Unable to load image

X's algorithm was changed in mid-July 2024 to boost right-leaning accounts :marseyschizowall:

https://news.ycombinator.com/item?id=42160659

Mastodon thread with lots of sneed

Link to paper

Award for best comment:

In fascism, nothing matters more than the ruling party's power. There was no question from jump that that was where Elmo's hand was on the scale.

Being a POS racist from POS South African white supremacy, it's no surprise that he found a welcoming home here in the economic and political remnants of Earth's most brutal racial oppression of Black folks by white folks.

There is no escape, however, from one's karma. There is no drug, ketamine included, that can assuage the guilt or pain of a lifetime of being an ahole. As Bob Marley said, tho, "Time will tell." Elmo has no loving family, just power and wealth. At the end of the day, that does not give one the inner peace earned by a lifetime of selfless, compassionate service to one's fellow human beings.

As for Black folks, they have never been under any delusion that America is anything more than what they showed this year. For them and those of us fighting for equality and equity for ALL human beings, we move forward, letting the fools tell us who they are and who and what they serve.

The only question any of us face is, "Do you fight for compassion, or against it?"

https://news.ycombinator.com/item?id=42160804


https://old.reddit.com/r/Asmongold/comments/1gswsyx/xtwitter_algorithm_changed_in_midjuly_2024_to/

67
Jump in the discussion.

No email address required.

This methodology is offensively terrible. They only tracked 5 left-wing accounts (Alexandria Ocasio-Cortez, Bernie Sanders, Cory Booker, Kamala Harris, and Stacey Abrams) and five right-wing accounts (Jack Posobiec, Tucker Carlson, Ben Shapiro, Donald Trump Jr., and Libs of TikTok). Notice how all of the left-wing accounts are politicians and the right-wing accounts are professional posters.

They find that all metrics for all accounts go up the same amount after July 12th, with the only difference being the Republican accounts saw a larger increase in views. They say

The findings suggest that there is likely some kind of pro-Republican bias for view counts after the detected structural break on July 12. This is not the case for the structural break we detect for retweet and favourites counts, which occurs on July 14 for both engagement metrics and appears to affect both groups equally.

But if the algorithm started increasing the amount of views for Republican accounts, wouldn't all other metrics also go up a statistically significant amount? All this study proves is that elected politicians get more engagements per view than people whose job it is to post ragebait. Oh, and that all political posts got more engagement after Trump's assassination attempt and as election day got closer

Jump in the discussion.

No email address required.

Lmao PHDcels are a joke of a job

Jump in the discussion.

No email address required.

PhD dissertations are essentially "this is what it might look like if I had a job" and are graded by people who have never had jobs

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.