Recent searches
Search options
Administered by:
There's been a lot of hype about #BlueSky allowing people to choose their own #algorithm and their own #moderation on the #ATProtocol. I've been trying to put into words something that makes me feel uncomfortable about it. This is the best I've come up with so far:
It feels like a libertarian strategy to take power away from communities and give it to individuals. On the surface it sounds like more freedom for everyone. In the end I expect it will be more freedom for some and more trouble for others. 
The #BlueSky discussion seems to falsely conclude that a recommendation algorithm only affects the person consuming the recommendations. But it doesn't. It affects everyone being recommended (or ignored) too.
Look at the recommendation algorithms of #BigSocial. They generally tune their algorithm to keep people on their platform for as long as possible. They’ve discovered that upset people tend to keep scrolling. So what do they do? Slip in a bunch of posts that are likely frustrating to you. If you’re a reasonable person, you might see a bunch of posts from fascists. If you’re a fascist, you might see a bunch of posts from marginalized people who, like everyone, deserve basic human rights. What does this lead to? A bunch of decent people being harassed by fascists. Why? Because of the #algorithm.
Of course, lots of different kinds of algorithms could be made. Some could really work on ways to make people happier and kinder. That’s kind of cool to think about theoretically. But opening the door to such things also means that algorithms could be made that are even worse for the community as a whole. Elon Musk has shown that pretty clearly by taking an algorithm that (let’s be real) was far from perfect and making it substantially worse.
Then there’s the part where it’s important to understand how complex a recommendation algorithm is. The average person will not know or understand that the algorithm they are using is tuned to hide/suppress certain information, like, I don’t know, criticism of the oligarchs. Or pro-democracy efforts. Or the human rights violations Russia is perpetrating in Ukraine. Or lots of other stuff.
I am not an anti-recommendation-algorithm absolutist. But I do think it is concerning that some of these things have not been addressed AT ALL in anything I've seen regarding the #ATProtocol.
I think there are a number of reasons why we need to work on discovery and curation in the #Fediverse. But I think it's critical to look at such things in an opt-in setting, not just as a content consumer, but also as a content producer.