Algorithm competitions are economic systems to push
the state-of-the-art of solving a computational problem.
TopCoder and similar platforms are prominent examples.
Our new competition design is desirable from two viewpoints.
(1) We minimize the effort on the competition administration
by distributing the work of evaluation among participants
while avoiding unfair evaluation.
(2) We minimize collusion so that the strong participants
cannot be outnumbered by colluding participants.
Our design is axiomatic in that we formulate
axioms (including a collusion-resistant axiom)
for ranking functions and prove a representation
theorem. It shows that all ranking functions satisfying
the axioms must have an elegant property useful for
collusion-resistant mechanism design.
Our design approach introduces a new class of games,
called side-choosing games, a generalization of
semantic games, a well-studied area in logic.
In addition to algorithm competitions, our system and
its theoretical foundation is also
useful to pushing the state-of-the-art in formal sciences
and in education in formal sciences. Indeed, the system was
developed in the context of education (Algorithms and Software Development Courses) to facilitate fair
peer grading and focused communication among students.