Recordings made across several decades grew steadily louder, and the loudness was achieved by removing the difference between quiet and loud passages. The reasoning behind it was sound and the outcome was not.
Louder wins a direct comparison
Presented with two versions of the same material at different volumes, listeners consistently prefer the louder one, and describe it as fuller and more present.
The preference operates below conscious judgement, which means it cannot be corrected by knowing about it.
Anyone whose record might be heard next to another therefore had an incentive to be the louder of the two, and that applied to everybody at once.
Compression is how loudness is bought
A recording has a maximum level it cannot exceed, so the only way to raise the average is to bring the quiet parts up and hold the loud parts down.
Compression and limiting do exactly that, reducing the distance between the softest and loudest moments so the whole can be pushed higher.
Applied gently this is ordinary craft. Applied hard it flattens the material until the level barely moves from start to finish.
The cost is impact and fatigue
Dynamic range is how music creates emphasis. A chorus feels larger because the verse before it was smaller, and removing the difference removes the effect.
Drums suffer most, since a snare depends on a sharp transient that limiting is specifically designed to reduce.
Sustained listening to heavily compressed material is also tiring, because the ear receives no variation and never gets a moment of rest.
Normalisation removed the incentive
Streaming services adjust playback so that tracks play at a comparable perceived level, turning quieter recordings up and louder ones down.
Under that system a heavily compressed master gains no advantage, and it arrives with its dynamics already destroyed and no compensating loudness.
The competitive logic that drove the practice for decades was eliminated by a single playback decision applied across catalogues.
Why the habit outlived its reason
Mastering conventions are slow to change, and a generation of listeners learned to associate the compressed sound with records they liked.
Some genres also use compression as a deliberate texture rather than as a volume strategy, and there the practice is a choice rather than an arms race.
The distinction that matters is whether the compression is producing a sound someone wanted or simply winning a comparison that no longer takes place.