My notes and other stuff


On Theories and Tools

Always loved page 140 From William Kent's Data and Reality Ch. 9, titled Philosophy:

Theories tend to distinguish phenomena. A theory tends to be analytical, carefully identifying all the distinct elements and functions involved. Unifying explanations are abstracted, relationships and interactions are described, but the distinctness of the elements tends to be preserved.

Good tools, on the other hand, intermingle various phenomena. They get a job done (even better, they can do a variety of jobs). Their operation tends to intermix fragments of various theoretical phenomena; they embody a multitude of elementary functions simultaneously. That’s what it usually takes to get a real job done. The end result is useful, and necessary, and profitable.

Theories tend toward completeness. A theory is defective if it does not account for all aspects of a phenomenon or function.

Tools tend to be incomplete in this respect. They incorporate those elements of a function that are useful and profitable; why bother with the rest? The justification for a tool is economic: the cost of its production and maintenance vs. the value of its problem- solving functions. This has nothing to do with completeness. (In 1975, a government official asked to have his job abolished, because nobody actually needed the services of his office. His job did have a well-defined function, in theory. “Completeness” would have dictated that his job be retained.)

Useful tools have well-defined parts, and predictable behavior. They lend themselves to solving problems we consider important, by any means we can contrive. We often solve a problem using a tool that wasn’t designed for it. Tools are available to be used, don’t cost too much, don’t work too slowly, don’t break too often, don’t need too much maintenance, don’t need too much training in their use, don’t become obsolete too fast or too often, are profitable to the toolmaker, and preferably come with some guarantee, from a reliable toolmaker. Tools don’t share many of the characteristics of theories. Completeness and generality only matter to the extent that a few tools can economically solve many of the problems we care about.

Thus the truth of things may be this: Useful things get done by tools that are an amalgam of fragments of theories. Those are the kinds of tools whose production and maintenance expense can be justified. Theories are helpful to gain understanding, which may lead to the better design of better tools. This understanding is not essential; an un-analytic instinct for building good tools is just as useful, and often gets results faster.

Of course there's also a good first line on page 141 to close it up:

It may be a mistake to require a tool to fit the mold of any theory. If this be so, then we’d better be aware of when we are discussing theory and when we are discussing tools.