One of the reasons for not disclosing content recommendation algorithms is to combat spam or generally exploitation of the algo.
You’d be exclusively recommended content from whoever is trying the hardest to game the algorithms, and not however the algo was actually intended to work.
Curious as someone who absolutely does not understand things. Would it be possible to create an open source algorithm that could be used successfully? As in, you view certain types of content, similar content is recommended, home feeds are personalized, et cetera, but without the generally horrible shit that comes along with algorithm feeds currently?
Yes, there are many successful open source algorithms, so there is no reason to believe a content recommending algorithm that is open-source could not be successful. If you’re using Lemmy and you sort by “hot”, you are already using one…
One of the reasons for not disclosing content recommendation algorithms is to combat spam or generally exploitation of the algo.
You’d be exclusively recommended content from whoever is trying the hardest to game the algorithms, and not however the algo was actually intended to work.
Curious as someone who absolutely does not understand things. Would it be possible to create an open source algorithm that could be used successfully? As in, you view certain types of content, similar content is recommended, home feeds are personalized, et cetera, but without the generally horrible shit that comes along with algorithm feeds currently?
Yes, there are many successful open source algorithms, so there is no reason to believe a content recommending algorithm that is open-source could not be successful. If you’re using Lemmy and you sort by “hot”, you are already using one…