Abstraction is a generic technique that allows the scientist or engineer to focus only on certain features of a system while hiding others. Scientists in all disciplines typically use abstractions as a way to simplify calculations for purposes of analysis, but computer scientists also use abstractions for purposes of design: to build working computer systems. Because building systems is the central focus of much work in computer science, the use of abstractions to cope with complexity over a wide range of scale, size, and levels of detail is central to a computer scientist’s way of thinking.

The focus of the computer scientist in creating an abstraction is to hide the complexity of operation “underneath the abstraction” while offering a simple and useful set of services “on top of it.” Using such abstractions is the principal technique for organizing and constructing very sophisticated computer systems, and they enable computer scientists to deal with large differences of scale. For example, one particularly useful abstraction uses hardware, system software, and application software as successive layers on which useful computer systems can be built. This illustrates one very important use of abstraction in computer systems: each layer provides the capability to specify that a certain task be carried out without specifying how it should be carried out. In general, computing artifacts embody many different abstractions that capture many different levels of detail.

A good abstraction is one that captures the important features of an artifact and allows the user to ignore the irrelevant ones. (The features decided to be important collectively constitute the interface of the artifact to the outside world.) By hiding details, an abstraction can make working with an artifact easier and less subject to error. But hiding details is not cost-free—in a particular programming problem, access to a hidden detail might in fact be quite helpful to the person who will use that abstraction. Thus, deciding how to construct an abstraction (i.e., deciding what is important or irrelevant) is one of the most challenging intellectual issues in computer science. A second challenging issue is how to manage all of the details that are hidden. The fact that they are hidden beneath the interface does not mean that they are irrelevant, only that the computer scientist must design and implement approaches to handle these details “automatically” (i.e., without external specification).

SOURCE: Adapted from Computer Science and Telecommunications Board, National Research Council, Computing the Future: A Broader Agenda for Computer Science and Engineering, National Academy Press, Washington, D.C., 1991.