After Twitter made the code for its algorithm open source at noon Pacific Time on March 31, users began to pick through it, and discovered that tweets judged to be about Ukraine were down-ranked – meaning users were less likely to see them in their feed.
Twitter user Aakash Gupta (@aakashg0) got together with a group of others to sift through the code for the algorithm and discovered the secrets to boosting your follower numbers on the site – as well as the fact that like topics judged to be “misinformation,” the topic of Ukraine is highly down-ranked.
Anecdotally, Twitter users who post frequently on Ukraine topics had noticed less engagement with their accounts since Musk took over the platform in October last year. Musk’s own position on support for Ukraine is ambiguous.
While Musk has aided Ukraine by providing it with Starlink satellite Internet terminals, which the country has used to keep communications up and running for both military and civilians during Russia’s full-scale invasion, some of his other public positions regarding the war have come in for criticism – even from Ukrainian President Volodymyr Zelenskyy.
Musk in early October last year tweeted a “peace plan” for Ukraine that highly favored Russia’s position, and a poll of users asking whether "the will of the people" should decide if seized regions remain part of Ukraine or become part of Russia.
Zelenskyy himself fired back with a twitter poll asking "which Elon Musk do you like more?": "One who supports Ukraine" or "One who supports Russia."
Musk replied that he still very much supported Ukraine, but he was afraid that a “massive escalation” of the war would hurt not only Ukraine but the whole world.
Musk had not commented on the down-ranking of posts about Ukraine by the Twitter algorithm as of the time of publication.
Meanwhile, news that the Musk had made the Twitter algorithm open source was welcomed as a step towards transparency by many users, although some had worries that by revealing how tweets are ranked, it was now much easier to game the algorithm.
Twitter user Brandon Fish (@brandonfish), replying to a Musk tweet about making the Twitter algorithm open-source, commented: “With the recommendation code and API access, someone could build a tool to predict how many impressions a tweet would get.”
Another user, LeGate (@williamlegate) noted: “seems like a terrible idea if you’re worried about large botnets which could, quite easily, be self-trained using such a model.”
Read the original article on The New Voice of Ukraine