Every mobile social app re-sorts your feed continuously with a scoring pass over recent posts. This 2D twin exposes the three signals separately as adjustable weights instead of folding them into one fixed formula:
Score(post) = ( wR·Recency + wE·Engagement + wA·Affinity ) / (wR + wE + wA)
Recency(Δt) = e^(-λ · Δt)
Engagement = 1 - e^(-0.15 · (likes + 3·comments))
Affinity = relationship strength, 0..1
- Recency — exponential time-decay since the post was made, same shape as EdgeRank's decay term. A higher λ makes the feed forget old posts faster.
- Engagement — a saturating function of likes and comments (a comment counts 3× a like), so a post can't earn infinite score just by racking up interactions.
- Affinity — your relationship strength with the author. Liking or commenting on a post also raises that post's own affinity, exactly like an engagement signal feeding back into the ranking model.
The weighted score is a genuine weighted average of the three normalized signals — drag a weight to 0 to remove that signal entirely, or push one to maximum to see the feed collapse to pure recency, pure popularity, or pure relationship ranking. The right-hand column always shows the same posts in plain chronological order (newest first, no scoring at all) so you can see exactly which posts the weighted algorithm promotes above — or demotes below — where they'd sit in a plain timeline.