How Facebook got addicted to spreading misinformation
technologyreview.com
technologyreview.com
Since then, other employees have corroborated these findings. A former Facebook AI researcher who joined in 2018 says he and his team conducted “study after study” confirming the same basic idea: models that maximize engagement increase polarization. They could easily track how strongly users agreed or disagreed on different issues, what content they liked to engage with, and how their stances changed as a result. Regardless of the issue, the models learned to feed users increasingly extreme viewpoints. “Over time they measurably become more polarized,” he says."
Facebook has repeatedly shown that it will maximize its bottom line and growth (in particular, as measured by engagement metrics) above everything else. Fair enough, they're a for-profit corporation! I just can't really take seriously their claims of 'self-regulation'. They simply can never be trusted to self-regulate because it's just not in their interest, and they've never shown meaningful actions in this regard. They will only respond to the legal stick - and it is most certainly coming.
"But anything that reduced engagement, even for reasons such as not exacerbating someone’s depression, led to a lot of hemming and hawing among leadership. With their performance reviews and salaries tied to the successful completion of projects, employees quickly learned to drop those that received pushback and continue working on those dictated from the top down."
The more shocking it is, the more eyeballs it attracts and the more add revenue it gets.
News broadcast is a better analogy but still not as addictive as something that continually reaches out to you throughout the day.
Even the most nihilist shock-jock has to pick a speed that won't alienate half his audience or drive all the advertisers away from the network. In these systems, those counterweights don't exist.
The scale of reach is tremendous, faster and optimized in a way newspapers/media could only dream of.
It's also not publicly clear what is being widely disseminated and to whom, other than when outside parties analyze what's happening within Facebook on a given day (and this has limits).
Really, I see Facebook as the A-bomb of virality and fomenting extremism. It is devastatingly effective. That's why they've made so much money. Much more than the NYTimes or National Enquirer or Drudge Report. It's not just another media source, it's a game changer at harnessing people's attention and outraging them as quickly as possible. At scale, with 1 billion users, this has world-changing effects.
Tabloids, newspapers, news broadcasts are a single artifact delivered to many people. They have the same incentive to sensationalize, this is tempered though by the fact that they leave a public record that people can discuss, scrutinize, and will ultimately impact their reputation.
When a paper consistently prints outlandish stories of bat-boys and lizard-ufos, everyone can see that and the paper will get a reputation as being a rag.
In contrast, the feed of posts you see on your facebook (or twitter, or instagram, or other social media feed site) is tailored to you. There's no opportunity for other people to see the crazy thing that hit the front page and write blog posts critiquing it, give talks debunking the methodology, etc.
The algorithm can show your eyes, and your eyes alone, a steady stream of posts that are right on the edge of what you will find acceptable and engaging. The public at large won't know what is on your feed or even the feeds of powerful actors in society like politicians, billions, and celebrities.
It's like we are all accessing our own personal newspaper everyday, whose stories have been tailored to be as interesting as possible to us, but without any accountability or scrutiny besides what we bring to the table. You are welcome to see something that confirms your world view and either accept it as gospel or do the leg work to prove or disprove it.
What we are finding though is that when people are presented with things that confirm their beliefs, they tend to accept them. This isn't really earth-shattering or even surprising. The ramifications of having a population constantly consuming a stream of media tailored exactly for them is something are starting to understand.
Whenever the advertising industry get excited about a new medium or platform it should be a big red flag for the rest of us.
Speed, scale, pervasiveness, profitability, automated algorithmic microtargeting based on individual psychological profiles.
>The more shocking it is, the more eyeballs it attracts and the more add revenue it gets.
It's the same concept, just taken to a never before seen technological extreme.
This is what the public (government) should be responsible for. That way regulation is uniform (at least one can hope) across the industry.
But how are you going to legislate the engagement / polarisation tradeoff?
The bar to get banned from any online platform is pretty high. Just look at how long it took for the most prominent Twitter offender to get rejected.
You can ban all public forums you want, then people will just continue sharing BS on WhatsApp without you even noticing.
The problem isn’t Facebook but people. Governments should punish people, not outsource the policing to companies.
Governments shouldn't be in the business of deciding what is truth and what is "misinformation", and they certainly shouldn't be punishing people based on the outcomes of such determinations.
Otherwise, we'd have to ban the church and jail all the priests.
There are people who are deliberately spreading misinformation to make money. I think that we absolutely should be investigating and prosecuting such people. We simply lack the policing resources to effectively combat online fraud, online bullying, and online extortion. I think we need a fundamental restructuring of policing on our society to make this feasible.
The US constitution prevents bans on televangelists.
In the US, to achieve your goal you'll have to start off with a constitutional amendment repealing the first amendment.
