Hardness is useful bur in English is a misleading synonym. Hard science is more quantitative and has more rigor in terms of hard proof and mathematical models and less room for fuzziness.
"Soft" science is fuzzier with more qualitative elements. It doesn't necessarily mean less difficult.
Another distinction is that the level of tools limits things. Say studying color before and after the ability to analyze wavelengths.
From my perspective, it's a term, coined by a 19th century philosopher, that gets dragged out by people with physics envy.
Modern biology, for example, is incredibly quantitative: molecular biologists collect massive genomics datasets, neuroscientists record from hundreds of sites in the brain, and ecologists, whom everyone thinks of as flannel-clad outdoorsmen, fit sophisticated statistical models. Psychology is getting there too, especially if you're willing to lump in related industries like advertising: Google and Facebook have massive datasets on people's activities and preferences.
It's true that these fields have had less success with mathematical modeling, but you also have to look at the sheer number of interacting variables vs the ~10 terms in Maxwell's equations. It's possible that the underlying phenomena might often just be irreducibly complex, as this perspective argues: (preprint: https://www.biorxiv.org/content/10.1101/764258v3 and was just published in Neuron).
"Soft" science is fuzzier with more qualitative elements. It doesn't necessarily mean less difficult.
Another distinction is that the level of tools limits things. Say studying color before and after the ability to analyze wavelengths.