Comment by herbstein | Hacker News Reader

Comment by herbstein | Hacker News Reader

Parent Full thread

You can know that you're spreading misinformation, creating political division, and harming children's mental health without being able to point to whether a specific piece of content is doing that.

If we rephrased the GP post slightly it would say something about creating a system of incentives that, knowingly or not, created these outcomes. It's in fact entirely possible the current system was implemented with no ill intent. The creation of the system itself isn't, as such, the critique. Rather, the critique is how they, after seeing what the outcome of the system is, choose to keep it in place.

These things are complex and unpredictable and fixing it isn't as simple as setting a boolean flag. What we want to see is the public realization from the executives at Facebook that what they're doing is harming vast amounts of people and a decisive commitment to concrete, proactive, and fundamental changes to the system. All of this backed up by short-term visible results in both the short-term and long-term behavior of the system and company.

You can know that, sure. But any "concrete, proactive, and fundamental changes to the system" require first deciding what content is "okay" and "not okay". Or somehow stopping people from spending time on Facebook products, which just moves the problem to TikTok (or whatever social network takes it place).

I see it as either we allow all content or no content regardless of the effect, any other method has facebook (or again a human moderator or ML model) deciding whats okay and not okay for our eyeballs to see. I don't want that.

We can't depend on tech companies to fix human societal issues. People have been spreading misinformation on the radio and tv, political division by partisan "news" sites, and children's mental health is affected when they compare themselves to peers in school. We can't blame facebook for these. The reality of the situation is that if Facebook makes any changes everyone will jump ship to another platform and the cycle will continue.

> any "concrete, proactive, and fundamental changes to the system" require first deciding what content is "okay" and "not okay"

But they're already doing this at massive scale with their algorithm. The issue is that the content they think is "okay" is the most inflammatory and tribal, because it generates the most interaction. If the idea is to let people have all the information and decide for themselves what's true and what's not, that's not what's happening now. What's happening now for many people is that the marketplace of ideas is restricted to bias-confirming conspiracy theories, and no other information is allowed to be seen by those people.

Not exactly. It's the other way around. Just like herbstein said earlier theres no boolean flag that tells facebook that something is "inflammatory and tribal". Its not like facebook has some algo that says if a republican posts and a bunch of democrats comment then boost this to a wider audience (repeat ad nauseum for every category).

Their algorithm is just identifying content that lots of people interact with and then it boosts that to a larger audience. That content happens to be inflammatory and tribal content because people, unsurprisingly, like interacting with that sort of content.

If people didn't interact with it, it wouldn't get boosted. So ultimately its society's fascination with inflammatory and tribal content that causes the algorithm to boost it.

Again people pretend like facebook somehow looks at a picture and says "wow this person is attractive, wouldn't it be nice to affect someone's mental health by showing this to them so they can compare themselves?" Thats not how it works.

It sounds like you think that I think that they’re showing people content with the intent of sewing unrest. I don’t; I believe their intent in all cases is to make money and expand their empire. That’s their intent when they’re boosting inflammatory content, suppressing content that might challenge people’s biases, and that’s also their intent when they ban users whose content is so inflammatory that it actually invites real-world violence.

It also sounds like the most important aspect to you is their intent? Like, did they mean to censor something, or were they just trying to increase engagement?

So let’s say an engineer comes up one day and says, hey boss, I did some A/B testing with three different algorithms: “A” is just a chronological timeline, but B and C each use a different method of boosting and suppressing content. B increases engagement with some users but also drives away some users altogether. C shows content that generates fewer engagement “hot spots” but shows higher user retention. And the boss says, we’re in it for the long haul; let’s go with C! By the way, what’s it do? And the engineer says, well, an AI is trained to look for posts containing political opinions outside the mainstream and shows a cat picture in the timeline instead. Guess it turns out people really like cats.

So would that be ok? After all, in this hypothetical situation their intent isn’t to harm anyone; it’s just to increase their user base and make more money.

Agreed, but I think what many critics of FB don't seem to realize is, the only solution to this that will really work, is for there not to be a social media network this big. Scale is its own problem. No executive will decide, "I'm too successful, let's split apart into dozens of smaller companies". The problem isn't how FB is run, it's what they are.

Reply on news.ycombinator.com