In the last few years, the computational geometry community has felt a growing concern over the health of the field. Although computational geometers continue to produce exciting research results, many are troubled by the gap between their field and the application areas that should be its primary constituency.
A task force of leaders of the field, organized by Bernard Chazelle, has proposed in a recently released report that computational geometry renew its focus on applications. This is unquestionably worth doing, but I believe that we need to go further.
Computational geometry results need to be communicated to potential users. When a problem is motivated by an application, geometers must make an effort to carry the results back to the application area--Chazelle's report stresses this. Computational geometry already has results that are needed in practice; those results would be more widely known and used if they were available in the form of easy-to-use implementations.
In this note I argue that existing computational geometry algorithms are often directly relevant in industrial applications. However, the knowledge of those algorithms is not widespread, and robust, easy-to-use, well-publicized implementations are rare. We should strive to package our best geometric algorithms into easy-to-use software tools that can be used by non-specialists. Our goal should be to make geometric tools that are (almost) as easy to use as the Unix ``sort'' command and system call. Such tools would dramatically enlarge the set of potential users of geometric algorithms.