Designing coding assessments that are fair

A coding assessment is only useful if two equally skilled candidates get the same result. That starts with the problem statement: state the input format, the expected output and the constraints in plain words, and give at least one worked example.

Test cases should cover the normal path, the edge cases and one or two larger inputs. Hidden cases keep candidates honest, but the visible ones should be enough to let someone check their understanding before submitting.

Time limits deserve the same care. Set the limit from how long a competent engineer needs for the task, then add a margin. A limit that only the fastest typists can meet measures typing speed, not reasoning.

Finally, publish the scoring rules. When candidates know how points are awarded, they spend their time on the parts that matter.