has said that its overzealous downranking of accounts was the result of
errors in their system, just three days before CEO Jack Dorsey is set to
testify before the House Energy and Commerce Committee on the subject of
social media censorship.
company said that in its previous system, accounts determined to have a
“higher likelihood of being viewed as abusive” were downranked in public
conversations — up until now.
the platform’s previous system didn’t merely seek to downrank accounts
that were actually abusive, but accounts that could be viewed as
abusive. And not even that — accounts that were likely to
be viewed as abusive.
did not explain whose determinations of “abusiveness” were being treated
said that they turned off the feature, explaining that a “higher level
of precision” is needed. The acknowledgment that its system of
downranking was flawed comes just a few days before Jack Dorsey is set
to answer lawmakers’ questions on the subject of social media
censorship. The evidence session is the result of recent controversy
over Twitter’s blocking and “shadowbanning” practices, which has affected top
Republican politicians but not Democrats.
company said that it would continue to factor “behavioral signals” into
how the platform ranks tweets, in order to serve “conversational
health.” What these signals include remains a mystery.
has thus far refused to acknowledge that its downranking algorithms implicitly
favor the left. Because Twitter factors in the number of times an
account has been blocked or muted when deciding whether to downrank its
content, it has set up a system that favors the easily-offended: people
who like to shut themselves off from contrary opinions. It isn’t merely
a stereotype that these people tend to be found more frequently on the
left — research has found that Democrats are three
times more likely than Republicans block or unfriend people
over their political opinions on social media.