Defining science and engineering

[Durandal]

We've had a lot of threads over the years about what is and isn't a science. Well, we've got a lot of smart people here, many of whom are scientists or engineers. Let's try and create a SD.Net-approved description of science that is not tethered to any colloquial usage. So just because it's common for every department in a university to have the word "sciences" appended to its name doesn't mean that we need to respect that in our definition.

And while we're at it, we might want to do the same for engineering as it relates to science. Are the two fields separate? Where do they intersect? Is engineering a subset of science?

So, some key, opening questions. If a field of study is pursued empirically, is that field of study a science? Must science describe strictly natural behaviors of the universe? Must science employ mathematics? Do we wish to distinguish between "hard" and "soft" sciences?

[Surlethe]

Obedience to the scientific method must be a necessary condition for a discipline to be considered a science. Because the scientific method has as its goal a description of the universe, model-building must also be a necessary condition. So, a science must be a discipline that uses the scientific method to build descriptive models. In fact, this could make a good definition: a discipline is a science iff its goal is to only create increasingly accurate descriptions of the universe (or some subset thereof) using the scientific method.

The distinction between hard and soft sciences lies in whether the models are chiefly qualitative or chiefly quantitative. Hard science builds quantitative models; soft science builds qualitative models. The difference is the extent of the use of mathematics.

In the other thread, detective work was brought up as a counterexample; this doesn't seem to satisfy the definition above because the detective's overriding concern is not simply the creation of models, but also the application of those models to catch criminals.

This definition seems to include all of the basic areas of science -- physics, biology, chemistry, geology, astronomy, cosmology, etc. Notably, it excludes mathematics and computer sciences, which do not make use of methodological naturalism, though they do make building descriptions their goal (in some sense, they build intrinsic rather than extrinsic descriptions). It also excludes the applied sciences, since they do not have as their chief goal model-building, though they do use the scientific method.

[The Dude]

My preferred definitions:

Science is the creation, testing and revision (or rejection) of hypotheses and theories.

Engineering is the application of scientific principles (usually from the physical sciences) to practical technical problems.

These are more activities than professions. Most engineers and many scientists do some of both.

[Alyrium Denryle]

Obedience to the scientific method must be a necessary condition for a discipline to be considered a science. Because the scientific method has as its goal a description of the universe, model-building must also be a necessary condition. So, a science must be a discipline that uses the scientific method to build descriptive models. In fact, this could make a good definition: a discipline is a science iff its goal is to only create increasingly accurate descriptions of the universe (or some subset thereof) using the scientific method.

The distinction between hard and soft sciences lies in whether the models are chiefly qualitative or chiefly quantitative. Hard science builds quantitative models; soft science builds qualitative models. The difference is the extent of the use of mathematics.

In the other thread, detective work was brought up as a counterexample; this doesn't seem to satisfy the definition above because the detective's overriding concern is not simply the creation of models, but also the application of those models to catch criminals.

This definition seems to include all of the basic areas of science -- physics, biology, chemistry, geology, astronomy, cosmology, etc. Notably, it excludes mathematics and computer sciences, which do not make use of methodological naturalism, though they do make building descriptions their goal (in some sense, they build intrinsic rather than extrinsic descriptions). It also excludes the applied sciences, since they do not have as their chief goal model-building, though they do use the scientific method.

Actually forensics could fit the definition. It depends on the scale of the model that you accept and the particular methodology of the teram involved. For example, in solving a murder a group could create a testable model for the circumstances surrounding the murder (place, method of death, perpetrator) and then running experiments that seek to reject these models. They are building a description of a very small highly specified part of the universe.