Density decay happens when a game runs out of unique concepts. It starts to repeat itself or stretch itself out. Often, it does both at the same time.
A game’s systems afford a limited number of very distinct situations. If all of these situations are explored in the beginning of the game, the player is well prepared to handle the rest of the game, but there are not going to be any significant systemic surprises left. Instead of anything new, they are served combinations of things they have already experienced.
Most people don’t finish the videogames they buy. Many don’t even make it halfway through. Frontloading all of your interesting things makes economic sense. The people who keep playing beyond the beginning of density decay are the most invested, so they are much more likely to keep pushing through. The time invested into the game takes the place of the joy of discovery as motivator.
Density decay also often occurs accidentally, when a game is developed linearily. It is very common for developers to spend a lot of time in their first area, adding little handcrafted flourishes and bespoke reactivity before realising that they are running out of time and need to finish the rest of the game, which then plays off of the already established concepts and flourishes, instead of receiving the same amount of unique care.
Density decay is not a problem or a “flaw” (it makes a lot of sense to frontload unique things, to pull people in), it’s just something I have observed in many different games.