I think one of the major distinctions between "hard" and "soft" or "physical" and "social" or however you want to define it is the level of determinism you get.
with math there is very hard determinims
with chemistry too
physics gets a little less - quantum mechanics deals with probability fields for subatomic particles
medicine deals with probabilities in treating illnesses because of variability in the patients to both the disease and the medications. Thus we see things like 90% survival in 20 years with treatment.
meteorology deals with probabilities of weather systems based on past systems and models of systems that input known variables and assume others.
the "social" sciences deal almost exclusively with probability systems
What this means is that experiments cannot be done simply with a small sample and get valid results, but that they have to include large samples and have controls for other "pushing" factors that can bias the data (thus requiring even larger sample bases for the data)
It also means that they cannot make "hard" conclusions but "soft" ones.
Science in general deals with a range of "hard" to "soft" data fields.
Evolutionary Psychology is probably the softest I know of to date that can still be called a science (some will debate this) because it is hard {heh} to do a social experiment with historical data. But it does have {testable\falsifiable} hypothesis, and that, bottom line, is the distinction between science and pseudo-science.
Don't know if that helps.
we are limited in our ability to understand
by our ability to understand
RebelAAmerican.Zen[Deist
... to learn ... to think ... to live ... to laugh ...
to share.