Hacker News Reader

Comment by mrweasel | Hacker News Reader

One reason why Facebook keeps failing in increasing "meaningful social interaction" could be because they at the same time tries to sell ads.

It seems like it would be easy to simply exclude "reshares" from users news feed. They could try to: "Sort by date" and "only posts posted/shared by my friend directly". Personally I think that would remove much of the toxicity. Secondly, simply ban news sites from the platform.

Facebook is trying to fix an algorithm that can't possibly work, because it has two opposing jobs, increase time spend on the platform and at the same time it has to downplay the content people engages with the most.

View on HN

I think another factor is that they’re trying to quickly turn a massive ship that has, for years, self-selected users intent on heading in the direction of outrage and toxicity. The people who are the most “engaged” with the platform are the people who get a dopamine rush from arguing and baiting and instigating. The nice people who just want to see what their friends are up to have either left or significantly scaled back their use.

I think that makes any algorithmic “fix” much harder. Or perhaps even impossible.

I think the solutions you’ve suggested would have worked had they been implemented five years ago. I’m not so sure they would work today.

Anecdotally this is what I've been seeing over the years. Most of my friends and acquaintances seem to have left Facebook behind and moved to Insta. The people that stayed on FB are mostly those who like to express strong opinions and engage in pointless discussions.

> engage in pointless discussions.

Pointless to some, I'm sure, but if they're that engaged it's meaningful to them. Viewed through that lens, then attempting to externally manipulate and cajole their conversations was never going to have a positive outcome or even impact on the rest of the userbase.

It feels like this is a core point that's somehow got lost here.

There's a lot of assumptions in these discussions that connections and sharing are only good and useful if the results look a certain way. People sharing dog photos: good. Sharing approved mainstream news stories: good. People posting personal politics: bad. People disagreeing with each other: also bad. Etc. But it's a purely subjective take. Maybe our society needs to talk about politics way more than it does. Historically, instead of talking there's often just been fighting, and everyone agrees that talking is far healthier.

Facebook's original working assumption was that connecting people and letting them communicate and share is inherently a valuable thing. Is that actually wrong? Clearly, there is enormous demand to discuss politics that was unleashed by the internet, whereas before then only a small number of people had the access to present their opinions to those around them via opt-in channels like TV, the press, etc. Other people had to make do with phone calls. Something once restricted to a very small group is now available to all. Unsurprisingly, the sort of people who feel ill when they see the results are those whose views are the same as those who previously had a monopoly on group communication of various kinds. The ones most engaged are the ones who feel the media ignore their worldviews.

It's pretty dangerous to just say, well, if people are arguing about stuff that the establishment wouldn't let them argue about before, then it's "toxic". It's certainly risky to blame social media as the cause rather than merely a symptom of disagreements that would be there anyway. Far better to have "toxic" social media than a revolution!

The problem I see on FB is less with what people argue about, but how.

It feels like people are just saying things at each other trying to "win" an argument.

Of course, that's a problem with education, not FB. And I think it's a good thing that it's surfaced on FB.

FB can act as toxic behavior multiplier, though.

FB can act as toxic behavior multiplier, though.

Indeed, the medium can alter the message. For example, supposedly any extra lag in voice communication makes participants think the other person is angry.

We also see groups that never had close proximity now interacting directly and finding that they disagree pretty strongly on fundamentals.

So there may be ways to facilitate the same discussions, but more productively and humanely. And it simply might not be possible for people with diametrically opposed beliefs to do this -- there may need to be a game of a few layers of telephone, or at least an active moderator as in a formal debate, to help translate meaning and intent across the epistemological divide.

I don't know if it's a problem with education but more likely, deeply held implicit assumptions that aren't being articulated.

Most political disagreements fall into this category. E.g. arguments about COVID stuff are usually rooted in differences in assumptions about things like the trustworthiness of the establishment, the corrupting strength of profit, whether events are best explained as conspiracy or cockup, etc. Because they're sort of all-encompassing and vague people don't or can't easily spell them out, leading to the "talking at" not "talking with" problem. But that isn't to do with Facebook. It's inherent to humanity.

