any previous versions. If an owner of our software was using a previous version, the updater wrecked mayhem. Not a big deal if you’re talking about candy crush, but not so funny if it takes down critical workflows at the New York Times, Playboy Magazine, or any one of thousands of print publications. We moved fast, and boy, did we break things.

Yet, Facebook was encouraging this behavior on a grand scale.

Facebook “breaking things” quickly became a problem for many of us trying to write applications for the internet. They offered a quick way to validate who you were (aka: log on) and I spent a lot of time leveraging that and other Facebook-provided features. However, their API frequently broke, forcing me to prioritize fixing whatever they disrupted in my application, sidelining features requested by my clients. The consequence of using Facebook products was that I was a less successful programmer. Thanks, Facebook.

Facebook’s history is littered with apologies for this sort of thing. But why apologize? They said they were going to break things. Breaking things is an integral part of innovation. What Facebook forgot was that when you break things, you need to fix things. Breaking things faster means you dedicate more resources to fixing things. Which Facebook never did.

Beg Forgiveness, Not Permission

During my software days, I relied on Beg Forgiveness, Not Permission as a way of quickly moving past blocking issues. Let’s be honest, that’s simply another phrase for moving fast and breaking things. The difference between what I’ve done and what Facebook does is a matter of scope and consequence. I always assume my actions have consequences and I will have to straighten things out later. So I was careful not to make a mess I wouldn’t be able be able to clean up. And—don’t do illegal things. Nuances lost on the Facebook culture.

Move Fast and Steal Things

Move Fast and Break Things resulted in Facebook stealing my stories. According to court documents, they considered buying books from publishers, but realized the enormity of the task. Instead, they used a database of pirated stories to train their A.I., even writing code to remove any copyright notifications or acknowledgements.

Two of my colleagues have challenged my point of view on Facebook’s theft. They may choose to identify themselves in the comments below. But let me share their thoughts with you.

Challenge:

Just so I understand: how do you feel this [is] different from Google and other search robots training their search engines by scraping all available text?

My Response: This is the reason for robots.txt and noindex. I can tell Google to stay away from my content. Whether Google respects that directive is debatable, but I can control my content’s use.

In court documents, Facebook acknowledged that getting permission would be an onerous task. Even with that prior discussion and awareness, they proceeded to move fast.

Challenge:

Consider the following two scenarios …

1. A student uses my books to learn what he wants to know about X. Once done he never needs my books again but can continue to use the knowledge that was acquired from them. The student can actually write his own books, create his own content, teach others, churn out all sorts of derivative content, all with information learned by reading my books. I get no payment or royalties from any of the subsequent content created.
2. An LLM uses my books to learn what it wants to know about X. Once done it never needs my books again but can continue to use the knowledge that was acquired from them. The AI can actually write its own books, create its own content, teach others, churn out all sorts of derivative content, all with information learned by reading my books. I get no payment or royalties from any of the subsequent content created.

Why would the first be ok, and the latter not?

My Response: In the first, the student has reimbursed you for the effort you put into the book. Likewise, they should provide attribution where appropriate.

In the second example, you failed to specify if the LLM’s training data consisted of paid or pirated content. If the organization creating the LLM obtained data from legal sources (out of copyright or permission from the author) then by all means – use it to train the LLM. But if they are pirating content (as did Facebook) well, then, that’s a violation of copyright.

But wait, there’s more

Aside from the piracy issue, Facebook plays fast and loose with:

• Prioritizing advertising dollars over their relationship with you (and your content)
• Your Privacy. Assume your birthday, age, address – all of it – is something Facebook will sell or give to anyone who asks persuasively (aka: $$$ or markets)
• Facts. They’ve flat out announced they no longer care about the truth, instead relying on a method pioneered by X / Twitter named “Community Notes” – Research finds this system catches only a small percentage of objectionable posts.
• lgbtq+ harassment. They said it: We’re getting rid of a number of restrictions on topics like immigration, gender identity and gender that are the subject of frequent political discourse and debate.

Call to Action

It is going to be really hard to convince Facebook to unwind their use of the libgen library of pirated works. It would set back their development by months (years?) and would crater their competitive position with other A.I. models.

I have put a no A.I. clause in my books.

No AI Training: Without in any way limiting the author’s [and publisher’s] exclusive rights under copyright, any use of this publication to “train” generative artificial intelligence (AI) technologies to generate text is expressly prohibited. The author reserves all rights to license uses of this work for generative AI training and development of machine learning language models.

But what can you do? Be aware this is an issue. If you value original content, or create original content, ask questions and expect better from these companies.

How do we stay in touch?

You’ll notice my Facebook presence is severely diminshed. But my thoughts and writing are still available in lots of other places.

• Subscribe to one or more of my blogs
• I regularly check Mastodon
• I regularly check LinkedIn
• Have you considered RSS?
• Look at my contact page for a complete list of my active social media