# Memory Is a Derivation: The Distributed-Evidence Paradox in Long-Term Agents

arXiv: https://arxiv.org/abs/2609.36130 · HTML: https://mlsanity.com/paper/2609.36130/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Agent memory systems, one of 16 directions in LLM agents and tool use.
- Builds on 25 of its 32 references that are in the map.
- First posted on arXiv 28 Sep 2026.

## Placed in

- LLM agents and tool use / [Agent memory systems](https://mlsanity.com/agents/agent-memory-systems/)

## Builds on (most cited first, 25 of 25)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [MemGPT: Towards LLMs as Operating Systems](https://mlsanity.com/paper/2310.08560/) (2023)
- [FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation](https://mlsanity.com/paper/2305.14251/) (2023)
- [A-MEM: Agentic Memory for LLM Agents](https://mlsanity.com/paper/2502.12110/) (2025)
- [On Faithfulness and Factuality in Abstractive Summarization](https://mlsanity.com/paper/2005.00661/) (2020)
- [ExpeL: LLM Agents Are Experiential Learners](https://mlsanity.com/paper/2308.10144/) (2023)
- [Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory](https://mlsanity.com/paper/2504.19413/) (2025)
- [LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory](https://mlsanity.com/paper/2410.10813/) (2024)
- [A Survey on the Memory Mechanism of Large Language Model based Agents](https://mlsanity.com/paper/2404.13501/) (2024)
- [MemoryBank: Enhancing Large Language Models with Long-Term Memory](https://mlsanity.com/paper/2305.10250/) (2023)
- [Understanding Factuality in Abstractive Summarization with FRANK: A Benchmark for Factuality Metrics](https://mlsanity.com/paper/2104.13346/) (2021)
- [Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions](https://mlsanity.com/paper/2507.05257/) (2025)
- [Evaluating Factuality in Generation with Dependency-level Entailment](https://mlsanity.com/paper/2010.05478/) (2020)
- [ChatGPT as a Factual Inconsistency Evaluator for Text Summarization](https://mlsanity.com/paper/2303.15621/) (2023)
- [HaluMem: Evaluating Hallucinations in Memory Systems of Agents](https://mlsanity.com/paper/2511.03506/) (2025)
- [FactGraph: Evaluating Factuality in Summarization with Semantic Graph Representations](https://mlsanity.com/paper/2204.06508/) (2022)
- [Localizing Factual Inconsistencies in Attributable Text Generation](https://mlsanity.com/paper/2410.07473/) (2024)
- [MemTxn: A Transaction Boundary for Source-Supported Updates and Complete-State Recovery in Agent Memory](https://mlsanity.com/paper/2607.27834/) (2026)
- [ConsistencyGate: Preventing Memory Contamination in LLM Agents via Self-Consistency Admission Control](https://mlsanity.com/paper/2607.22962/) (2026)
- [Mitigating Provenance-Role Collapse in Long-Term Agents via Typed Memory Representation](https://mlsanity.com/paper/2605.25869/) (2026)
- [Eywa: Provenance-Grounded Long-Term Memory for AI Agents](https://mlsanity.com/paper/2605.30771/) (2026)
- [SUCEA: Reasoning-Intensive Retrieval for Adversarial Fact-checking through Claim Decomposition and Editing](https://mlsanity.com/paper/2506.04583/) (2025)
- [VISTA: Verification In Sequential Turn-based Assessment](https://mlsanity.com/paper/2510.27052/) (2025)

## Abstract (arXiv)

Long-running LLM agents compress past interactions into persistent memories that may be reused as premises for later tasks. This creates a distinct derivation problem: whether the memory actually follows from what the interaction history supports. Relevant evidence may be scattered across earlier interactions, while compression can introduce relations or event status that the history never established. A valid memory may therefore appear unsupported because its citations omit relevant evidence, while individually supported facts may be composed into a stronger statement the history never established. We characterize this problem through three coupled requirements: (1) Evidence scope; (2) Compositional validity; (3) Admission reliability. We therefore ask whether the interaction history available at write time supports what enters persistent memory. We introduce DerivAudit, a framework for auditing whether a memory is actually supported by the history available when it was written. The audit separates three questions: whether supporting evidence lies beyond writer-provided citations, whether the composed memory introduces unsupported meaning, and how write-time admission decisions affect later memory use. Across two natural memory corpora, audits using broader pre-write history recover support for nearly 60% of memories that appear unsupported from citations alone, while 17-21% remain unsupported after expansion. Yet broader evidence does not by itself make admission reliable: unsupported memories are still frequently admitted across verification models, and evidence expansion alone worsens it on two backbones.