The Loudness War is/was about reducing the dynamic range (DR). The DR is the difference between the average loudness and the peak loudness. Because louder sounds better, and peak loudness is always limited to 0dBFS, engineers reduced the dynamic range with compression and limiting, to make tracks louder than others. You basically squeeze the average loudness closer to that upper limit of 0dBFS, for example to -6 dBFS, which is very little. Heavy reduction of dynamic range is considered bad and taking the 'life' out of music.
But I was talking about the peak loudness. Peaks should simple be as loud is possible without distorting. Most systems are at their best with the signal as loud as possible, as the noise stays constant, so you end up with the best ratio of Signal-to-Noise. You can't go over 0dBFS as that adds distortion. (Sometimes used for desirable in the recording/mixing process, but technically impossible in the digital world of streaming and WAV files).
The advent of streaming platforms has actually weakened the Loudness War. Platforms normalise the (average) loudness tracks to around -14dBFS, while asking for peaks at no more than -1 dBFS. That means music can have 13dB of dynamic range without ever sounding less loud than other songs. That is much more than is customary, at least in EDM, pop, hip-hop, rock etc. A small note: the quality of the mix itself has a lot of influence over perceived loudness, and that doesn't show up in the numbers. It is completely possible to have a muddy and soft sounding mix with the same numeric loudness as a punchy clear mix, this is the art of mixing