Sadly, some of my friends now use Insta to propagate the same level of political vitriol as they did on FB. They just leverage text overlays to accomplish their goal.

I see the same on WhatsApp too. Family members blindly forwarding whatever outrage meme of the moment lands in their message queue. Especially for family outside the US who use WhatsApp as their primary way to communicate with groups.

At this point the exact posts are just shared around platforms. Tweets posted on HN. Tweets posted to reddit. Facebook memes on instagram. Ticktock posts to instagram and reddit. Doesn't matter what platform you go on, the same fodder ends up everywhere.

Living in a country of 1.5 million people, it often feels devastating how big a part of political/intellectual discussion in our (tiny) language actually takes place in Facebook. I have worked in journalism, so maybe it's sour grapes in a way, but: most of the great writers are there. All the politicians are there. Our prime minister first announces crucial things on her FB wall, etc.

In short, if you want to participate in the public sphere of our tiny society that yet speaks its own tiny language, you'd better have a FB account. The account provided by an American Megacorp, that is. This actually feels really strange.

I've been more or less off FB for 5-6 years, occasionally feeling lonely and intellectually isolated. I would love to see things like the Fediverse/Mastodon or tilde.town take off for speakers of our language, but I'm having doubts whether this could ever happen.

For the lingua francas, there will always be a considerable chunk of alternative culture and great, bright discussion outside dominant platforms (howdy, HN!). For smaller cultures, this relationship status is surely complicated. FB is just so damn big.

I feel you can apply the maxim 'in politics nothing happens by accident' to Facebook despite their earnest attempts to claim unintended consequences. They are a highly important component of our divide and rule political era on multiple levels and sadly are doing an excellent job.

True. In the early 2010s, I used to ponder whether journalism is actually dying -- meaning, now that we have social media, traditional gatekeeper-journalism [1] is simply not needed any more. During the passed decade, it has become obvious that as long as a social network is also a megacorporation, this idea won't work. We still need the better parts of traditional journalism to check what's going on behind the curtains of profit-driven social networks. Really messy, but also truly interesting times for journalism for sure.

1: https://en.wikipedia.org/wiki/Gatekeeping_(communication)

The huge challenge for journalism is the reality that professionals are almost always paid to promote the agenda of their employer and advertisers. Independent, investigative reporting is very rare in corporate media these days. Many (Chris Hedges, ex NYT foreign correspondent is a good example) set up as independents in the golden era of blogging prior to the social media/smartphone tsunamai.

I would argue people such as Matt Taibbi, Matt Stoller, Michael Tracey, Chris Hedges in the US are the true gatekeepers right now based on my understanding of the planet. These people appear to survive by publishing books and keeping up a strong social media presence critiquing the platforms that host them, and with companies like substack paying them to promote their platforms.

> I think that makes any algorithmic “fix” much harder. Or perhaps even impossible.

Optimize for another metric than engagement.

> I think another factor is that they’re trying to quickly turn a massive ship that has, for years, self-selected users intent on heading in the direction of outrage and toxicity.

But are they actually trying to turn? They say they are but all evidence points to them maintaining the status quo because fear and outrage boost engagement and engagement is king.

Well, that and rationalization. Everyone's subject to rationalization.

You can say 'maintaining the status quo because fear and outrage boost engagement'…

…and they can think 'we represent the social engagement of ALL people, how dare you tell us not to support genocide fanciers when our data shows that's a solid 27% of our actual platform'…

…and then they can say 'we are trying to turn the massive ship, Senator' and tell each other 'stay the course, we have a responsibility to ALL our users', and the whole time, engagement will still be king.

People don't always know why they're doing what they're doing, and even if they do, they won't always tell you.

Relevant CGPGrey: https://m.youtube.com/watch?v=rE3j_RHkqJc

Right on. I've seen the posts/posters in many local fora in the UK have two defining characteristics. They are judgemental, and they are nosy in equal measure. [1]

I put the recent increase in this down to people's familiarity with the technology, having grown up with it -- those people only 10/15 years older have seen life before and after the revolution, and behave differently. Specifically, when local fora started to make an appearance, they were a bit more diffident in their postings.

