Alcun Atirutan BBS

Big thanks to honestly. Although he misleadingly frames it as "your" fault, what he is explaining here is that 's algorithm is designed to pour oil on every fire it sees.

Some of us have been pointing out how dangerous that is for years now.

Elom Musk tweets:
"Trashing accounts that you hate will cause our algorithm to show you more of those accounts, as it is keying off of your interactions.

Basically saying if you love trashing *that* account, then you will probably also love trashing *this* account. Not actually wrong lol."

Twitter always tried to downplay it, but they learned early on that outrage drives clicks. A short character limit encourages Hot Takes without nuance or context; an algorithm that shows you your personal hot-button tweets will make you tweet in outrage.

We've heard a lot about disinformation in the last few years, but far too little about the structural problems with corporate social media.

They've done huge damage to our society, and it's time for change. Luckily, one deranged billionaire has made that really really obvious.

My personal rule for Twitter now is that I will not ordinarily contribute value to it. I will tweet criticism and explainers about Twitter, and I will promote alternatives to Twitter. That's all Elon is getting from me.

@moh_kohn It's not only the algorithm, probably not even mainly algorithm. People with names and addresses were whipping up hate as their jobs at Twitter.

@moh_kohn Well, it's not so much pouring oil on the fire, but because you're interacting with something, it assumes that other people who interact with those people will have other connections you would like to interact with. The tenor of the interaction matters less, the fact that you created content in response is the key. You ignore them, they assume it's useless to show you similar things. So, interact positively with things you like, don't interact with what you hate.
replies
0
announces
0
likes
0

@kazriko that is incorrect. It is optimising for "engagement" and learned that, statistically, anger is engaging. Twitter's engineers probably can't tell you exactly what the basis of their ML models' decision-making is.

The responsibility lies with the creator of the technology, who could use different models and target different metrics, not with the users.