I’ve been exploring how different platforms decide what reaches the front page, and I’m interested in which ranking algorithms you know and which you think work best. Here are a few examples with links to their explanations:
-
Lemmy’s ranking – vote-based score with time decay https://join-lemmy.org/docs/users/03-votes-and-ranking.html#sorting-posts
-
Weighted/outlier-based approach – normalizes performance by community size, highlights statistical outliers using z-scores, and applies time decay https://reddthat.com/post/52386265
-
Menéame’s promotion system – dynamic karma thresholds, weighted votes (user-karma × vote; anonymous votes = 4), plus a minimum-vote requirement https://www.meneame.net/faq-es
My preference is for algorithms that give niche communities a fair chance, rather than allowing large communities to dominate. I’d also like to see a user setting that counts only votes from one’s home instance, so each instance develops a more distinct front page instead of everything looking the same across the network.
Another issue is that many ranking systems would benefit from clearer naming and more coherent category boundaries. For example, Lemmy’s New Comments sort is effectively a classic forum-style bumping model and would be more intuitive if labeled accordingly. Meanwhile, Top Comments is relatively weak in its current form and would be far more useful if it mirrored the “Top” family by offering consistent time-window variants (day, week, month, year, all-time). More broadly, overlapping sorts such as Active, Hot, and Scaled can blur together and confuse new users without delivering meaningfully different discovery experiences. Renaming them or supplementing them with brief tooltips could help clarify their purpose.
Which ranking algorithms are you familiar with, and which do you think work best?