By Bret Swanson

AEI recently published two thoughtful essays on the online censorship question. The Cato Institute’s John Samples (also a member of Facebook’s Oversight Board) argues concerns about online censorship are understandable but misplaced and that nearly all proposed solutions are too aggressive. Libertarian law professor Richard Epstein thinks concentrated power over information is more serious and, despite preferring competitive innovation as a regulator of bad behavior, believes a common-carrier approach may be necessary. I’ll address Epstein’s essay later on. 

Samples’ argument is twofold: Critics who allege social
media firms suppress information faultily assume Big Tech’s censorship is
“systematic” and violates free speech rights. Samples covers a lot of ground,
and I agree with many of his warnings about regulatory overreach, but new information
may partially refute his arguments. 

On the day Samples’ report was published, a trove of emails and presentations from the Centers for
Disease Control and Prevention (CDC) became available via Freedom of
Information Act request. The materials suggest social media firms’ suppressive
efforts are in fact systematic and possibly violate the First Amendment due to
cooperation with the government.

The documents show that throughout the pandemic, the agency
and others in government corresponded frequently and held meetings with social
media employees. The CDC created “be on the lookout” reports for “misinformation”
about COVID-19 and asked the companies to incorporate this information into
content moderation policies. The CDC asked Google to help “push back against
the misinformation” and to “signal boost” the World Health Organization’s
“infodemic” efforts. The CDC asked the companies to attach “contextual
information” to user posts, which they did via warning labels. The companies
asked the government for information to include in fact-checking efforts and
for examples of “problematic content.” Facebook gave the CDC $15 million in
advertising credits. These examples are the tip of the iceberg. 

Systematic Misdiagnosis

Samples argues social media algorithms aren’t smart enough
to target viewpoints with any particularity or sophistication. Even if they
could, “Systematic viewpoint
discrimination by social media platforms is almost impossible to prove.”

Samples may be correct that algorithms are dumb, clunky, and
generate faulty suppressions that are often reversed on appeal. But he fails to
address social media companies’ concerted efforts to banish or down-rank
particular views and speakers through human decision-making and algorithm management. Samples
may not define these human efforts as “systematic,” but they are highly
organized, extensive, and focused. How, for example, did social media firms
suddenly become so irate about certain approaches to COVID-19? An algorithm
didn’t dream up bans for physicians contradicting official storylines. These
acts of censorship—and the humans behind them—create real-world disasters.

Perhaps Samples’ rejection of “systematic viewpoint”
discrimination is grounded in his assumption that most social media critics are
“conservatives” and that most haphazard censorship is partisan and obnoxious
but not consequential. He correctly notes that many well-known conservative
pundits enjoy huge audiences on social media, whereas if their censorship were
truly “systematic,” zero conservatives would be allowed.

But Samples misdiagnoses the problem as primarily political and
partisan. Many physicians and scientists who were censored by Twitter,
Facebook, YouTube, and LinkedIn during COVID-19 considered themselves liberal
or moderate Democrats. Twitter permanently banned Dr. Robert Malone, a pioneer
of mRNA vaccines and a Joe Biden donor and voter, for speaking skeptically
about his own technology. If conservatives stay in their lanes without violating
specific taboos, they are mostly allowed. But if anyone—regardless of political
orientation—crosses certain thresholds, watch out.

Modern First Amendment Questions

Samples and most scholars agree that, unlike government,
social media companies aren’t bound by constitutional free speech guarantees
and can moderate as they see fit. Samples acknowledges, however, that private
firms might violate the First Amendment if they act as vehicles for
government. If public officials coerce platforms to censor, content
moderation begins resembling state action and thus violates the First Amendment.
Samples largely dismisses the possibility this is actually happening. But the
new documents are the latest proof of not only coercion but hand-in-glove
cooperation to suppress speech. 

Remember, a previous stash of emails showed how Drs. Anthony Fauci and Francis Collins of the National Institutes of Health directed their staff to work with media to discredit eminent epidemiologists Jay Bhattacharya of Stanford, Sunetra Gupta of Oxford, and Martin Kulldorff, then at Harvard. YouTube and other platforms suppressed the release of the trio’s Great Barrington Declaration, which proposed an alternative policy strategy to COVID-19. Facebook allowed a spring 2020 video of Dr. Scott Atlas criticizing the COVID-19 lockdowns right up until he was appointed as a White House advisor, when they took it down. 

There seems to be lots of direct collusion between social media and the government. But even if the coercion and/or ideological partnerships were more indirect, we are walking uncomfortably close to some unsettled modern First Amendment questions.

My Response to Bret Swanson

By John Samples

Mr.
Swanson has me at a disadvantage. I have not reviewed the collection of emails
and presentations recently released. They require close attention for at least
two reasons. First, there is the question of government action. Did the federal
government, through the Centers for Disease Control (CDC), use Facebook to
censor political speech? Collaboration is not enough to make Facebook a de
facto state actor, and requesting labels on posts is not censorship. But the
documents may yet show that the government coerced Facebook into censoring
speech during the pandemic. It will be hard, I think, to discern “state
action” in Facebook’s moderation in these matters. But again, the
documents may say otherwise.

Second, Facebook may well contend that its employees sought only authoritative knowledge from the CDC about the pandemic to prevent “imminent physical harm” caused by falsehoods posted on the platform. Facebook is a company built on expertise; its employees might well have an overly strong and unwarranted faith in scientific authority. But why only government authorities? And was the “authoritative knowledge” used by Facebook correct in retrospect? If not, might a broader designation of authorities have fostered more trustworthy and thus legitimate content moderation regarding misinformation about COVID-19? Meta itself has sought policy advice about these questions from its Oversight Board. We welcome public comments as we review Meta’s policies.

I think proving “systematic viewpoint discrimination” would require data (a population or a valid sample) about moderated posts that we are unlikely to have. We do not know if Meta in general treats left and right differently in similar circumstances. We do know that citizens do not trust content moderation, and it is plausible that such mistrust could undermine confidence in elections and policymaking. On the other hand, Meta likely has a right to editorial judgment (as the Eleventh Circuit found with regard to the Florida social media law), and users have no right to equal or fair moderation by Meta. For better or worse, the Court has recognized that preventing declining confidence in political institutions is a “compelling public interest” that might justify imposing modest limits on Meta’s editorial judgment (i.e., content moderation). I have my doubts about that, but others will disagree.

Bret Swanson is a nonresident senior fellow at the American Enterprise Institute, where he focuses on the impact of technology on the US economy, telecommunications, and internet regulation. John Samples is a vice president at the Cato Institute and serves on Facebook’s Oversight Board.

The post Evidence of Public-Private Collusion Complicates Online Censorship Debate appeared first on American Enterprise Institute – AEI.