nucleation

LIT-327: Thermodynamics of Prediction

1 min read

Properties5
StatusActive — read in full 2026-09-29 (NOTE-295); worth reading as the short, exact source of "memory that does not predict costs dissipation" and of the nostalgia-augmented Landauer bound. Read it knowing three things. The bound prices only the *nonpredictive* part of memory. The nonnegativity of the per-step nostalgia needs a Markov drive, which the paper does not say. And nothing in it supplies a k_BT ln 2 floor per judgment.
Version2
Tagsinformation-theory, thermodynamics, natural-sciences, learning-theory
Cited as “paper” byNOTE-295 — Thermodynamics of Prediction
Cited as “source” byTHEORY-026 — For a system driven without feedback through fixed kernels, the average work dissipated at each switch of the drive equals the memory that fails to predict the next input; that is nonnegative when the drive is Markov and, summed, bounds the protocol's dissipation from below when each kernel keeps its equilibrium stationary, so there only nonpredictive memory is wasteful; with feedback, in one binary channel no predictive agent is efficient, THEORY-030 — For a memory evolving reversibly with its environment, Landauer's bound prices only steps that lose logical distinguishability, at least k ln 2 of entropy per bit lost, exported to the environment and not necessarily as heat, so copying onto a blank register, measuring or evaluating a bit has no minimum cost; learning a label has a floor only in total entropy production, at least the information learnt, which in Goldt and Seifert's perceptron can be paid in the weights' own entropy with vanishing heat

Graph View

Susanne Still, David A. Sivak, Anthony J. Bell, Gavin E. Crooks (2012), Physical Review Letters 109, 120604 — DOI-10.1103/PhysRevLett.109.120604

Standing in the record

Registered on 2026-09-29 at the owner’s request, because the owner’s prior-art novelty map (a private working document on concepts as projectors, the linear representation hypothesis and their neighbours) cites it: row 13 and §4 must-cite: “thermodynamics of learning” field that owns the physics side of the predicate-cost bridge. It is seeded from metadata and published descriptions and has not been read here.

Identification notes: Not held as an entry in nucleation; nucleation LIT-041 (Work Capacity of Channels with Memory) mentions “thermodynamics of prediction” in its body but is a different paper.

It was filed Deferred, unread. NOTE-295 is the close reading of 2026-09-29, and it placed the work: Active — worth reading as the short, exact source of “memory that does not predict costs dissipation” and of the nostalgia-augmented Landauer bound. Read it knowing three things. The bound prices only the nonpredictive part of memory. The nonnegativity of the per-step nostalgia needs a Markov drive, which the paper does not say. And nothing in it supplies a k_BT ln 2 floor per judgment.

Backlinks