Washington (AFP)

Donald Trump appears as the first result of the accounts suggested by typing in English the word "racist" in the Twitter search. The result of an algorithm that should not fix the already strained relations between the American president and the platform.

The British newspaper The Independent is the first Wednesday to have noted this curiosity in the context of widespread protests across the United States against racism and police violence.

"If an account is regularly associated with certain terms, they can emerge together into recommendations through an algorithm," said a Twitter spokesperson.

The social network for the first time on May 26 reported a tweet from Donald Trump as misleading, before pinning another three days later for "apology for violence".

The Republican President, followed by more than 80 million people on the platform, had responded in the meantime by signing a decree aimed at limiting the judicial protection of social networks.

For Greg Sterling, editor of the site specializing in search engines Search Engine Land, the fact that Donald Trump was suggested first when searching for "racist" on Twitter is not a new episode in this open conflict.

But rather the reflection of "a large number of people using the words + racist + or + racism to describe or respond" to the president; or the result of a "concerted effort to associate the Trump account with these terms."

The algorithm of Twitter, explains the specialist, uses a whole bunch of variables making in principle the platform safe from any attempt at manipulation.

It is also supposed to be the case of Google, however victim in 2003 of a so-called "bombing" maneuver so that the name of former president George W. Bush appears at the top, searching for the expression " miserable failure ".

Only an in-depth analysis would make it possible to explicitly shed light on the mechanisms associating Donald Trump and racism, notes Kjerstin Thorson, a professor at the University of Michigan specializing in politics and social networks.

"The platforms have gone to great lengths to avoid giving the slightest impression of bias," she said.

© 2020 AFP