I'll say this about multiband processing; it tends to make sounds take up more space. This can be super helpful for making sparse arrangements sound more full. But in the context of a more dense arrangement, there's value in having sounds that poke through but then get out of the way so that other things can be heard, for example, a bass's pluck has plenty of high-mid content. But that's short in duration, while the sustain is mostly fundamental. Limiting this signal has the effect of reducing the apparent volume of the attack, and increasing the relative volume of the sustain (which might be necessary, for say, slap bass). But a bass which takes up more space in the mix is more likely to get turned down so that other things can be heard. That doesn't mean it's universally bad to compress. Just, like anything in music and art and audio, it's highly subjective.
