Exact Record Omission in Delta Attention: A Transport Criterion, Its Cost, and a Replay Certificate

RESUMO

An assistant can stop repeating a deleted statement while its recurrent memory still carries that statement's influence. We examine this distinction by saving the state difference immediately after a record, transporting this receipt through later updates, and comparing the correction with the state built from the same conversation with the record omitted. Unrolling the recurrence gives an exact criterion: transport reaches this never-stored state if and only if the additional differences created by later updates cancel after transport. We connect this criterion to eleven conditions on 40 prospectively selected synthetic Kimi Linear contexts, with 40 separate calibration contexts and paired whole-record uncertainty estimates. Attention masking removed all 40 exact greedy target answers, yet a fixed three-query candidate-scoring attack attained AUC 0.740 (95% interval 0.695--0.809); the measured false-positive rate was 7.5%. Prompt-only forgetting still returned 33 of the 40 targets. Checkpoint replay matched the complete declared active state and audit logits in all 80 contexts, while preserving all 120 retained answers in the evaluation cohort. An independent Qwen cohort and matched original-bf16/8-bit Kimi controls confirmed native transport mismatch above their empirical arithmetic floors. Replay also supported record replacement and sequential deletion with an intervening admission. These results provide a practical audit of both the requested memory change and the assistant's remaining behavior, with replay work determined by the surviving suffix.