Algorithms - Writing for Computer Science - page 138

Chapter 10
Algorithms
Mostly gobbledygook …
Eric Partridge, defining computation
Usage and Abusage
In many computer science papers, the core contribution is a family of algorithms.
These algorithms are often the product of months of work; the version that the
researchers have decided to submit for publication is typically based on a great deal
of discussion, brainstorming, prototyping, testing, analysis, and debate over details.
Yet in many cases this effort is not reflected in the presentation. Not only are the
steps of the algorithm often not made clear, but there is no discussion of why the
reader should believe that the algorithm is correct, or believe that its behaviour is
reasonable. An algorithm by itself is uninteresting; what is of value is an algorithm
that has been shown to solve a problem.
The topic of this chapter is effective description, analysis, and explanation of
algorithms—perhaps the most challenging single task in writing of papers in com-
puting. Experimental assessment of algorithms is considered in Chap. 14 . Here the
focus is on helping the reader to understand what the algorithm does, what effect it
has, what its value is, and what its properties are.
Presentation of Algorithms
When an algorithm is presented in a computer science paper, the details of the
algorithm by themselves—the program steps, for example—do not show that it is
of value. You must demonstrate that the algorithm is a worthwhile contribution: for
example, show that given appropriate input, it terminates with appropriate outcomes;
or perhaps show, by a combination of proof and experiment, that it meets some
claimed performance bound.