Hard Versus Soft Sciences

I actually don't believe there is any distinction between hard sciences and soft sciences. Below are some of my writings from Wikipedia:

One perceived difference supporting the distinction is the degree to which conclusions in different fields are controversial within those fields. Some believe that conclusions from physics or chemistry tend to be less controversial among physicists and chemists, versus how much of political science is controversial among political scientists. However, in most physical sciences there has been extensive debate about issues like whether atoms exist and whether randomness is inherent in subatomic particles. Russ Roberts from George Mason University claims that although many people romanticize about the objectivity of the so-called hard scientists, many physical scientists constantly engage in controversies and arguments.

There is much difficulty distinguishing between soft and hard sciences because many social sciences like economics use the scientific process to formulate hypotheses and test them using empirical data, i.e. econometrics. Furthermore, many social scientists engage in experimental work within the field of experimental economics. In most cases the methodology used by practitioners of the so-called soft scientist are the same as those used by practitioners of the hard sciences and the only difference is the object studied. Physical scientists tend to look at atoms, energy, waves, etc while social scientists tend to look at societies, individuals, firms, etc.

In all experimental or empirical sciences there is a need to set up experiments. One necessary feature of experiments is the need to control for all factors. It may be hard to control for all factors in an experiment because the experimenter may not account for all factors. This problem exists in the social sciences and the physical sciences. To establish causation the experimenter needs to have a control group where only one variable, the variable of interest, is changed, and all other variables held constant. The difficulty is in how to control for all other variables when there could potentially be infinite variables.