Compressibility is not Feedback: A Random-access Gap in Causal Semantic Repair
Abstract
Semantic feedback can repair a noisy first decode without retransmitting the whole message, but current end-to-end evaluations hide where a feedback design fails. An oracle that sees both the source-side state and receiver observation may describe a useful correction with few bits, while the receiver, which sees only its own observation, may not know which correction to request. We formulate these as distinct direct and receiver-constrained coding problems, connect repair distortion to retained local utility and prove a random-access separation: one direct bit is exact although receiver utility vanishes with the number of source-conditioned alternatives. A synthetic ladder verifies the exact law. In paired populations of eight Reuters text-JSCC codecs, the trained recurrent population retains over three times more direct 16-bit repair value than the trained DeepSC population despite comparable full-repair opportunity. Nevertheless, an entropy-valid receiver audit across all eight recurrent checkpoints does not resolve against repetition, source-only or shuffled-index controls. The results show that compressible repair is an upper bound, not evidence of deployable feedback, and motivate reporting both rate curves under matched resources.
Related
- Federated Learning for Semantic Communications over Multiple Access Channels
- Rates and Nearly-Lossless State over a Multi-Access Channel with Causal State Information
- Foundations of Stochastic Lexical Calculus: Semantic Descent and Random Dynamics on Probability Simplices
- The Statistical Cost of Causal Discovery with Feedback
- Capacity of a Mixed Multiple-Access Channel with Causal Observation
BibTeX
@misc{bafarassat2026compressibility,
title = {{Compressibility is not Feedback: A Random-access Gap in Causal Semantic Repair}},
author = {Milad Bafarassat},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2609.25020},
eprint = {2609.25020},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.25020},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Compressibility is not Feedback: A Random-access Gap in Causal Semantic Repair
AU - Milad Bafarassat
PY - 2026
DA - 2026/10/09
PB - arXiv
DO - 10.48550/arxiv.2609.25020
UR - https://arxiv.org/abs/2609.25020
AB - Semantic feedback can repair a noisy first decode without retransmitting the whole message, but current end-to-end evaluations hide where a feedback design fails. An oracle that sees both the source-side state and receiver observation may describe a useful correction with few bits, while the receiver, which sees only its own observation, may not know which correction to request. We formulate these as distinct direct and receiver-constrained coding problems, connect repair distortion to retained local utility and prove a random-access separation: one direct bit is exact although receiver utility vanishes with the number of source-conditioned alternatives. A synthetic ladder verifies the exact law. In paired populations of eight Reuters text-JSCC codecs, the trained recurrent population retains over three times more direct 16-bit repair value than the trained DeepSC population despite comparable full-repair opportunity. Nevertheless, an entropy-valid receiver audit across all eight recurrent checkpoints does not resolve against repetition, source-only or shuffled-index controls. The results show that compressible repair is an upper bound, not evidence of deployable feedback, and motivate reporting both rate curves under matched resources.
ER -