Can ranking manufacture a star when everyone starts equal? Each dot is one creator at a fixed position; its height is the square root of its share of that feed's views, so small shares stay visible. The cyan dots stay level, while in the amber feed one usually pulls ahead.
Try this Run Equal odds, then Strong feedback (ranking-strength 160, no discovery): ranked-top-share usually ends above 90% while equal-top-share stays near 5%, and your diamond gets almost no ranked views. Then try Room for discovery, which sends half the views through the equal-odds draw.
About this model Where it ends up
A MathPets experiment on feedback in ranked feeds. Left, cyan: every eligible creator has an equal chance of being shown. Right, amber: past views buy better odds. Both feeds share the posting schedule, every random draw, and a fixed number of viewing slots, so any difference between them comes from the ranking rule.
Each tick awards one view in each feed; attention-slots sets how many ticks make a simulated hour, and the run stops at 336 hours (4,032 views per feed at the default 12). The ranked feed picks among eligible creators with probability proportional to (1 + views)^(ranking-strength/100), except for the discovery-share percent of views that use the equal-odds draw. Each creator reuses one random number in both feeds, so at ranking-strength 0 the feeds agree view for view.
The diamond in each feed is you. Connecting X sets only your posts per day (0 to 48). The other 23 creators each post six times daily; posts remain eligible for six hours, and everyone starts with one eligible post. Creator quality and audience size are held equal. Ranking and discovery are assumptions you can change, not estimates of the X algorithm.
Idea MathPets · org
MathPets MathPets · org
