A problem on black hole thermodynamics in a Chinese Physics Olympiad involves the misuse of the concept of temperature as a black hole’s mass approaches zero. Frontier research in finite-system thermodynamics reveals that the conventional concept of temperature is no longer applicable when the mass of a black hole approaches zero. This case thereby exposes an educational contradiction more critical than a mere problem-solving error: exams tend to require students to accept the given assumptions and reproduce “standard answers,” whereas scientific research demands that researchers confront truth, examine the validity of concept definitions, the domains of validity of theories, and whether conclusions correspond to actual natural processes. When competition scores are tied to scarce college admission opportunities, this epistemological conflict may transform into a structural contradiction between scientific honesty and personal prospects. This paper proposes that both “in-model problem-solving” and “out-of-model review” should be incorporated into the evaluation of physics talents. Furthermore, combining international competition error-correction mechanisms, Goodhart’s Law, and the development of artificial intelligence, we suggest establishing systems for scientific dissent protection, tiered grading, and red-team auditing for exam proposition. Technical explanations regarding finite-system thermodynamics, non-thermal corrections to Hawking radiation, and the final stages of black hole evaporation are detailed in the appendix.