[1] (I don't want to start splitting hairs on which might be greater. :-))

I forgot about the incessant hand-wringing and virtue signalling posts (typically made after someone reports a crime such as a street robbery), and these are also made from behind anonymous usernames. (any social capital created anonymously like this is useless in real life, so the motivation must be for some other reason such as entertainment or personal satisfaction)

Note, this is not me passing judgement on people -- quite the opposite -- the whole point is that they can't and wouldn't be like this in real life, but the medium of the forum allows for it, which makes this cocktail of usage characteristics on these fora a recipe for quite strange reading, over time.

Not really: ads have a budget of attention within an envelope that Engagement (the team whose work is described here) is trying to maximise. There’s detailed estimates to how much ads take away but it’s minimised and isolated from that conversation completely.

There are many issues, some addressed there; some more nuanced (figuring out that is a good post or comment is genuinely hard with crude metrics because clickbait and flamewars look very much like compelling content); some that I can’t mention publicly.

There are non-scalable ideas for solutions (my PhD was around complex network so check Jure Leskovec’s research for ideas like isolating cliques and defining them as meaningful or not). One idea that I’ve floated, but less than I should, and many other have too, but more than they should (or at least not articulated well around this problem of sorting good from bad content to inform the News Feed algorithm): have secret negative and positive reactions per post and comment. The fact that Likes are public makes them performative, both very compelling (it’s who you want your friends to see in you) and counter-productive. But giving people the ability to say: "look, I will comment on every flame bait that I see, because I can’t help myself, but please, remove from from my feed" would have helped argue against what is still a very linear model of weighting acting that are uniformly seen as "engagement".

Source: I was the DS looking into those questions in 2015.

While I'm sure there are some people at fb genuinely trying to "fix" toxicity, overall the issue is that they're not trying to fix it. They want to do the minimum possible to protect their image / bottom line, they don't care if it actually works. Actually fixing toxicity is one way to improve their image, but appearing to try can be good enough.

I think all of the problems they have are earned. They've got, what, 3 billion users? How many 9s (99.99999) effective does the solution need to be to wrangle edge cases? Moderation is an eternal issue.

It's really difficult to make that kind of change from the rank and file. Once revenue is involved, it's amazing how nervous people get and nobody wants to approve some change that might lose as little as fractions of a percent of the money stream. There has to be a clear, explicit mandate from leadership that acknowledges there will be a revenue hit, and absolves any project members from bad outcomes / or rewards them for participation. I.e. it's hopeless :-)

Yup, which points out the fundamental problem: Facebook is a product of its incentives. Its incentive is to maximize engagement. Why? Because more engagement equates to more ad revenue. The externalities of this behavior do not align with what society seeks. Therefore, if we wish to fundamentally transform Facebook, then we must change the incentives. The incentives are not going to change on their own. Enough time has passed to demonstrate that the market will not correct for this. This can be deemed a market failure. This means some intervention by the state must be made. I propose that having the platform free should be illegal. Instead, the company should be forced to charge customers a recurring fee, sufficient enough to cover operation costs, and to no longer obtain revenue through advertising. This, I believe, will shift the incentives to produce the behavior we as a society seek. Facebook will no longer optimize for engagement, as that no longer drives revenue, and must instead compete on something else, like quality of experience on the platform. Of course, such policy would have to be applied across the board for all social networks as to not undermine Facebook's existence with a competitor offering the same service for free.

Couldn't a company outside the US just set up a free Facebook competitor then?

True. There is nothing that would prevent a software company within Sweden, let's say, from opening a free Facebook competitor. Given the reality, such a law as I proposed would be seen as unfair from Facebook's perspective. In many ways, I would say I agree with them. To remedy this, the law could be constructed so that it is only illegal within the confines of the United States. In other words, a paid subscription is mandatory within the United States, while it can be offered for free for anyone outside.

This is my take as well. It's the business model itself that's toxic, or at least predatory, as the necessary outcome of it is to lean as hard as possible on known frailties of human behavior and psychology in order to cultivate advertising placement and revenue.

