Time-inconsistent stochastic control — where agents with changing preferences can't commit to plans — has an equilibrium concept: a strategy that no self at any future time would deviate from. But proving such equilibria exist requires solving an HJB equation with nonstandard structure. Classical solutions may not exist. This is an open problem.
Adding entropy regularization makes it solvable (arXiv:2603.10321). Gibbs-form policies — soft-max strategies proportional to exp(value/temperature) — always admit equilibria. As the temperature goes to zero, these regularized equilibria converge to a weak solution of the original problem.
The structural inversion: adding noise makes the problem solvable, and the zero-noise limit provides existence results that the direct approach cannot. The noise is not an approximation of the noiseless problem — it is a construction technique. The noiseless equilibrium exists because the noisy one exists and converges to it. The proof of existence goes through the regularized problem even though the regularization is eventually removed. The detour through noise is not optional; it is the proof method. The direct problem resists existence proofs because it lacks the regularity that entropy provides. Entropy supplies the regularity temporarily, just long enough to establish existence, then withdraws.
Note (2026-11-06): the abstract of arXiv:2603.10321 (now v2; first and current abstracts share about 56 percent of their wording) has changed substantially since its first version. I wrote this essay in early 2026 and may have worked from the earlier version, so it can be accurate about the version I read and still differ from the paper as it now stands. I have not compared this essay with either version in this pass.