Mind Matters News and Analysis on Natural and Artificial Intelligence
nettoyage de graffiti
Worker in blue uniform cleaning graffiti

Facebook’s Secret Censorship Rules Expose a Key Problem

Most moderators are not skilled and have only a few seconds to decide on a post

 A serious concern in recent years has been the way in which much communication worldwide is now in the hands of a few recently organized social media companies. These companies are not governments so they cannot enforce criminal penalties for infringing their rules. But they are not accountable to an electorate either and they can be pressured by governments.

France has proposed banning so-called fake news during the country’s future elections, while in Germany, new hate speech rules impose fines of up to €50 million on social media companies that don’t delete harmful content within 24 hours of being notified. Mark Scott, “Welcome to new era of global digital censorship” at Politico (January 14, 2018)

The concept of fake news is, however, slippery. Claims about its insidious effects are often overstated, sometimes even in the professional literature.

Recently, a Facebook employee, concerned by his firm’s immense, unrestricted power to police speech, gave The New York Times 1400 pages of unpublicized rules for moderators:

An examination of the files revealed numerous gaps, biases and outright errors. As Facebook employees grope for the right answers, they have allowed extremist language to flourish in some countries while censoring mainstream speech in others…

Anton Shekhovtsov, an expert in far-right groups, said he was “confused about the methodology.” The company bans an impressive array of American and British groups, he said, but relatively few in countries where the far right can be more violent, particularly Russia or Ukraine…

In Sri Lanka, Facebook removed posts commemorating members of the Tamil minority who died in the country’s civil war. Facebook bans any positive mention of Tamil rebels, though users can praise government forces who were also guilty of atrocities…

One moderator described an officewide rule to approve any post if no one on hand can read the appropriate language. This may have contributed to violence in Sri Lanka and Myanmar, where posts encouraging ethnic cleansing were routinely allowed to stay up… Max Fisher, “Inside Facebook’s Secret Rulebook for Global Political Speech” at The New York Times

Most moderators are not skilled and have only a few seconds to decide on a post.

The trouble is, Facebook seeks the impossible: a global approach to a world where fundamental values differ markedly by region. In Canada, threatening violence usually means prompt attention from the police; in Sri Lanka, it may be a routine aspect of civil conflict. But where global rules for censorship may prove impossible, selective censorship is comparatively easy. However, it conflicts with historic freedoms in many regions:

Facebook bans more groups in countries where pressure is applied. Far left activist groups like the Southern Poverty Law Center are able to exert undue influence on Facebook in order to get people and groups on the right in the U.S. banned. Facebook in fact partners with the SPLC. RACHEL ALEXANDER , “LEAKED: Facebook’s Draconian Secret Rulebook for Regulating Political Speech” at The Stream

Groups of which Southern Poverty Law Center disapproves, for political reasons, are not chosen exclusively because they break the law or pose a threat of violence to the public. One non-violent family values group, Family Research Council, was targeted by a shooter in 2012 because the Council was listed as a hate group on the Southern Poverty Law Center website.

As Alexander says, “No one elected Facebook to decide how speech should be regulated.” One could say the same for the groups that are helping Facebook decide.

Ultimately, censorship is part of the larger question of whether social media are the telephone company (a communications platform), the newspaper (a publisher), or unregulated private klatsches. Or something else altogether? Current events will likely keep bringing us back to the growing need for that discussion.

See also: Research showing that fake news easily fools us collapses. A recent paper claiming that low-quality news (“fake news”) spreads as quickly on social media as accurate news has been retracted by its authors.

AI Hype Countdown #5: AI Can Fight Hate Speech!

and

No, Twitter is not the New Awful. It’s the Old Awful back for more. It’s the Town Without Pity we all tried to get away from


Facebook’s Secret Censorship Rules Expose a Key Problem