New Task Types at the Canadian Computing Competition

In the 2006 competition workshop held at Dagstuhl, Germany, there were many fruitful discussions about the difficulties facing computer science competitions today. Our competitions have several purposes: to foster interest in the discipline, to create a community, and to promote achievement, for example. Balancing these various purposes may require many tradeoffs. Several participants identified areas where we need to improve our competitions. Tom Verhoeff (2006) discussed the problem of giving a meaningful ranking to incorrect solutions. Maryanne Fisher and Tony Cox (2006) pointed out that some groups of students are disadvantaged by the present format. Many participants made suggestions for improving the competitions. One of the suggestions in our paper (Cormack et al., 2006) was open-ended tasks. A task is open-ended if there is no known optimal solution to the problem. Points are awarded for correct submissions in proportion to how well they do. A vast number of real-world applications, such as pattern recognition, information retrieval, and compiler optimization appear suitable for this purpose. At Canada’s national informatics olympiad, the Canadian Computing Competition, we have been exploring several of these suggestions. In this paper we describe the experiments we have performed and we analyze whether the objectives have been achieved.