There's no tinkering around the edges of that control loop that's going to fix the problem. What we observe as the deleterious effects on society are in fact systematic (as in, "design failures") faults in the thesis of Facebook's business model.

> Facebook is trying to fix an algorithm that can't possibly work, because it has two opposing jobs, increase time spend on the platform and at the same time it has to downplay the content people engages with the most.

Would it be fair to say they're trying to increase time spent by existing users on the site? Does Facebook have such a huge number of users there's no growth from building a better product? I personally think the product is awful.

I have an account, but I only log in often enough to keep it active. If I had a nice, reverse chronological feed of things posted (not shared) by my friends I would read it.

Maybe the subset of people like me is too small to worry about, but there's definitely some of us where the current algorithms work against engagement.

> Does Facebook have such a huge number of users there's no growth from building a better product?

https://www.statista.com/statistics/264810/number-of-monthly...

>How many users does Facebook have? With roughly 2.89 billion monthly active users as of the second quarter of 2021

>MAU: A registered active user who logged in and visited Facebook through the website, mobile app, or Messenger application in the last 30 days as of the date of measurement.

It would be physically possible to have better engagement, yes. But they're already the most wildly used website/software platform/media outlet in the history of humanity. Given from how much regulatory attention they're attracting, you could make a solid argument as a PM that it would be better for the company to reduce MAUs and spike user logins. From how Congress is talking, they're not far from just ordering Facebook to be shut down, which will reduce MAU to 0.

The Facebook Integrity team agrees with you, Zuck just doesn't want to give up the cash.

"Early tests showed how reducing [downstream MSI- the likelihood of creating long chains of reshares] of the algorithm for civic and health information helped reduce the proliferation of false content. Facebook made the change for those categories in the spring of 2020. When Ms. Stepanov presented Mr. Zuckerberg with the integrity team's proposal to expand that change beyond civic and health content -- and a few countries such as Ethiopia and Myanmar where changes were already being made -- Mr. Zuckerberg said he didn't want to pursue it if it reduced user engagement, according to the documents."

The article makes it seem like people at Facebook have ideas on how to potentially thread the needle, but Zuckerberg (and probably a handful of other senior executives) keep putting the kibosh on it.

We know there are some people with pretty idiosyncratic political beliefs and unsavory connections on Facebook's board and within its senior executive leadership. It's not hard to imagine that maybe they like it being an engine for stoking outrage and perpetuating disinformation for reasons unrelated to the financials or for any intelligible design/engineering logic. It doesn't even necessarily have to be malicious. They could literally just be blinkered by a weird ideological belief in "absolute" free-speech.

The only time I ever have meaningful interactions on Facebook is in niche groups like meme groups with a few thousand people.

Every single one gets banned eventually. My favourite Top Gear meme group has been through I think 6 groups now.

Facebooks potential to destroy culture if they throw a hissy fit is dangerous and not being paid attention to.

>My favourite Top Gear meme group has been through I think 6 groups now.

I think I know the one you mean, that group's been through quite a few "Zuckings".

This is a social science experiment with 3 billion guinea pigs.

Every time you rerun it, if someone somewhere (and the people connected to them get effected) how many attempts do you get?

Facebook does most of their experiments by rolling out changes to a smaller subset of users. The number of guinea pigs is much smaller.

Animal testing is regulated activity. Do you know why?

Agreed.

But I think big tech didn't use their power years ago when advertisers started demanding their ads don't appear next to X, where X is the offensive thing/position/etc.

Big tech has power but it folded and immediately started policing when this happened. Instead of a united front each of them caved to advertisers and now the content on each platform must be kosher according to the ethics that the platforms' rulers have.

Big tech never had that power versus advertisers. There's no possible scenario where major mass market advertisers like Coca-Cola or Toyota would ever tolerate having their ads displayed on highly offensive content, like say a Facebook group for white supremacists. If the social media companies hadn't imposed stricter content censorship then the advertisers would have simply dropped those channels to protect their brand images.

Reply on news.ycombinator.com