Mind Matters Natural and Artificial Intelligence News and Analysis
Blue bird
Blue bird

Twitter Twitted Over Clumsy Political Censorship

Getting the algorithm to censor users can backfire and sometimes the results are funny

I once wrote an online forum system called Areopagus. It had a simple “bad words” (profanities, etc.) filter. It looked for common bad words and common variations of those words and automatically prevented any post containing those words or their variations from being posted. This is called “mechanical censorship” and, on its own, it is nothing new. Some users came up with cleverer variations of the bad words than the filter was set for and then moderators had to spot and remove the posts.

During the recent presidential election campaign, Twitter has taken mechanical censorship to new levels. Instead of censoring ways of speaking, Twitter has decided to censor specific ideas and thoughts. Automatic censorship of ideas is a problematic use of technology in itself but Twitter has been so inept that its efforts boomeranged. All the things Twitter aimed to repress were instead highlighted and Twitter itself became the butt of the joke.

By now, everyone has probably heard about Twitter’s attempt to censor the New York Post’s story about damaging materials found on Joe Biden’s son Hunter’s laptop.

In an unprecedented step against a major news publication, Twitter blocked users from posting links to the Post story or photos from the unconfirmed report. Users attempting to share the story were shown a notice saying: “We can’t complete this request because this link has been identified by Twitter or our partners as being potentially harmful.” Users clicking or retweeting a link already posted to Twitter are shown a warning the “link may be unsafe”.

Kari Paul, “Facebook and Twitter restrict controversial New York Post story on Joe Biden” at The Guardian

Twitter attempted to remove the Post story, “Smoking-gun email reveals how Hunter Biden introduced Ukrainian businessman to VP dad” (October 14, 2020), from its entire website. Nothing in the article was really new and most people who knew about the allegations either already believed them or didn’t. But MIT’s Technology Review shows that attempted censorship doubled the exposure of the article to audiences.

At the time, I tested the censorship myself, and below is the screenshot of my own attempt to post the story to Twitter:

The image at the bottom says, “We can’t complete this request because this link has been identified by Twitter or our partners as being potentially harmful. Visit our help center to learn more.”

The outrage vented at Twitter showed that the social media giant couldn’t single-handedly prevent distribution of the article. Twitter finally allowed the article to be posted on its platform, but kept the New York Post in “Twitter jail” for many days thereafter.

The Wall Street Journal’s Editorial Board published a critical piece pointing out that Twitter’s reasoning for its actions made no sense; it seemed to modify its policies at will: “When social-media companies sanction political speech they don’t like, they always point to one policy or another that was supposedly violated. The truth is they are often making up the rules as they go.”

This week, in the post-election craziness, there have been a number of allegations of fraud. Some allegations involve hacking of voting machines using pieces of software called “hammer” and “scorecard.” Unfortunately for Twitter, there was no single source for this claim. So Twitter simply turned the censorship dial up to 11. Any tweet containing the words “hammer” and “vote” was automatically tagged with a message saying, “This claim about election fraud is disputed.”

Several Twitter users decided to have some fun with that. Below is a screen cap of my feed (username is removed because it contains bad words):

After innumerable users started having their own fun with the clumsy mechanical censorship, Twitter eventually relented and turned off the filters.

For now, Twitter’s attempts to insert itself into its users’ conversations have backfired. But what happens when the techniques get better? What happens if Twitter’s authority as a censor is accepted?

The point isn’t the content. It is dangerous for Twitter to be telling people what topics of conversation are permitted and which things are true or false. Even if the Twitter authorities start out being correct, such a position is ripe for abuse. Thankfully, for now, the users have forced Twitter to back down. How long will that last?

You may also wish to look at:

What;s the main thing we should learn from the big Twitter hack? Yes, Twitter got control of its platform back but not before its credibility in security matters was significantly weakened


Journalists’ reliance on Twitter “may” lead to pack journalism. A 2018 study tracked the importance that journalists ascribe to what they hear on Twitter and the results are not good. (Jonathan Bartlett)

Jonathan Bartlett

Senior Fellow, Walter Bradley Center for Natural & Artificial Intelligence
Jonathan Bartlett is a senior software R&D engineer at Specialized Bicycle Components, where he focuses on solving problems that span multiple software teams. Previously he was a senior developer at ITX, where he developed applications for companies across the US. He also offers his time as the Director of The Blyth Institute, focusing on the interplay between mathematics, philosophy, engineering, and science. Jonathan is the author of several textbooks and edited volumes which have been used by universities as diverse as Princeton and DeVry.

Twitter Twitted Over Clumsy Political Censorship