How do we compare two “good” schedules?

RobinX and the case for shared problems, explicit constraints, and reproducible comparisons.

A shared problem before a shared score

RobinX provides a classification and data format for round-robin sports timetabling. The associated repository includes problem instances and solutions. Its underlying paper is by David Van Bulck, Dries Goossens, Jörn Schönberger, and Mario Guajardo (2020).

The editorial lesson is straightforward: two scheduling systems should not be compared as if they solved the same problem unless their inputs, restrictions, and goals actually match. A fast result with relaxed requirements cannot be read as equivalent to a slower result satisfying all of them.

What an honest comparison would disclose

Publish the input version, tournament format, mandatory restrictions, preferences, and evaluation definitions. Report validity separately from quality. Add time limits, computing conditions, and repeated-run behavior where relevant. Keep failed or infeasible runs in the report.

A useful evaluation could include late-slot distribution, opponent repetition, travel, and changes to a published schedule. These are proposed reporting categories, not a claim that a single RobinX score captures every operational concern.

Open questions

Can a benchmark include the cost of collecting missing requirements? Can it measure whether operators understand why a solution is good? Mathematical performance matters, but so does getting a schedule approved and used. These are research questions for future study, not completed findings.

Read the original

Van Bulck, D., Goossens, D., Schönberger, J. & Guajardo, M. (2020). RobinX: A three-field classification and unified data format for round-robin sports timetabling. European Journal of Operational Research, 280, 568–580. Authors’ project and repository ↗