Mutable Transcripts: Mitigating Context Pollution through Editable Conversation State
Dan Barry and Andrew Hines at University College Dublin (NeurIPS 2026) propose mutable transcripts, a chat interface where users revise earlier turns through natural-language edit requests so outdated context is removed from the conversation instead of piling up.
Ask this paper
Context pollution. Standard chat treats history as append-only, so corrected or abandoned instructions keep influencing later responses.
Transcript-level edits. Users ask in natural language for earlier turns to be changed, which supports retroactive corrections, new global constraints and pruning of stale context.
User study. In a controlled study with 17 participants, mutable transcripts beat standard chat on transcript cleanliness, state clarity, confidence and ease of use.
Fewer restarts. Participants reported lower intent to abandon the conversation and start a new one.
Transcript analysis. Representative conversations became shorter and no longer carried obsolete context. Prototype code is released as ReChat.
Abstract
Contemporary large language model (LLM) chat systems treat conversation history as an immutable sequence of turns that defines the model's working context. However, user intent in real interactions is not static: it evolves through correction, refinement, and shifting constraints. This mismatch between dynamic intent and static transcripts can result in context pollution, where outdated or irrelevant information persists and continues to influence subsequent responses. We introduce mutable transcripts, a new interaction paradigm that enables users to revise prior turns through natural language edit requests, allowing the conversation history itself to be updated rather than appended. This reframes the transcript from a passive record into an editable representation of conversational state. We present a working prototype that integrates transcript-level revision into a standard chat interface and evaluate its feasibility through a controlled user study (n=17) and an illustrative transcript analysis of representative interaction scenarios. Participants significantly preferred mutable transcripts over standard chat across measures of clarity, confidence, and ease of use, with reduced intent to restart conversations. Transcript analysis of representative user study conversations shows that mutable transcripts can reduce conversation length and eliminate obsolete retained context. These findings provide initial evidence that user-driven revision of conversational history can improve interaction quality and help maintain a more current representation of user intent. The source code and prototype can be accessed at https://github.com/QxLabIreland/ReChat