Eight years ago, Mark Zuckerberg made a lofty announcement: Facebook would use its resources to get billions of people online for the first time. One part of that effort was Free Basics, a partnership with mobile service providers that allows users to access a curated set of websites, including Facebook, at no cost. Zuckerberg framed the program as part of a noble philanthropic mission, but activists and researchers decried it as “digital colonialism” and said it violated net neutrality, the principle that all web traffic should be treated equally. 

“Gradually, we realized that [what] Facebook was looking for was not necessarily to connect the world, but to see how they can capture more of the market in the name of connectivity, and to connect people on their own platform,” said Osama Manzar, founder of the nonprofit Digital Empowerment Foundation in New Delhi, who was contacted by Facebook when it tried to bring Free Basics to India. 

The backlash eventually led the Indian government to block Free Basics — a decision The Guardian described at the time as Facebook’s “biggest setback” — but the program has quietly continued operating in dozens of other countries across Africa, Asia, and South America, helping bring over 100 million people online. Now, a new study finds that another Facebook connectivity project has adopted many of the same qualities that enraged Free Basics’s critics. The research suggests the social network is still positioning its products as a gateway to the internet in order to capture new users in developing countries, though this time without any grandstanding from its CEO. 

The new study — published last month as part of the ACM Conference on Human Factors in Computing Systems — looked at the rollout of Discover, an app and mobile platform similar to Free Basics that Facebook launched last year to little fanfare. Discover lets users scroll through a text-only version of any website for free, up to a certain daily data cap set by their mobile provider. But unlike Free Basics, Discover supposedly allows people to view the entire internet, instead of only a handful of preselected resources. Users don’t need a Facebook account to get started, and the company said it doesn’t collect user data for advertising.

The research was conducted in July and August last year by scholars at the University of California, Irvine and the University of the Philippines, and focused specifically on how Discover functions in the Philippines — a country with high levels of internet usage and where Facebook is already enormously popular. Facebook told Rest of World that Discover has now replaced Free Basics there entirely. 

To save data, Discover routes all traffic through a proxy server, which strips features like video and audio streaming, as well as some images. It essentially gives users free access to a pared-down version of any website. But the researchers found that when they accessed Facebook through Discover, it wasn’t redacted at all — and just 4% of images were removed from Instagram, compared with more than 65% of images on other popular sites like YouTube and e-commerce platform Shopee. In other words, the study found that Discover rendered Facebook’s own services far more functional than those of its own competitors. 

Facebook via ACM Conference on Human Factors in Computing Systems

“Free Basics was initially critiqued for violating net neutrality by creating this walled garden, [where] Facebook gets to approve what’s in the free internet,” said Lucy Pei, a PhD student at the University of California, Irvine, who co-authored the study along with PhD student Benedict Salazar Olgado and informatics professor Roderic Crooks. Discover was supposed to fix that problem by letting users surf the open web, but stripping certain types of media is still “a very top-down approach to moderating what kind of experience you can have on the internet,” Pei said.

In response to inquiries from Rest of World about the study’s findings, a Facebook spokesperson said the company hadn’t intended to favor its own sites on Discover, blaming the issue on a  “technical error” that had since been corrected. “As this report identified, there was a proxy error in the Discover app that resulted in inconsistent image loading across many websites that load images involving HTTP redirection,” the spokesperson said in an emailed statement. “​This was a technical error that has since been resolved and all websites are being proxied the same as intended.”

Facebook’s frequently asked questions page for Discover said that it “does not select which sites are available, and the technical limitations are applied equally to all sites, including Facebook.” The social network told Rest of World that it redacts sites using a highly technical process, and that due to sheer volume, it’s impossible to proactively check that each one will be rendered correctly in Discover.

Toussaint Nothias, the research director of Stanford University’s Digital Civil Society Lab who previously studied Free Basics in Africa, said it’s disingenuous for Facebook to claim it’s treating all sites equally, especially if it can’t predict how they will be displayed. “Facebook has been doing a range of zero rating offers for a very long time,” he explained, referring to the practice of offering mobile data at low or no cost. He said the social network “needs to acknowledge that removing audio and video is not just a technological choice; it’s a political choice, it’s an economic choice.”

“Removing audio and video is not just a technological choice; it’s a political choice, it’s an economic choice.”

The researchers found that some popular websites, such as delivery site Foodpanda and gaming platform Roblox, were redacted to the point of being non-functional, and it was also impossible to make purchases on many e-commerce platforms. In some instances, users couldn’t create or log into accounts on different sites because Facebook’s image redaction prohibited them from completing the required CAPTCHA test.

Olgado calls the redaction systems built into Discover a type of “form moderation,” meaning that it alters how a site looks, rather than the specific content it hosts. He said that much of the discussion about content moderation has focused on individual posts that platforms leave up or take down, “but we also need to look at form moderation. Content and form go hand in hand. If you change the form of content, the meaning changes.”

Mong Palatino, a former member of the Philippines House of Representatives who has studied Free Basics in the Philippines for the nonprofit advocacy group Global Voices, warned that because Discover displays few images, it could inadvertently help spread misinformation. “This makes it more difficult for readers to process information, and means users will react based on text or headline alone,” he explained. Palatino suggested Facebook should instead direct its efforts in the Philippines toward curbing content that promotes hate and violence. 

Facebook via ACM Conference on Human Factors in Computing Systems

Facebook initially said Discover was being tested in Peru, as well as the Philippines, Thailand, and Iraq. But Nothias told Rest of World that tests using a VPN last month showed it was also available in Chile, Malawi, and Myanmar. (Nothias wasn’t part of the new study.)

Facebook declined to list the countries where the program has ever operated, but said Discover is currently available in Peru, Chile, Thailand, the Philippines, and Iraq, and noted that all of its services are currently blocked in Myanmar (a spokesperson didn’t clarify whether Discover operated there prior to the military coup in February).

The researchers noted that Discover also used data from the web analytics company SimilarWeb to suggest a list of popular websites to users. They found the rankings overwhelmingly listed sites from U.S.-based entities, making them more readily accessible than local resources in the Philippines. (Facebook said the ranking page has since been replaced with a screen highlighting information about Covid-19.)

“Anybody who is playing a role in content creation or production, their interest in connectivity is always [raising] a question about creating a monopolistic situation,” said Manzar from the Digital Empowerment Foundation in India. “Where is the public good in Free Basics, in Discover? If you really wanted to solve the connectivity problem, you could have said: Any entrepreneur who wants to do last-mile connectivity, we are going to supply you with the equipment.”