Nieman Foundation at Harvard
HOME
          
LATEST STORY
From shrimp Jesus to fake self-portraits, AI-generated images have become the latest form of social media spam
ABOUT                    SUBSCRIBE
Sept. 28, 2017, 11:12 a.m.
Aggregation & Discovery

All the news that’s fit for you: The New York Times is experimenting with personalization to find new ways to expose readers to stories

“Instead of thinking about having stories compete for limited space on the homepage, we’re trying to shift the conversation to a different understanding of our distribution.”

In the future, “All The News That’s Fit to Print” will depend a lot more on who’s reading.

One of The New York Times’ greatest offerings to readers is its editorial judgment; whether it’s Page 1, the Times app, or the homepage of NYTimes.com, the package of stories selected by Times editors comes with an implicit statement: This is the most important stuff to know right now.

But over the past few months, The New York Times has been conducting a series of small experiments aimed at customizing that story selection to the individual reader, based on a variety of signals — such as past user behavior, location, or time. Just as your Google search results aren’t the same as mine, your New York Times might be subtly different from your neighbor’s.

“We produce a lot of great journalism; we just don’t want to waste it,” said Caroline Que, editorial director of the Times’ news desk. “Instead of thinking about having stories compete for limited space on the homepage, we’re trying to shift the conversation to a different understanding of our distribution.”

These tests are subtle, for the most part. In one ongoing experiment, the Times has tested personalizing part of the homepage based on where readers are located: Visitors in New York City or the surrounding area see a section with a handful of recent stories about the city; those who aren’t, don’t.

In the Smarter Living section of the homepage, which is home to lighter, more service-oriented stories, the Times has experimented with using personalization to expose readers to new evergreen content. Many readers who visit the page see editor-selected stories, but others see stories based on clusters of topics both broad and narrow such as health, weddings, and air travel, based on their reading behavior. “All of this comes from the recognition that we have this great diversity of interest among our readers and that broadcast model where everyone sees the exact same thing all the time is not the only way to expose people to our work,” Que said.

The Times has also considered tweaking the homepage based on the last time a reader visits the site. If, for example, the newspaper publishes a particularly enterprising story on a Monday, but a reader doesn’t visit the site until later in the week — long after that story has left the homepage — the Times could personalize the user’s homepage to showcase that story, particularly if it’s about a topic that that reader has repeatedly demonstrated interest in. This kind of personalization could also work in tandem with custom-targeted push notifications, which the Times is also interested in. This sort of thing may be table stakes for most tech companies today, but for news organizations that have so far lacked the rich user data or expertise needed, it’s new and uncharted territory.

The Times has also considered ways of using geotargeting to decide which readers see specific stories. Que pointed to coverage of events like this summer’s solar eclipse, some of which was only relevant to users who were actually in the eclipse’s path. There are also some smaller, more granular tweaks in the works. Readers in the U.K., for example, could see stories with units of yards and Fahrenheit swapped out for kilometers and Celsius. The same goes for date formatting.

For the Times, this kind of personalization, while still relatively new, represents a new, powerful way to get “our most ambitious, innovative journalism in front of the most relevant audience as we can,” Que said. “We want to make the experience as graceful and obstacle-free as we can for all of our readers.”

Many of these experiments were teased in a March column by then-public editor Liz Spayd, who acknowledged the promise of personalization while also cautioning against potential drawbacks. There are some understandable privacy concerns, for one, as well as the larger questions about the loss of editorial judgment and the risk that personalization will make it harder to expose readers to stories about unfamiliar topics or opinion columns with differing views.

And some Times readers will push back against any effort at personalization. Back in 2014, then-public editor Margaret Sullivan fielded a complaint from a reader about an earlier Times effort at personalized article recommendations: “I find this offensive and ridiculous, since I feel competent to choose articles to read on my own.”

Que said that the Times appreciates the concerns but is committed to honing the balance between the traditional shared experience of the newspaper and a more bespoke model that readers have come to expect from their news online. She noted that readers have always taken their own personalized routes through the print newspaper, guided by their own interests as much as editorial decisions. “We always want to have that shared experience where everyone can pick up the same newspaper and see on the front page the stories the senior editors think are the most important,” she said. “But you can also open the paper and grab the sports or arts section first if you want to,” she said. “It’s all about that balance of presenting that shared experience but also meeting readers where they are in their lives.”

Photo of the unique snowflakes all Times readers are by Danilo Urbina used under a Creative Commons license.

POSTED     Sept. 28, 2017, 11:12 a.m.
SEE MORE ON Aggregation & Discovery
Show tags
 
Join the 60,000 who get the freshest future-of-journalism news in our daily email.
From shrimp Jesus to fake self-portraits, AI-generated images have become the latest form of social media spam
Within days of visiting the pages — and without commenting on, liking, or following any of the material — Facebook’s algorithm recommended reams of other AI-generated content.
What journalists and independent creators can learn from each other
“The question is not about the topics but how you approach the topics.”
Deepfake detection improves when using algorithms that are more aware of demographic diversity
“Our research addresses deepfake detection algorithms’ fairness, rather than just attempting to balance the data. It offers a new approach to algorithm design that considers demographic fairness as a core aspect.”