An English-Language, Algorithmically-Personalized News Aggregator, Based In China — What Could Possibly Go Wrong?

from the no-social-graph-required dept

Techdirt has been exploring the important questions raised by so-called “fake news” for some time. A new player in the field of news aggregation brings with it some novel issues. It’s called TopBuzz, and it comes from the Chinese company Toutiao, whose rapid rise is placing it alongside the country’s more familiar “BAT” Internet giants — Baidu, Alibaba and Tencent. It’s currently expanding its portfolio in the West: recently it bought the popular social video app for about $800 million:

Toutiao aggregates news and videos from hundreds of media outlets and has become one of the world’s largest news services in the span of five years. Its parent company [Bytedance] was valued at more than $20 billion, according to a person familiar with the matter, on par with Elon Musk’s SpaceX. Started by Zhang Yiming, it’s on track to pull in about $2.5 billion in revenue this year, largely from advertising.

An in-depth analysis of the company on Ycombinator’s site explains what makes this aggregator so successful, and why it’s unlike other social networks offering customized newsfeeds based on what your friends are reading:

Toutiao, one of the flagship products of Bytedance, may be the largest app you?ve never heard of — it’s like every news feed you read, YouTube, and TechMeme in one. Over 120M people in China use it each day. Yet what’s most interesting about Toutiao isn’t that people consume such varied content all in one place… it’s how Toutiao serves it up. Without any explicit user inputs, social graph, or product purchase history to rely on, Toutiao offers a personalized, high quality-content feed for each user that is powered by machine and deep learning algorithms.

However, as people are coming to appreciate, over-dependence on algorithmic personalization can lead to a rapid proliferation of “fake news” stories. A post about TopBuzz on the Technode site suggests this could be a problem for the Chinese service:

What’s been my experience? Well, simply put, it’s been a consistent and reliable multi-course meal of just about every variety of fake news.

The post goes on to list some of the choice stories that TopBuzz’s AI thought were worth serving up:

Roy Moore Sweeps Alabama Election to Win Senate Seat

Yoko Ono: “I Had An Affair With Hillary Clinton in the ’70s”

John McCain’s Legacy is DEMOLISHED Overnight As Alarming Scandals Leak

Julia Roberts Claims ‘Michelle Obama Isn’t Fit To Clean Melania’s Toilet’

The post notes that Bytedance is aware of the problem of blatantly false stories in its feeds, and the company claims to be using both its artificial intelligence tools as well as user reports to weed them out. It says that “when the system identifies any fake content that has been posted on its platform, it will notify all who have read it that they had read something fake.” But:

this is far from my experience with TopBuzz. Although I receive news that is verifiably fake on a near-daily basis, often in the form of push notifications, I have never once received a notification from the app informing me that Roy Moore is in fact not the new junior senator from Alabama, or that Hillary Clinton was actually not Yoko Ono’s sidepiece when she was married to John Lennon.

The use of highly-automated systems, running on server farms in China, represents new challenges beyond those encountered so far with Facebook and similar social media, where context and curation are being used to an increasing degree to mitigate the potential harm of algorithmic newsfeeds. The fact that a service like TopBuzz is provided by systems outside the control of the US or other Western jurisdictions poses additional problems. As deep-pocketed Chinese Internet companies seek to expand outside their home markets, bringing with them their own approaches and legal frameworks, we can expect these kind of issues to become increasingly thorny. We are also likely to see those same services begin to wrestle with some of the same problems currently being tackled in the West.

Follow me @glynmoody on Twitter or, and +glynmoody on Google+

Filed Under: , , , , , ,
Companies: bytedance

Rate this comment as insightful
Rate this comment as funny
You have rated this comment as insightful
You have rated this comment as funny
Flag this comment as abusive/trolling/spam
You have flagged this comment
The first word has already been claimed
The last word has already been claimed
Insightful Lightbulb icon Funny Laughing icon Abusive/trolling/spam Flag icon Insightful badge Lightbulb icon Funny badge Laughing icon Comments icon

Comments on “An English-Language, Algorithmically-Personalized News Aggregator, Based In China — What Could Possibly Go Wrong?”

