If a post gets downvoted, it could be a geinuenly awful post. But another post that gets downvoted, but is actually empiracially scientifically true, then it is treated equivolent as the other even though they are the same.
I don’t think this is the answer but one idea is to add points to people, or products, who are verified to be awesome. So that would be a scientist or compassionate politician gets more votes or a healthy product gets a subsidy.
It’s a methodology for achieving what you describe, without manually having to assign social scores to each person.
can you define this? this seems like almost a semantic nightmare in practice. Like tagging all the the topics they up/downvote
That’s what happens: every post one up or downvotes is public. You can just download that data.
Uh huh… like web scraping? There seems to be some confusion-making
You don’t need to scrape. The votes made on one instance need to be propagated to the others. So the information “person X up/downvoted post Y” is openly transmitted as part of the defederation protocol.
The short answers are a bit infuriating for me. I am tempted to ask more questions, but this is becoming an unfun game for me. I guess one could see the votes, but the categorizing the types of topics per user and/or cluster of users sounds like a difficult, combinatorically complex job.
Ah ok, now I see the confusion. I thought you were enquiring how to gather the data.
There’s well known clustering algorithms for that. It’s a long time known and solved problem: https://en.wikipedia.org/wiki/Cluster_analysis
These techniques have the benefit that you don’t have to inject your own biases into the score assignment, you can just let the data speak for itself.
I respect that you’re learned and an expert on this topic. I am getting frustrated how the comments are getting cut off prematurely each time. It keeps begging the question. I’m at a point like “ok great so just program the thing” because anything short of that feels like a waste of time
With a swift google I found this example: https://chrispaynehome.github.io/html/voters.html
The author uses such cluster algorithms to find clusters in British political voting behaviour.
One could do the exact same with lemmy up or downvotes. Find cohorts of similarly behaved voters. Then display each posts score per cohort, instead of the aggregate.
Perhaps it’s not a great thing that, so quickly, a topic feels like a waste of time, simply because you lack knowledge.