Ask HN: To what extent does “shadow suppression” occur on social media? | Hacker News Reader

Ask HN: To what extent does “shadow suppression” occur on social media?

20 pointsrapnie18 comments

In another thread [0] I commented on the moderation practice of Shadowbanning. But I wonder to what extent there are even more insidious methods, which I coined "shadow suppression"?

Say, if instead of an all-out shadow ban a moderator (or automated scripts, for that matter) would tweak some algorithmic parameters that instruct the platform AI to limit someone's freedom of speech in very specific ways.

Just like a person based on their PII and content can be accurately targeted by ads, they can be accurately singled out and subtly limited in the amount of influence they can have on the platform.

Unlike a shadow ban this may never be noticed. A person's non-controversial opinion may pass through, while their political statements are directed to a small echo chamber of other people. They will never reach the audience they are intended for.

The possibilities for shadow suppression are limitless. Without algorithmic transparency how do we even know how far such practices are applied?

[0] https://news.ycombinator.com/item?id=28343848

I can't speak generally about all of Facebook because I don't know, but I can say that there are many surfaces on Facebook's products where people widely believe that shadow banning occurs, when in fact it does not, at least not in the way you're describing.

Generally what is perceived as shadow banning is just algorithms learning that the poster has low engagement, so it ranks their posted content (comments, posts) lower in the future.

> that the poster has low engagement

Beautifully said, this is a big elephant in the room.

There is a preliminary problem: the mission. "Intellectual curiosity", "Stay for the empathy", "Truth finding", "Thank you for sharing", "Ask the expert", "Feel the outrage", "Access knowledge", "Take this latest"¹ (and of course, "Be addicted!") [²] - these do not bring to the same place. And the wealth in one may conceal that you are in the wrong place.

(¹ "Take this latest": with reference to implementations that show the recent but irremediably hide the past. Like press under some dictatorships. Because if you were interested in Habermas' "chirps" on the events of May 2013, and to read Ferguson's interventions like a book, but you are supposed to be only interested in yesterday...)

Edit:

² I forgot a big one, implicit: "Build peer pressure by discovering others with a similar comfort zone". "...Today a page, tomorrow the world".

> algorithms learning that the poster has low engagement

Not specific to FB, but engagement is not the only thing a platform can optimize for, if a company wants to maximize revenue. Optimize the AI to avoid incentives for government regulation or lawsuits, or just to appeal to a country's regime, individual politicians or lobbyist groups can have huge impact on profits.

"We really want to operate in your country, but we don't want to be seen as submitting to your, umm.. particular netiquette guidance. Shall we design some policies together, that match both our interests?"

It's quite common. I worked on two small to medium sized social networks and it was already implemented early on. I was told it was used because of bots and not really to be used in human users.

PS. I know that Ask HN needs more upvotes to reach front page, but it strikes me as peculiar that with 14 points and 10 comments this hasn't happened yet (not on page 2-10 either), while earlier links submitted near my entry entered front page with just 3 upvotes and zero comments.

When I moderated reddit I always shadow banned bigots but left their messages to get seen by their friends to not arouse suspicion. Works very well

That scares me a little. Who gets to decide who's opinions are bigoted?

The guy above and all other volunteers

Have you ever considered that in doing this you've made it harder for said individuals to realise the error in their ways? Placing people inside echo chambers (involuntarily or otherwise) only allows said views to fester, potentially leading to something more dangerous.

The trolley problem. And A vicious cycle. Attribution error at the heart of it.

Opinions dissenting from Correct Thought seem to fare poorly on social media, including this site.

It's a cultural thing (if you meant that "Correct Thought" was different depending on the platform). I'm pretty ure this came from the US, and to be more precise, from McCarthyism, and bled through the cultural dominance to other western countries.

Actually "Correct Thought" is a Gene Wolfe reference =>

> Although Ascian is their mother tongue, adult Ascians don't understand plain Ascian sentences, unless they are direct quotations from governmental propaganda materials (called Correct Thought). So, in order to communicate, an Ascian has to know by heart thousands of these quotations (sentences) on many different topics.

https://en.m.wikipedia.org/wiki/Ascian_language

Even worse. So much of it is digital lookism. Failing to use the right words or not genuflecting appropriately. People don’t even engage with thoughts. They haven’t any.

> including this site

You figure the rest.

HN has all potential (the participants) to make clash fruitful; unfortunately, it lacks systems to prevent dead ends. (It is a pity though: well thought and mature ideas lost diluted in the space, and not meeting their natural matches in strenghtening, refinement, correction.) It is not "perfect" as a debate platform (it's not built to be a perfect debate platform as a goal - that would require proper exposure of different ideas and ingenuous organization of the confrontation spaces and methods).

Do opinions dissenting from Correct Thought fare better elsewhere?

There are very few human interaction spaces where such opinions are given oxygen, and it's usually done in a highly-codified context.

> including this site.

What I find most interesting is that on HN, the Correct Thought depends on the time of day. I have seen quite a few comments that either didn't get any votes or even got a few upvotes in the European morning. Then, later that day, when the US folks start reading it, the upvotes disappear and the comments got downvoted. Seen that several times...

How is that even possible on Reddit? You can configure Automod to remove comments by users but AFAIK you can't make it visible to friends.

Reply on news.ycombinator.com