Hacker News new | ask | show | jobs
by itsoktocry 214 days ago
They created 9 accounts for the study, and had "experts" qualify content as left or right wing. They could have at least tried before publishing this story for the 100th time.
1 comments

>Sky News collaborated with digital consultancy firm 411 to train a Large Language Model - an AI tool - to categorise the content into political and apolitical categories. The definitions for those categories were created in consultation with multiple computer science academics and social media experts. The 5,981 accounts we categorised accounted for 67% of all of the posts in our dataset. See our full methodology here[0].

[0] https://news.sky.com/story/how-sky-news-investigated-xs-algo...

From the article:

>"Specifically, the Sky News team ran a study where they created nine new Twitter/X accounts, three left-wing, three right-wing, and three politically neutral, and then tracked what content got dumped into their “For You” tabs on the Elon Musk-owned social media service during a one-month period in 2025"

Yes, the link I provided gives much more information on their method including the names of the people who set the categories.

I don't find any mention of how/if they separated the accounts by using different IPs/types of internet connection - seems like a bit of an omission.