This pisses me off. It's open source, you can check yourself. If you can't understand the code, that's fine, just ask someone. Don't jump straight to accusing people of white supremacy with no evidence because you're bored.
The code is right here.
It searches for a high (today's score - yesterday's score), then, if that's the highest it's seen, takes that number squared and uses it as the trending score, halving the score every max_score_halflife. If the peak is older than max_score_cooldown, it forgets and lets it trend again.
max_score_halflife and max_score_cooldown are admin-configurable, but the defaults are 4 hours and 2 days, respectively.
Gaza isn't trending because the number of people tagging it is never substantially higher than were tagging it yesterday.
No? The news is that this is an ongoing genocide that's been happening in high gear for months and more slowly for decades, and it is not actually something new that happened since yesterday.
while the lead dev is a weido, i think it's just "working as intended", something isn't "trending" if you're talking about it every day for the past 6 months, but if 10 people are posting about planking or w/e then it is trending
Surely that's just part of the code on each instance? If there were a bias it would have to be present there.
I don't use any kind of platform besides this site but unless there's some kind of global analytics system run by somebody affiliated with Mastodon that every instance is set to reference by default I don't know how this would be possible.
The main mastodon dev has some pretty weird opinions so it could be possible they've hidden some implicit bias in the trending function. It would be interesting to see analysis of behavior across other activitypub implementations.
Who the hell still uses kolektiva after the admin's house got raided?
