There has been much press about the New York University study, “False Accusation: The Unfounded Claim that Social Media Companies Censor Conservatives.” It is being touted by the media as establishing that any allegations of bias against conservatives is “disinformation,” the term used by authors Paul M. Barrett and J. Grant Sims of the NYU Stern Center for Business and Human Rights. That term of course is now used as a basis for flagging or censoring material. The problem is that the study is largely conclusory and, though buried in the study, acknowledges that it is not based on any real hard data and is therefore “inconclusive.”
Conservatives have hit the study as funded by Craig Newmark, a billionaire tech titan who is a major donor to Biden and Democratic candidates. Newmark has reportedly supported groups in favor of blacklisting and called on Big Tech to take action against “bad actors.” However, the authors did reveal Newmark’s funding and, while Barrett has also been attacked for a record allegedly supporting Big Tech, the credibility of the study would be judged on its actual support and analysis.
It is really two lines in the study that is the most serious problem in my view:
The question of whether social media companies harbor an anti-conservative bias can’t be answered conclusively because the data available to academic and civil society researchers aren’t sufficiently detailed. Existing periodic enforcement disclosures by Facebook, Twitter, and YouTube are helpful but not granular enough to allow for thorough analysis by outsiders.
So the study is not actually based on a review of individuals and groups censored by these companies because the companies refuse to release the data. Much of the rest of study seems strikingly conclusory and at points apologist. For example, the blocking of the Hunter Biden story before the election was a disgraceful decision for which the Twitter CEO Jack Dorsey later apologized. The authors largely dismiss the entire incident as a simple mistake:
“The Post/Biden imbroglio, in retrospect, seems like a case of reasonable decisions wrapped in mystifying processes. Facebook generally tries to stop posts from spreading if there are “signals” of falsehood. But as in the Post/Biden case, the company doesn’t disclose what those signals are, leaving onlookers to speculate. For its part, Twitter froze the Post/Biden story based on a rule against sharing hacked material. But under fire from conservatives, Twitter backed down, saying that from now on, it would ban hacked material only if it is directly shared by hackers or their accomplices.”
The problem is that the story itself clearly stated that the information did not come from hacked material but Hunter Biden’s own abandoned laptop. Moreover, major newspapers regularly publish information from insiders and whistleblowers like that contained in the New York Post story. Yet, the authors just dismiss that matter as confusion while citing conservative pressure on the company for its changing positions.
At every turn, the authors portray controversies as at best anecdotal or unsupported while acknowledging that it does not have the data to actually determine if there is a pattern of bias. It does not indicate any substantial insights or new information on the internal deliberations at companies like Twitter. Most of what is in the study is based on previously known and discussed facts. It simply presents those facts as a case in support of Big Tech on the allegation of bias.
The study also offers little on the striking contrast of companies banning figures like Donald Trump but allowing others to continue to justify violence or spout hate toward conservatives. We previously discussed professors who have called for killing Trump supporters. Others seem to justify violence. Rhode Island Professor Erik Loomis who writes for the site Lawyers, Guns, and Money and declared that he saw “nothing wrong” with the killing of a conservative protester. (A view defended by other academics). Other professors have simply called for all “Republicans to suffer.”
In the end, I would be interested in looking at the actual data and would keep an open mind on the results. Ironically, my problem is with the entire premise of the expanding system of censorship regardless of how it is implemented. Free speech is my bias. The authors openly support “moderation” which is the preferred term for censorship in our new world of speech controls. This is why I recently described myself as an Internet Originalist:
The alternative is “internet originalism” — no censorship. If social media companies returned to their original roles, there would be no slippery slope of political bias or opportunism; they would assume the same status as telephone companies. We do not need companies to protect us from harmful or “misleading” thoughts. The solution to bad speech is more speech, not approved speech.
Yet, this study is neither conclusive or particularly compelling. It read more like an extended, 20-page opinion editorial. It does seem itself to have a pronounced bias, particularly in declaring allegations of bias as “false” and “disinformation” while quietly noting that it cannot conclusively say whether there is bias.