Subscribe: RSS Leave a comment
PaulT (profile) says:

“Roy Moore Sweeps Alabama Election to Win Senate Seat”

Wow. There’s “fake news”, stuff that might be plausible but someone actually made up. There’s “fake news” which is sensationalist clickbait that nobody in their right mind should think is actually true. Then, there’s actually stating the polar opposite of reality.

Being someone who tries to carefully consider sources and remove known partisan/fake sources from my daily intake, I’m not sure, but – is it actually common for “fake news” to mirror reality like this? No wonder some people are so confused and angry when they stagger outside their echo chambers, if they’re believing the complete opposite of what’s happening in the real world.

Anonymous Coward says:

Re: Re:

I think of “fake news” as being mostly rumors where they say that someone said something about someone else. The last three examples would fall into that category if the scandal is someone claiming John McCain did something.

With Roy Moore winning a Senate seat, that just seams like such an easy thing to check that I wouldn’t believe someone lied about it if the article didn’t explicitly say it was “fake news”.

ECA (profile) says:

Re: Re:

Fake news?
Best lie is 1/2 truth and 1/2 lie/fantasy..

Words to AVOID..may have, Could have been, MIGHT have..Anything that relies on ALTERNATIVE INFORMATION..

” A person was Shot at ????” is news..”It might have been his ex-wife” IS NOT NEWS, it is opinion/speculation/estimation/Fantasy/and many other words.

Knowing and understanding abit about the internet, Every time a computer is HIT. and hearing RUSSIA/CHINA/N. KOREA DID IT…makes me PUKE.. There are ways to track the net, but they are NOT EASY..

REALITY SUCKS, dont make it worse or sugar coat it..

PaulT (profile) says:

“The last three examples would fall into that category if the scandal is someone claiming John McCain did something.”

That’s the headline that could really be anything. It could be referring to scandals around McCain himself, or it could just be a sensationalised write-up of something tangentially related to something he worked on. You can’t tell without following the headline, clickbait at its purest.

The other two are more obvious – specific claims about what a celebrity has supposedly said about a political figure are almost always made up or taken out of context.

“that just seams like such an easy thing to check “

The primary reason why “fake news” is so problematic is that a lot of people really don’t check. They’ll nestle in their echo chamber, blissfully unaware that they’re being lied to. Unless their same feed/group/whatever feeds them the correct result later on, they’ll not even think to check if the original claim was correct.

Wendy Cockcroft (user link) says:

Re: Re:

In my experience, they don’t give a rat’s bum about whether they’ve been lied to or not as long as those lies appeal to their biases. I’m seeing this the way a scientist sees mould in a petri dish: fascinating but yuck.

I mean, we try to teach our kids about resisting peer pressure, but hypocritically give in to it ourselves. What sort of message does that send them?

And ultimately, that is what the echo chamber thing is: peer pressure.

PaulT (profile) says:

Re: Re: Re:

“In my experience, they don’t give a rat’s bum about whether they’ve been lied to or not as long as those lies appeal to their biases.”

My experience is that they do care about being lied to, but that if you try to introducing new evidence to prove that they’re wrong, they’ll just attack the messenger. What they won’t do is re-evaluate their knowledge and conclude that their first conclusion was flawed. The person introducing new information has to be wrong or lying, never the first person they believed.

It’s less “peer pressure” from what I see, but more of a cult/religious thing. They hate being lied to as much as anybody else, but it undermines their entire worldview to think that the sources they trust by default might be lying. So, they turn their ire again anyone trying to tell them that, no matter how true it is.

John85851 (profile) says:

Feature or bug?

It sounds like the AI is learning all right, but it’s learning that more people click on fake news and entertainment than on real news. After all, which is more popular- the crazy story that everyone knows is false or the true, correct version of the story?

Like PaulT said above, how many people won’t bother verifying the information? They’ll scroll through the stories or click to read it, then move on to the next thing in their feed.

Add Your Comment

Your email address will not be published. Required fields are marked *

Have a Techdirt Account? Sign in now. Want one? Register here

Comment Options:

Make this the or (get credits or sign in to see balance) what's this?

What's this?

Techdirt community members with Techdirt Credits can spotlight a comment as either the "First Word" or "Last Word" on a particular comment thread. Credits can be purchased at the Techdirt Insider Shop »

Follow Techdirt

Techdirt Daily Newsletter

Techdirt Deals
Techdirt Insider Discord
The latest chatter on the Techdirt Insider Discord channel...