Album leaders have to earn the ceiling
The first era rule I shipped was a clamp: songs from a given year could not outrank a ceiling I had picked for that year. It did suppress recency. It also handed a forgettable album track a reputation it had not earned, and it flattened three Maroon 5 singles to the same number despite one of them leading on every chart measure I actually had.
The rule now is the other way around. Each year still has a base cap. A track only rises through it if the evidence pays. Album leaders do not inherit the album's fame. They climb with streams, or a decisive chart peak and certification, or curated acclaim. Ordinary hits stay on the base.
The base caps
| Year | Base cap |
|---|---|
| 2020 and later | 90 |
| 2015–2019 | 92 |
| 2010–2014 | 93 |
| 2000–2009 | 94 |
| Before 2000 | 99 |
Pre-2000 is effectively uncapped. The ladder and the classic-smash floors do the work there, not a year bracket. That is deliberate: I will not pretend a 1979 radio era needs a popularity ceiling the streaming era invented.
What earning looks like
Stream relief is reserved for generational numbers. The bar is 2.5 billion verified plays for +2 and 4 billion for +3, and it is harder for a brand-new song because streams accrue faster now. A song out this year has to clear about 2.08 times those totals. The multiplier decays to 1 by age six, so established hits are not punished. That is how The Weeknd's Blinding Lights (2020, base 90) sits at 93 in the catalog without dragging every other 2020 single up with it.
Chart relief is the other door, and it exists because stream volume alone used to flatten records that were obviously not peers. On Maroon 5's V, Maps, Animals and Sugar used to truncate to exactly 93. Sugar is a #2, certified 8× platinum. Maps is not. They now separate: Sugar 95, Animals 94, Maps 93. The extra points come from peak and certification, taken as a max with the stream bonus, never summed, so the ceiling stays bounded.
Acclaim is the third door. Kanye West's Runaway carries a curated acclaim of 92. Under the old clamp it could not pass the 2010 cap of 93, so it sat next to ordinary hits of that year. It is 94 now. The album-level cap is the maximum earned cap across the album's tracks, so a strongly certified non-leader is not held down by a streamier labelmate.
What does not earn it
Big Sean's Wolves (2020) is 87. Latto's Big Energy is 87. Both would have looked like "era cap 90" songs if the ceiling were a gift. They are ~400 million-stream cuts. The stream ladder puts them in the high 80s, and they stay there. The regression suite fails the build if either of them reaches 90.
That check exists because the first version of this system loved round numbers. A cap is a magnet. Anything nearby sticks to it unless you force the evidence to speak.
The thing I will not do
I will not raise a whole album because the famous song on it is famous. That was the original sin: inherit the ceiling, then wonder why filler tracks from celebrated years outranked singles from quieter ones. Age adjustment of the input streams is the recency tool now. The output cap is earned. Those two jobs used to be one badly designed clamp. Splitting them is written up here.
This still does not answer biggest versus greatest. A 93 that got there with 5 billion streams and a 94 that got there with acclaim are still sitting next to each other. The cap is a ceiling, not a philosophy. That argument is open.
Play #1 Debut