Internet Intermediaries’ Long History of Censorship

Recent decisions by technology companies, especially “upstream” infrastructure technology companies, to drop neo-Nazis as customers have captured public attention—and for good reason. The content being blocked is vile and horrific, there is growing concern about hate groups across the country, and the nation is focused on issues of racism and protest.

But this is a dangerous moment for Internet expression and the power of private platforms that host much of the speech on the Internet. People cheering for companies who have censored content in recent weeks may soon find the same tactic used against causes they love. We must be careful about what we are asking these companies to do and carefully review the processes they use to do it. A look at previous examples that EFF has handled in the past 10+ years can help demonstrate why we are so concerned.

Complaints to “Upstream” Speech Intermediaries

This isn’t just a “slippery slope” fear about potential future harm. Complaints to various kinds of intermediaries have been occurring for over a decade. It’s clear that Internet technology companies—especially those further “upstream” like domain name registrars —are simply not equipped or competent to distinguish between good complaints and bad in the U.S. much less around the world. They also have no strong mechanisms for allowing due process or correcting mistakes. Instead they merely react to where the pressure is greatest or where their business interests lie.
Here are just a few cases EFF has handled or helped from the last decade where complaints went “upstream” to website hosts and DNS providers, impacting activist groups specifically. And this is not to mention the many times direct user platforms like Facebook and Twitter have censored content from artists, activists, and others.

• The U.S. Chamber of Commerce sent a complaint about a parody website created by activist group The Yes Men not merely to its hosting service, May First/People Link, but to that service’s upstream ISP, Hurricane Electric. When the hosting service May First/People Link resisted Hurricane Electric’s demands to remove the parody site, Hurricane Electric shut down MayFirst/PeopleLink’s connection entirely, temporarily taking offline hundreds of "innocent bystander" websites as collateral damage.
• Shell Oil sent a takedown notice to the ISP of activist group Oil Change International after it launched a campaign aimed at Shell’s sponsorship of New Orleans Jazz Fest. The ISP removed the site, abruptly halting the campaign.
• Unhappy with a single document published on the giant website Cryptome.org, Microsoft sent complaints to Cryptome’s domain name registrar and web hosting provider, Network Solutions. As a result, hosting provider Network Solutions pulled the plug on the entire Cryptome website — full of legal content — because Network Solutions was not technically capable of targeting and removing the single document. The site was not restored until wide outcry in the blogosphere forced Microsoft to retract its takedown request.
• Threats to the domain host of a critic of South African diamond conglomerate De Beers resulted in the temporary takedown of a New York Times spoof website that included, in part, a critical fake ad announcing that diamond purchases "will enable us to donate a prosthetic for an African whose hand was lost in diamond conflicts."
• Swiss bank Julius Baer pressured the domain name registrar for Wikileaks.org to lock the domain name after the organization posted documents demonstrating financial wrongdoing, and then obtained a court ruling confirming the censorship. In response to legal briefs by EFF and others objecting to this tactic, the district court dissolved the order, leading Julius Baer to dismiss its case.
• Media giant ABC sent a cease and desist letter on behalf of KSFO-AM radio in San Francisco to the webhost of the blog www.spockosbrain.com, after that site criticized the offensive and violent rhetoric on the radio station aimed at Congresswoman Nancy Pelosi and then-Senator Barak Obama.

You’ll notice that complainers in these cases are powerful corporations. That’s not a coincidence. Large companies have the time, money, and scary lawyers to pressure intermediaries to do their bidding—something smaller communities rarely have.

When Governments Get Involved

The story gets much more frightening when governments enter the conversation. All of the major technology companies publish transparency reports documenting the many efforts made by governments around the world to require the companies to take down their customer’s speech.[1]
China ties the domain name system to tracking systems and censorship. Russia-backed groups flag Ukrainian speech, Chinese groups flag Tibetan speech, Israeli groups flag Palestinian speech, just to name a few. Every state has some reason to try to bend the core intermediaries to their agenda, which is why EFF along with a number of international organizations created the Manila Principles to set out the basic rules for intermediaries to follow when responding to these governmental pressures. Those concerned about the position of the current U.S. government with regard to Black Lives Matter, Antifa groups, and similar left-leaning communities should take note: efforts to urge the current U.S. government to treat them as hate groups have already begun.

The Risks of Embracing Censorship

Will the Internet remain a place where small, marginalized voices get heard? For every tech CEO now worried about neo-Nazis there are hundreds of decisions made to silence voices that are made outside of public scrutiny with no transparency into decision-making or easy ways to get mistakes corrected. We understand the impulse to cheer any decisions to stand up against horrific speech, but if we embrace “upstream” intermediary censorship, it may very well come back to haunt us.