12
Policy Brief No. 45
Toda Peace Institute
publicly by outside partners. One partner, the Coral Project, is an open source comment
moderation project that originated as a collaboration between the New York Times and the
Washington Post.
Redirect Method (mentioned already when discussing MoonshotCVE) followed a similar
trajectory. It is a methodology originally developed at Jigsaw, after they helped convene a
2011 conference about violent extremism, which has now been implemented by outside
partners such as MoonshotCVE. Redirect Method attempts to redirect people who search for
extremist content to credible voices whose videos challenge the extremists’ arguments, and
was assessed in depth by RAND Corporation in 2018.
3. Brief Notes on Other Platforms: Twitter, Amazon, and Reddit
Like Facebook and Youtube, Twitter has deplatformed users, often after public outcry,
though in some high-profile cases they have resisted doing so longer than the other two
companies. The company used to have an internal slogan saying they were the “free speech
wing of the free speech party.” They also got a lot of praise for — and drew a lot of
encouragement from — supporting the Arab Spring soon after the company was founded,
so it is no surprise that Twitter employees feel honour-bound to support controversial
voices.
Yet although Twitter has generally considered itself pro-free-speech, there has always been
content that violated its rules and required deletion. Twitter has long faced the technical
challenge of moderating content at unprecedented scale. Their specific process is detailed
in a 2018 Logic Magazine piece by Tarleton Gillespie, and has many of the same problems as
Facebook’s.
The main difference between Twitter and other social media platforms is that it has a much
bigger harassment problem, especially for high-profile users and marginalised users. The
human rights organisation Amnesty International is the latest of many to release a 2018
report about Twitter being toxic for women. Because the harassment problem is a constant
burden on high-profile users, high-profile users have spent years asking for more effective
harassment management tools. A 2018 Bloomberg op-ed by Noah Smith is a good example
of an article that not only describes the problem, but asks Twitter to work on very specific
features in response. For instance, Smith suggests that “users should be able to lock individual tweets, closing them to replies.” (It is currently possible to lock tweets, but only by
locking the entire account.)
Smith, who is based in San Francisco and writes about technology, also notes that when he
talks to Twitter employees around town, “what frightens them most is the idea that Twitter
might be used to create echo chambers, where like-minded people are not exposed to
contrary viewpoints.” But the idea of social media echo chambers is widely believed to be
debunked — although it is possible that we do not know enough about Twitter, specifically,
to reach conclusions about its particular dynamics.
Early in 2018, Twitter put out a request for research proposals about “conversational health
metrics.” The two proposals they accepted focus on (1) examining echo chambers, and (2)
bridging gaps between communities, which supports Smith’s claim.