Story in The Economist showing Twitter was favoring conservative and the least accurate media outlets over accurate ones. Doesn't really fit Elon's narrative. If he wants people to take his narrative seriously he needs to give all this data to serious media outlets.
Its data shows a bias aiding unreliable media, regardless of ideology, and right-wing political parties
www.economist.com
According to Twitter, Twitter’s algorithm favours conservatives
Its data shows a bias aiding unreliable media, regardless of ideology, and right-wing political parties
Among the most hotly debated questions on social media is how algorithmic bias affects social media. In America conservatives claim that Facebook and Twitter bury or outright censor their views. The left retorts that right-wing conspiracy theories like QAnon flourish on these sites.
An unlikely arbiter recently emerged in this debate: Twitter itself. In October it released a paper showing that its algorithm, which picks which tweets users see in which order, favoured right-leaning American news sites. In six of the seven countries studied, the algorithm also gave a disproportionate boost to lawmakers from conservative political parties. Twitter shared its data with
The Economist this month, letting us test the authors’ claims.
The study relied on a large experiment. Until 2016 users saw tweets only from accounts they followed, shown in reverse chronological order. After launching its algorithm, Twitter kept 1% of users in the old system. This let it measure how often its algorithm served up certain tweets, compared with the “reverse-chron” method.
In April-August 2020 the authors used this approach on 3,634 accounts belonging to legislators from 32 political parties. Although they did not detect political bias in the treatment of individual lawmakers, they did find a slant when grouping accounts by party. In all countries but Germany, the algorithm’s “amplification ratio” was lower for members of leftist parties than for members of right-wing ones.
This discrepancy could arise for reasons besides ideology. To test alternatives, we fed Twitter’s data into a model that accounted for the amount of amplification in each country, political parties’ vote shares in the most recent elections and whether they were in government. Yet even after making these adjustments, the algorithm still favoured conservative parties.
In contrast, the evidence for bias aiding right-wing American media seemed less robust. The algorithm did give extra amplification to news sources that independent groups like Ad Fontes Media classify as conservative. However, ideology and accuracy (which Ad Fontes, among others, also scores) are correlated to each other. And among the sites studied, those with the strictest sourcing and fact-checking also tended to have left-of-centre politics.
In 2019 we studied how Google ranks news stories, and found that accuracy, not ideology, explained its rankings. This is also true of Twitter. However, whereas Google gave higher rankings to more reliable sites, we found that Twitter boosted the least reliable sources, regardless of their politics. Left-wing sites with poor accuracy scores, like tmz, were amplified more than credible, conservative ones like the
Wall Street Journal. ProPublica, a non-profit focused on public-interest investigations, had one of the lowest amplification ratios.
Because Twitter did not share the tweets it studied, we could not identify the type of content that its algorithm rewards. But if the company wants to reduce misinformation on its site, making tweaks to favour rigorous reporting might help.■
Sources: “Algorithmic Amplification of Politics on Twitter”, by Ferenc Huszár et al.; Ad Fontes Media; Media Bias/Fact Check; NewsGuard
