Σεμινάριο: "On the nature of human performance in competitive endeavors"

ΚΥΚΛΟΣ ΣΕΜΙΝΑΡΙΩΝ ΣΤΑΤΙΣΤΙΚΗΣ 2026-2027

Speaker: Ioannis Pavlidis, Eckhard-Pfeiffer Professor of Computer Science, University of Houston, Houston, USA

On the nature of human performance in competitive endeavors

ΑΙΘΟΥΣΑ 401, Ευελπίδων 47Α & Λευκάδος 33, 4^ος όροφος

ΠΕΡΙΛΗΨΗ

Heavy-tailed success is common in human competition, but it is unclear when it reflects runaway dominance versus a system in which skill has a real chance to accumulate. We study outcome distributions in three arenas: World War II fighter aces (victories), U.S. faculty competing for NIH/NSF awards, and Olympic medal totals in swimming and fencing. For each domain, we analyze the full historical record and then partition the data into periods aligned with institutional eras to test whether the “shape of winning” is stable or shifts over time. Across domains with stable rules and sustained chances to compete, the biggest winners follow a common pattern. We call it Relative Fairness: a system that is demanding yet gives skill a fighting chance through repeated opportunities. In contrast, temporary expansions of resources or relaxed selectivity shift the distribution toward a broad-award regime, while episodic dominance produces localized winner-take-all behavior. The result is a falsifiable, cross-domain diagnostic that distinguishes chance-dominated, relatively fair, and dominance-driven competition—and clarifies what different competitive designs imply for both fairness and performance. The broader implication is practical: by reading the “shape of winning,” organizations can diagnose whether a competition is excellence-building, broadly awarding, or winner-take-all—and adjust rules and resource levels to better balance opportunity, incentives, and performance.