Yahoo mixes editors and algorithms so readers don't only see what they already agree with
Top stories everyone sees sit next to stories picked for you, plus tools to compare viewpoints.
Newsroom Robots · Kat Downs Mulder: Inside Yahoo’s AI Strategy for the Future of NewsThe question put to Kat Downs Mulder, the senior vice president and general manager of Yahoo News, was about a familiar tension. Personalization lets people see only what they already care about, which can be bad for democracy, yet it can also bring in people who would otherwise avoid news entirely.
Her answer was a distinction she said guides Yahoo.
“We really highlight the need to know and the want to know,” Mulder said.
By that, she explained, Yahoo does not leave the front page and the news app to software alone. Editors recommend top stories from a variety of sources, and that selection sits next to stories chosen by software for a reader’s interests, favorite publishers, and favorite topics.
In Mulder’s view, that combination keeps a shared layer of news in front of everyone. However much the algorithm tunes itself to the individual, readers still see what Mulder called “the world is thinking about and talking about,” presented in a fact-based way that describes what happened rather than offering an opinion. That shared layer, she argued, is what keeps personalization from narrowing into a filter bubble, where software shows people mostly what they already like and they miss other views and common news.
Mulder pointed to reader appetite as the reason to preserve the mix. She acknowledged tribalism and loyalty, but said a lot of people do not want to be limited to a single outlook and find it valuable to see many sources on one topic.
For Yahoo, the next job is helping readers make sense of that variety. The company, she said, is building features that sort reliable accounts from the clutter and place different perspectives side by side, so readers have the toolkit to compare, synthesize, and form their own opinions rather than be told what to think.
Questions people ask
How do the two kinds of stories show up for a reader?
On Yahoo.com and in the news app, editor selections and personalized selections appear together, Mulder said. The personalized part reflects interests and habits, while the editor part reflects what editors judge most important. That editor layer is meant to keep a shared set of news in view even as the rest of the feed becomes more personal.
Do all Yahoo readers want that mix of views?
No. Mulder acknowledged tribalism and loyalty among some news consumers. But she said a lot of people do not want to be pigeonholed into one perspective, and those readers value seeing a wide variety of sources on a topic.
What is Mulder's larger worry about algorithms?
She said the past decade of algorithmic products has shown how things can go awry, with public discussion becoming "chaotic and toxic in a lot of places." Her argument is that quality information and quality conversation need deliberate support through product design and the right engagement between people and algorithms.