What would count as explaining the arrow of time?

Choose a measurable asymmetry

Possible targets include the relaxation of a temperature difference, a stationary dissipative current, the reliability of an accumulating record, or the probability of a trajectory compared with its reversal. These are related research questions, but success on one does not automatically answer all the others.

A useful statement specifies the system boundary, preparation, controls, observation window and uncertainty. It also identifies which variables change sign under physical time reversal. Reversing a plotted sequence without reversing the appropriate preparation and driving protocol can compare two different experiments.

The distinction between typical macroscopic behaviour and every microscopic realization matters. A probability statement can explain a highly reliable direction without forbidding every exceptional history. Conversely, a monotonic quantity in a special model does not establish a universal arrow for all systems.

Compare normalized path distributions

For a finite illustration, let two possible forward records have probabilities P = (0.8, 0.2). Suppose their paired reverse records, under a separately defined reverse experiment, have probabilities R = (0.5, 0.5). Define the dimensionless log ratio σ_i = ln(P_i/R_i).

The two values are ln 1.6 ≈ 0.470004 and ln 0.4 ≈ −0.916291. Their forward average is 0.8 ln 1.6 + 0.2 ln 0.4 ≈ 0.192745. An individual negative value is compatible with a positive average.

Moreover, Σ P_ie^−σᵢ = Σ R_i = 1. Jensen's inequality then gives the nonnegative mean ⟨σ⟩ ≥ 0. This calculation assumes normalized paired distributions with matching support; the example has strictly positive probabilities everywhere.

The result is a probability identity, not yet a thermodynamic derivation. Arbitrarily choosing R does not make σ physical entropy production. The connection requires a valid reversed process and dynamical conditions linking path probabilities to heat, work and state changes.

Supply the physical bridge

Crooks' fluctuation theorem provides such a bridge for specified microscopically reversible stochastic dynamics and paired driving protocols. In a thermal setting with equilibrium initial preparations, its work relation connects the path comparison to temperature and free-energy differences. The conditions are part of the result.

A new model should reproduce an appropriate benchmark before claiming a novel deviation. It must define its initial ensemble and reversed experiment, account for hidden variables, and state whether coarse records retain enough information to evaluate the proposed path ratio.

Finite data introduce another difficulty. If a rare reverse record is not observed, its probability is not automatically zero. A log ratio estimated from small samples can become misleading. Predictive probabilities, sample sizes and an uncertainty analysis should accompany any inferred asymmetry.

Explain why the relevant preparations occur

A relaxation theorem usually begins with a preparation or boundary condition. It does not thereby explain why the universe supplies that class of preparations. Lebowitz's analysis emphasizes the role of low-entropy initial conditions and typical microscopic states in the statistical explanation of macroscopic irreversibility.

For a laboratory claim, controlled preparation may be an appropriate input. For a universal claim, the origin and scope of that input become part of the problem. A generative model must show where the restriction enters instead of hiding it in a preferred initial archive or an assumed reservoir.

A concrete SMM/DTT research target

The book's three-arrow discussion and physical-reduction conditions suggest a precise project: derive a completed-record process from source dynamics, identify its stationary reference and exchanges, then calculate forward and reverse record probabilities.

Vary the retained memory while keeping the source law fixed. Determine which apparent asymmetries disappear when hidden correlations become accessible and which persist because of preparation or environmental exchange. This would connect inner generation and outer completion to a controlled observable comparison.

The new prediction should restrict an independently measurable quantity: a relaxation rate, a memory-dependent response, or a path-probability ratio under stated controls. A successful result would explain why those restrictions follow from the source principles. The present examples organize that task; they do not claim that a universal thermodynamic arrow has already been derived.

Sources and further reading

• G. E. Crooks, Entropy production fluctuation theorem (1999).
• J. L. Lebowitz, Macroscopic laws, microscopic dynamics, time's arrow and Boltzmann's entropy, Physica A 194, 1–27 (1993).
• Continue with the published dual-time research connection and matching claims with evidence.