It took decades to clearly define gambling enough to regulate it as we do now. Note that pinball was banned in major cities for decades because it was too similar to gambling devices.

And even with those definitions things weren't very clear. I used to work for financial traders here in Chicago. Every year there would be an unofficial snow futures market. It was a fancy way for all the traders and clerks to gamble on monthly snowfall at O'Hare.

But to my delight, the CME eventually created an actual, legal weather derivative for snow. What made one gambling and the other not? What makes any given bet a legitimate financial trade versus pure gambling? Let me know if you figure it out.

My point being that not having a clear definition won't stop us from grappling with the harms. Nor should it. Figuring out what new things really mean is always a messy process.

For me there is a simple test: how the algorithm decides what to put in my feed.

If the feed is made of posts belonging to the connections i follow (be it business pages or persons) in temporal order, then chances for addiction are low. If the algorithm decides for me what is relevant and from whom and in what order, then chances are high for addiction potential.

Algorithms are optimised for engagement (time spent there and interactions), not for my benefit (keeping up with my connections and interests). And they usually follow reinforcement/reward schedules -> specifically the Variable Ratio pattern. Just like gambling.

"For example, in a lab setting, this might involve delivering food pellets to a rat after one bar press, again after four bar presses, and then again after two bar presses. Gambling and lottery games are good examples of a reward based on a variable ratio schedule."

For FB, it could be how many irrelevant posts do i see before i get an update from a friend.

“Do you use personal / behavioral data / usage to shape the user experience and recommendations, creating a reinforcing loop?”

Roughly that.

The best idea ive seen is if you decide what a user sees in an algorithm way (beyond “natural” ones like chronological) you are liable for the content - you are no longer a common carrier and can be sued for damages. That should be step 1.

Tobacco regulations don't regulate the general case, they regulate tobacco products. I think the same will apply to social media. There's no need to define an addictive platform anymore than there's a need to define an addictive substance. Instead all that you need to define is a social media platform. And that can be expanded/adjusted as necessary.

As for Meta itself, and proving their case - I can't imagine it'll be difficult. Whatever they get in discovery + endless testimony from various people that have worked there will already likely leave it beyond any reasonable doubt.

A simple first step to me would be to update the DMCA so that platforms using engagement-maximizing algorithms lose their section 230 protections.

If you're targeting individuals with specific content algorithmically, you're hardly a dumb platform for user-generated content, and I would argue that it's well over the line that would classify you as a publisher.

In my humble opinion: anyone optimizing for "engagement" is trying to make a product inherently addictive.

Having a great product is not, but designing your product intentionally so people spend more time with it, than they wanted - intentionally not just let them write that one message, but distract them into other stuff - would qualify.

NAL, but looks like they're alleging something like negligence & deception. Addictiveness is just the mechanism of harm, so doesn't require a strict general case: just actual harm and the prior knowledge thereof.

So like the general argument is "company X was causing material harm, knew they were causing harm, and actively decided to continue causing harm while covering it up."

It'd be similar to suing someone who sold a dangerous toy under the label "safe for all ages". The lawyers don't need to ban specific features (sharp edges, fast motors, etc.). They just have to argue the way it's put together harms kids, the company knew about it, and they kept lying about it anyway.

The remediations could therefore be case-specific. They don't need to ban infinite scroll in general, just ban Meta from applying it to their products, since it's known to be harmful in that context.

You can just be explicit. We already do this with controlled substances, maintain a defined regulatory schedule, and give a designated regulatory body a legally prescribed process for adding to it, removing them, or changing their classification.

A trial is also the wrong venue. It isn't illegal to make an addictive website.

Instead lawmakers should be making new laws.

> We regulate gambling because it's a specific thing

Except when it's called "prediction markets" and therefore it isn't.

There are literally emails back and forth with Zuck where it's brought up that they're essentially making things as addictive as possible and it's bad for people, and he goes (to paraphrase) 'yeah, whatever. keep doing what you're doing'.

> But this feels legally unsatisfying in the general case

What if I told you that judicial outcomes are often nondeterministic?

It's hard to answer in general, but does it have to be answered in general? I could imagine supporting a policy that infinite scroll in particular is addictive and app developers can't use it without strong warning labels and time alerts.

Nit: Prediction markets are gambling and we dont regulate that.

I'm sure the companies marketing research divisions have very clear definitions.

We just use those.

[flagged]

[flagged]

Let’s start with banning algorithmic personalized feeds. That seems to be the common determinator of the bad apples.