Cutting off the advertising revenue of websites that spread hoaxes is the best way to end misinformation

The most effective way to combat misinformation is to try to ensure that the websites that spread it receive less advertising revenue. This is the conclusion reached by a study published recently in the scientific journal Nature, after researchers analyzed 1,276 misinformation websites and 4,209 legitimate websites between 2019 and 2021, as well as the behavior of 42,595 unique advertisers, who placed more than 9.5 million ads in that period.

According to the authors of the study, allowing consumers to know where companies invest their advertising budgets can result in a considerable cut to the funds being given to websites that spread lies. There are brands that don’t want to be associated with this type of content, but often finance it due to ignorance: they use automated digital advertising tools.

The problem is that these spaces, which are auctioned off — in processes organized by the so-called “ad exchanges” — not only include media outlets of proven reputation, but also websites that engage in misinformation. “Some advertisers aren’t aware of where their money is going,” writes Wajeehaa Ahmad in the article, which she co-published with her colleagues. She’s affiliated with the Department of Management Science and Engineering at Stanford University.

Following the money trail

The study confirms, firstly, that online misinformation is financed mainly by advertising revenue and, secondly, that the automation of the allocation of advertising space amplifies the financing of misinformation. Next, it examines how the financing of these websites affects advertisers and, finally, proposes measures to reduce this investment. According to NewsGuard, this sum isn’t minor: for every $2.16 of digital advertising revenue spent in legitimate media, American advertisers spend $1.00 on websites that misinform users.

The approach of this work is innovative because, until now, most interventions to try to counteract the proliferation of misinformation have focused on the consumer side: developing fact-checking websites, labeling responsible content, asking readers not to spread content they don’t trust, etc. The goal of Amad and his colleagues, however, was to take action on the supply side.

To identify the websites where misinformation is prevalent, the researchers turned to two main sources: NewsGuard — a company that rates the reliability of the information contained in 95% of the websites available in the five countries in which it operates — and the Global Disinformation Index (GDI), a not-for-profit organization based in the United Kingdom. The study considers a website to be engaging in misinformation when NewsGuard has repeatedly classified it as such between 2019 and 2021.

The first thing was to try to understand if companies directly place advertising on websites that misinform, or if this occurs automatically through digital advertising allocation tools. To achieve this, the team built a large database combining information from websites that publish hoaxes, along with the advertising activity of the media over a period of three years. At the same time, they surveyed company executives, asking them if they were aware that their organizations were supporting misinformation or disinformation. They found that, in many cases, they were not aware.

The next thing was to confirm if, indeed, consumers even care that the brands they buy support problematic content. To measure the level of discontent, the authors conducted a survey with a random sample of the American population. The objective was to see how consumption varies when consumers become aware that certain companies support disinformation websites. They also measured the response based on the intensity with which the companies in question spend money to finance misinformation.

According to the authors, consumers tend to stop betting on products from companies that support misinformation. This behavior persists, furthermore, despite the fact that the consumer is warned that investment in these sites often occurs without the knowledge of management. “The data show that consumer backlash was particularly strong for women and politically left-leaning consumers,” the article highlights.

Ahmad and his colleagues believe there’s a path towards reducing the funding of misinformation through two “scalable and low-cost” interventions. First, by improving transparency, so advertisers know where their ads appear and can subsequently reduce advertising on websites that misinform. This would especially benefit firms that were previously unaware that their ads appeared on these sites. And, second, this process would be faster if there were specific platforms dedicated to it.

The Check My Ads Institute — an American group — is one such platform. Founders Nandini Jammi and Claire Atkins are in charge of notifying large companies and ad exchanges — the platforms that organize the auction of advertising space — which websites are spreading hoaxes. Check My Ads made a name for itself following a great achievement by Jammi: she managed to deplete advertising revenue from the far-right website Breitbart News, owned by Steve Bannon, Trump’s star advisor in his presidential race and later his chief White House strategist, until his fall from grace in 2017. That same year, Check My Ads managed to get 90% of Breitbart’s anticipated annual revenue — around $8 million — removed by companies that were previously advertising with them. “This was possible because 31 of the 34 ad exchanges withdrew, and, with them, 4,000 advertisers,” Atkins tells EL PAÍS. “But we believe there’s still a lot of work to do.”

How to combat misinformation?

“Our results suggest that both simple information disclosures and comparative company rankings can reduce consumer demand away from companies advertising on misinformation websites,” the study notes. “Advertising companies may wish to account for consumer preferences in placing their advertising across various online outlets and exercise caution while incorporating automation in their business processes via digital advertising platforms.”

“Companies could use lists of misinformation outlets provided by independent third-party organizations — such as NewsGuard and the GDI — to limit advertising budgets being spent on misinformation outlets.” This would be especially convenient, the authors note, for companies whose target audience is women or left-wing individuals, the groups most sensitive to advertising investment in misinformation.

The researchers also recommend that portals where advertising space is automatically auctioned off specify whether they include misinformation websites among their destinations. Another measure could be to regularly publish rankings of companies that support misinformation in order of intensity, in the same way that Google Flights shows flights in order of carbon emissions.

The study, however, doesn’t assess how the generation of junk content using generative artificial intelligence (AI) tools will affect this equation. This factor will presumably facilitate the dissemination of misinformation and increase its volume.

The real effect of hoaxes and cancellation

Nature has also published two more scientific articles related to misinformation. In one of them, the authors debunk some myths about this content: that people’s average exposure to hoaxes is high, that algorithms are largely responsible for this exposure and that social media is the main cause of polarization. Rather, this research documents the opposite.

The authors conclude that there’s a relatively low exposure to hoaxes and lies among users, which is also highly-concentrated in social groups with a strong motivation to consume this type of content. Consequently, they recommend holding the platforms or websites responsible for the content they disseminate and ask for more transparency about what’s being published.

Another investigation corroborates that removing former U.S. president Donald Trump and 70,000 other toxic accounts from Twitter — today known as X — after the assault on the Capitol on January 6, 2021, significantly reduced the spread of misinformation across social media.

The researchers analyzed a panel of about 600,000 Twitter accounts that were active during the 2020 U.S. election cycle. They found that 1,361 of them — 0.25% — were canceled between January 8 and 12. However, this small subset of users was responsible for 24.13% of all misinformation shared on the panel.

The conclusion is that social media platforms may have the ability to partially control the spread of hoaxes, although the authors highlight that the data comes from a single country during a very specific period.

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition