Comment on Temperature change can solve the Deutsch-Jozsa problem: An exploration of thermodynamic query complexity

We show that the readout analysis in Phys. Rev. A 113, 012420 (2026) does not establish the claimed resource accounting of one thermal query followed by many statistically useful probe samples. A CNOT fanout of a diagonal post-query probe broadcasts its one-qubit marginal but produces perfectly correlated registers rather than independent copies. Consequently, neither the trace distance nor the relative entropy between the balanced and constant hypotheses increases, and repeated measurements of the ancillas provide only one Bernoulli observation. Independent samples instead require repeated preparation and heat exchange, which are repeated queries under the definition adopted in the original paper. We also show that the sample lower bound stated there does not follow from Pinsker's inequality: the inequality gives the opposite ordering for the reciprocal relative entropy. The thermal-kickback mechanism may still encode the decision in the probe temperature, but the one-query readout claim and the quoted necessity of approximately 116 samples require correction.