Could we use this sorting algorithm on Lemmy?
cross-posted from: https://lemmy.dbzer0.com/post/11252604
I found this link aggregator that someone made for a personal project and they had an exciting idea for a sorting algorithm whose basic principle is the following:
- Upvotes show you more links from other people who have upvoted that content
- Downvotes show you fewer links from other people who have upvoted that content
I thought the idea was interesting and wondered if something similar could be implemented in the fediverse.
They currently don't have plans of open-sourcing their work which is fine but I think it shouldn't be too hard to try and replicate something similar here right?
They have the option to try this out in guest mode where you don't have to sign in, but it seems to be giving me relevant content after upvoting only 3 times.
There is more information on their website if you guys are interested.
Edit: Changed title to something more informative
Not really interested in anything that even remotely mirrors "engagement" driven algorithms seen on other sites. It's predictably resulted in siloing of information and the explosion of "rage-bait" content that's pretty much taken over. Lemmy being different than that is a boon, not a deficit.
At the moment, upvotes and downvotes, while not used that way by many people, is more about what others will see, rather than what content you like. It's more like a community moderating and rating effort. Upvotes make posts more visible, by pushing them further up in what's currently popular. Downvotes do the opposite, and in my personal opinion, should be reserved for posts that don't fit the community they were posted in, spam, or things that break rules – typically the same reason why you would (and should) report a post. They are not "agree" and "disagree" buttons. Topics you disagree with can still spark interesting conversations.
Using the same mechanic, voting, to tell an algorithm whether similar posts should have higher visibility on your own feed, would be incompatible with this existing system. Posts that get a quick reaction or emotion out of you are even further encouraged, while things you simply don't want to see (but aren't necessarily "bad") get punished heavily.
This system works through subscribing to communities you are interested in and actively participating in improving the health of those communities, rather than passively consuming content. That takes some effort, yes.
All in all I think this proposed system is not compatible with Lemmy, and maybe not even a good idea.
Thanks for this insightful comment. I didn't see it from that perspective before and I agree that using this sort of algorithm makes it harder for community moderation to happen in any capacity.
We should work towards making algorithms that work for us instead of thinking we can rely on just chronological feeds even as we scale. We are sort of doing this by adding scaled and controversial sort but there I think that there should be something more customizable than what we currently have.
Maybe having the defaults but also having some area where it is possible to define a personal sorting algorithm would be nice.
I would add that factually wrong or misleading posts (e.g. that one the other day with an inflation percentage that didn't even match the other numbers in the same graph) are a good reason for downvotes too.
Why not implement it at the client level?
Is that possible?
It requires the information who upvoted and downvoted what, I don't think that is available to the client, is it?
It is in theory. I dont think its a practical lookup tho but it could be amortized to make it workable.
Tho i would just slap AI on it and call it a day /s
I think that approach drives polarization and echo chambers. It literally suppresses dissenting opinion.
I think that approach to content is directly responsible for the increasing division we see in political discourse.