A/B-ing a loudness-maximized master against the original source at matched level reveals the depth that maximization destroys

Greg Calbi’s quality check: take the original source, ignore overload limits, and A/B it against the maximized version at a matched peak level. Doing so reliably reveals that heavy level maximization has cost a whole bunch of depth — depth that listeners cannot recreate on playback. Matching level before comparing is essential, because the louder version otherwise seems ‘better’ purely from being louder.

Examples

An engineer levels-matches the squashed master against the untouched source and switches between them; the source is audibly deeper, exposing what the loudness push threw away.

Assessment

Why must you match levels before A/B-ing a maximized master against its source, and what does the comparison typically reveal?