Shallower algorithms have fewer overtones and the overtones tend to be harmonic.

Deeper algorithms have more overtones and often those overtones are enharmonic.

But also how you tune and mix plays a role…a shallow algorithm witH carriers tuned one way will sound consonant and with a different tuning will sound dissonant.

Or to put it another way, FM is awesome for sound design but like everything in music it is hard to learn quickly.