Signals Research

Can Improving Algorithms in Fact Improve News Quality?

Highly ranked stories, which might reward outlets financially, do not necessarily result in the production of quality journalism.

If higher-quality journalism is ranked higher in newsfeeds, will the overall quality of the news ecosystem also improve?

The NewsQ Initiative examines how news ranking and recommendation systems may promote journalistic quality. However, in this blog post I’m going to flip things around and will instead explore how such algorithms might be affecting the quality of journalistic production. 

Regardless of their actual content or their quality, news stories that appear at the top of our newsfeeds or go viral get more exposure, which translates into more engagement and more clicks, which in turn can generate revenue for news outlets. However, these highly ranked stories, which might reward outlets financially, do not necessarily result in the production of quality journalism.

Journalism Needs Engaged, Paying Readers

The interconnection between newsfeed visibility, reader engagement, and the quality of stories has profound implications for a news industry struggling to generate revenues. As advertising has declined, a growing number of news organizations, from national outlets like The New York Times to niche digital-only products like The Information, have increasingly relied on subscriptions, donations, grants, ticketed events, membership offers, and other alternative sources of revenue to run their operations. However, despite these emerging revenue streams, many news organizations, particularly at the local level, still rely on income from “clicks” on advertising.

Whether they seek to attract advertisers or foster alternative sources of revenue, publishers must attract readers. Writing in 2015 for the BBC when this trend was emerging, Ben Frampton described the growth of “clickbait”: 

Publishers increasingly use it for simple economics; the more clicks you get, the more people on your site, the more you can charge for advertising. A report by the Columbia Journalism Review highlighted the case of online magazine Slant, which pays writers $100 per month, plus $5 for every 500 clicks on their stories. Slant is far from unique in this respect and this business model is becoming increasingly common…

Over the past decade, more clicks to publishers began to be driven by the algorithmic ranking and recommendation systems of social media platforms and news aggregators. In turn, editors and reporters began to assign and produce content that would hopefully rank higher and rise to the top. At many publications, editors struggling with small budgets and even smaller revenue streams have assigned writers to rewrite and repackage reporting produced elsewhere, or report on topics that in the past may not have made the pages of a newspaper. For example, as of March 31, 2020, the top performing story in terms of traffic for The Miami Herald according to their reporter Sarah Blaskey is an article she authored from February 3, 2020 about President Trump fidgeting and “pretend-conducting the band” during the national anthem at the Superbowl. However a consequential and deep investigation by the Herald into the crimes and plea deal of Jeffrey Epstein, which ultimately led to renewed charges against the financier accused of multiple sexual abuses, did not enjoy the same amount of reader engagement.

Algorithms Reward Sensationalism, Rather than Quality

This is why the product managers and engineers who work on the ranking and recommendation algorithms that serve up the articles we see in our newsfeeds have a much larger impact. By ranking some articles above others — a role similar to editors at newspapers who lay out a front page — whoever designs, optimizes and maintains newsfeed algorithms has an impact upon which journalism is monetized and compensated. In the context of a revenue-hungry media industry, this means what kinds of journalism are more likely to get written and produced.

But algorithmic mechanisms can be opaque to those without engineering and data science backgrounds. For the better part of the past decade, journalists have been guessing what signals in their stories will influence ranking by algorithms. Author Nicolás Medina Mora who covered crime and breaking news for BuzzFeed in New York for three years once shared what this feels like in practice. Recalling a story he wrote about pair of undocumented immigrant siblings who migrated to Connecticut in 2014, he shared:

I intended to repackage the boy’s trauma into a digestible narrative I hoped would capture the attention of some hundred thousand internet users, who would then surrender valuable information about themselves to one or another technology baron, who would then reward the website for which I worked with a better starting position in the algorithmic rat race, which would allow the website’s owners to convince a handful of investors to keep funding the company, which in turn would allow my editors to pay me a salary.

Medina Mora’s representation of his publisher’s revenue model is not quite accurate, given that it misses the effect of automated programmatic-ad marketplaces. Nevertheless, his essay does capture the sense among journalists that they must sensationalize, and that they must grab readers attention before they scroll past to be rewarded by the algorithms controlling the presentation and distribution of their stories.

We can see the spread and impact of the tabloidization of our news by simply looking at the stories that received the most engagement on social media. Consider that 2019 was full of important stories, including:

  • A government shut down in the U.S. (link)
  • The Easter terrorist bombings in Sri Lanka (link)
  • A fire that decimated much of the Notre Dame cathedral in Paris (link)
  • Racist massacres in both Texas (link) and New Zealand (link)
  • A historic candidate field in the U.S. presidential primaries (link)
  • An impeachment investigation and trial that dominated the news cycle for much of the last third of the year

Yet, according to NewsWhip, a company that measures engagement metrics of news publishers on social media, only one of the year’s 25 most engaged stories was about one of the above topics — the Notre Dame fire ranked at #23 on the list.

Readers Actually Want to Read High-Quality Journalism

Yet there is evidence that many readers want more from journalists than what they are getting in their feeds. According to a study conducted by Pew Research Center published in 2020, 73 percent of American adults think that “it’s important for journalists to act as watchdogs over elected officials.” However, what the surveyed adults are seeing have led them to doubt that journalists are fulfilling this role well.

As I write this during the COVID-19 pandemic, people are turning to the news to find what is going on in their cities and states, what precautionary measures to take to protect against the disease, and what symptoms to look out for. People are looking for journalism that provides them with critical accurate information in a time of crisis.

Perhaps if higher quality work is rewarded in newsfeeds and other algorithms, and perhaps if this work gets ranked higher and therefore draws more eyeballs, then editors might assign reporters to work on stories that matter more overall. And then, in this way, we might see an overall improvement in the quality of our news ecosystem.

It’s also worth noting that, as a recent example, increased traffic due to coronavirus reporting has not directly helped revenue, as advertisers often will not have their brands associated with negative themes.

Of course, there is a big assumption that when people actually click on such higher quality stories appearing at the top of their feed, it will thereby drive revenue to those outlets through ads and subscriptions. But given what critical stories we might be losing, it seems a hypothesis worth trying out.