There is a long history of the US courts abrogating free speech. "Shouting fire in a crowded theater" originated as a metaphor to justify preventing pacifists from distributing anti-draft pamphlets.
If the televangelist can be demonstrated in court to not believe the stuff he is telling people to make money, I see no reason why fraud charges shouldn't stick.
As for the punishment, it could be as simple as community service or losing access to public benefits for some time.
“Doctors are killing people with the vaccine?” Ok, find and pay for your own doctor, you’re not welcome in public hospitals.
What about people saying you should be wearing masks at the beginning of the pandemic, when the CDC and WHO were saying you didn’t need them?
Anything can be "classified" as anything if you draw your venn diagram circles big enough; that doesn't make it so. Furthermore, "terrorism" isn't itself a crime, the violence that usually comprises it is (such as murder, bombings, et c). A rather good thing that, or otherwise people like yourself would love to "classify" more things as something illegal, bypassing the actual authority in our society tasked with making new laws.
Spreading falsehoods is protected expression, at least in the US.
To engage with your point tho, who is to blame for the spread? the person who posts the disinfo or the programmers who decide the disinfo is good for engagement?
I think there's some truth to this, however someone could easily try to use the same argument for different topics. "People are awful, the problem isn't heroin, it's people." There's some truth to that too, but frankly, if we could magically remove 100% of heroin from the world things would be slightly better. In other words, it's clear that Facebook is making this worse, even if we carry most of the blame.
Humans can be awful, and they can be wonderful.
>Facebook is just a catalyzer just like Twitter is. Every platform has their own disgusting self-centered bubbles and that’s not because of the platform.
Pretending like system design has no effect on user behavior is just silly.
♪We know it's not funny♪
♪But we'll stop beating this dead horse♪
♪when it stops spitting out money♪
Edit: actually regarding formulaic-ness I'm thinking about Country Song. Repeat Stuff is about how pop songs just repeat the same lines now, as you might guess from the title
> At the end of the day it's people who are more easily misinformed, who do not question, do not look elsewhere, are not skeptical about how content is making them feel
What is your reasoning for declaring that this is a matter of "choice"?
It's also not fair to just blame the media. How about the politicians? Any major political leader dare to elaborate a balanced view at all?
It seems to be deeply rooted in partisan politics.
And in some countries definitely the political class as well.
Wonder what fb thought they were getting into here with MTR!
They rely on a hammer to catch what is objectionable or misinformation, but optimize for engagement maximally with everything else. However the hammer misses plenty of newly emerging inflammatory content (for example, anti-Rohingya content would look very different from anything prior), and this content gets maximally amplified because the overall newsfeed algorithms optimize for engagement.
An alternate approach would be to sacrifice engagement overall (perhaps only a little!) to reduce the very real and negative consequences of undesirable content that slips past the detection algorithms (which it always will). I suspect some of the fixes that were proposed, but never implemented, effectively did this. But they were shot down because, well, the top-line engagement numbers would dip.
Facebook could do this too and get the same results. More DAU more money.
This feels very much to me like a "something must be done!"
it has been said elsewhere: engagement = addiction
dial up the addiction, dial up the profit. "Regulators! mount up"
I find it interesting that this article does not even mention some of the other platforms that have also grown off this strategy, and have contributed to the societal problems named in the article.
I joke with my wife that you can innocently watch a YouTube press conference featuring our moderate Republican governor discussing COVID policy one day, and before you know it you're being recommended videos by rabid anti-vaxxers and QAnon deep state conspiracy theorists.
It's not just a Facebook problem or an "AI" problem, it's a platform problem which requires wider solutions than the broken internal approaches described in this article.
Mastodon has been doing very well by simply offering the Email of social media.
I disagree that Mastodon (or other federated alternatives) are a good example of this because these projects don't have a business model to support, so obviously they don't have to care about maximizing engagement.
I'm not saying maximizing engagement is a good idea, or that engagement-based business models are a good idea, I'm just saying that comparing a platform with an incentive to make money to one that does not doesn't answer anything.
There are other articles that tackle this topic for YouTube [1]. But as far as I know, we're still waiting on any expose that shows that YouTube had done extensive research into the issue and opted to do nothing to continue growth. That dubious honor seems to be reserved for Facebook at least for now.
[1] https://www.usatoday.com/story/tech/2021/02/12/youtube-chann...
> [In 2017,] YouTube limited recommendations on those videos and disabled features such as commenting and sharing. But it didn’t remove them. The company said the crackdown reduced views of supremacist videos by 80%.
'Hate speech' these days means 'speech I hate and want silenced'. Freedom of speech protections exist for this exact reason.
Why should a social media app engage in political censorship and selecting right vs. wrong opinion?