# Context Compaction Theory

arXiv: https://arxiv.org/abs/2608.01326 · HTML: https://mlsanity.com/paper/2608.01326/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 45 references that are in the map.
- First posted on arXiv 2 Aug 2026.

## Builds on (most cited first, 20 of 20)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models](https://mlsanity.com/paper/2306.14048/) (2023)
- [Agent Laboratory: Using LLM Agents as Research Assistants](https://mlsanity.com/paper/2501.04227/) (2025)
- [LLMLingua-2: Data Distillation for Efficient and Faithful Task-Agnostic Prompt Compression](https://mlsanity.com/paper/2403.12968/) (2024)
- [Compressing Context to Enhance Inference Efficiency of Large Language Models](https://mlsanity.com/paper/2310.06201/) (2023)
- [The Case for Learned Index Structures](https://mlsanity.com/paper/1712.01208/) (2017)
- [ACON: Optimizing Context Compression for Long-horizon LLM Agents](https://mlsanity.com/paper/2510.00615/) (2025)
- [Recursive Language Models](https://mlsanity.com/paper/2512.24601/) (2025)
- [A Model for Learned Bloom Filters, and Optimizing by Sandwiching](https://mlsanity.com/paper/1901.00902/) (2019)
- [Adaptive Learned Bloom Filter (Ada-BF): Efficient Utilization of the Classifier](https://mlsanity.com/paper/1910.09131/) (2019)
- [Latent Context Compilation: Distilling Long Context into Compact Portable Memory](https://mlsanity.com/paper/2602.21221/) (2026)
- [Meta-Learning Neural Bloom Filters](https://mlsanity.com/paper/1906.04304/) (2019)
- [Fast Partitioned Learned Bloom Filter](https://mlsanity.com/paper/2306.02846/) (2023)
- [Communication Complexity (for Algorithm Designers)](https://mlsanity.com/paper/1509.06257/) (2015)
- [How to Approximate A Set Without Knowing Its Size In Advance](https://mlsanity.com/paper/1304.1188/) (2013)
- [Adversary Resilient Learned Bloom Filters](https://mlsanity.com/paper/2409.06556/) (2024)
- [Hallucination is a Consequence of Space-Optimality: A Rate-Distortion Theorem for Membership Testing](https://mlsanity.com/paper/2602.00906/) (2026)
- [On the Brittleness of LLMs: A Journey around Set Membership](https://mlsanity.com/paper/2511.12728/) (2025)

## Built on by (most cited first, 2 of 2)

- [An Architecture for Long-Horizon Agents: Levels, Ticks and Cascaded Intelligence](https://mlsanity.com/paper/2609.19519/) (2026)
- [Dude, Where's My State? Execution Information Requirements for Stateful Agents](https://mlsanity.com/paper/2609.32687/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) have a bounded context window. The context window is the maximum input size an LLM can consume for a single inference. AI agents rely on a process called context compaction to fit their state within the context window when calling an LLM. Despite its ubiquity, context compaction has received essentially no formal analysis. In this paper, we initiate a formal study of context compaction. We first introduce a framework consisting of two games that capture the two algorithmic strategies for context compaction used by contemporary AI agents in practice. The Context Selection Game models context compaction algorithms that select a subset of an agent's accumulated state to retain. The Context Generation Game models context compaction algorithms that summarize an agent's state by an arbitrary message of bounded length. We then prove an equivalence between the Context Generation Game and one-way communication complexity. The minimum context compaction budget for answering a set of queries within a target error is equal to the one-way communication complexity of the induced communication problem at the same error. Known bounds from communication complexity therefore transfer directly to context compaction. We also show that the Context Selection Game corresponds to a restricted class of one-way communication protocols. Any gap between selection and generation is therefore a gap between two classes of communication protocols. We prove that there exists a set of queries for which generation needs strictly less budget than selection. The equivalence between the Context Generation Game and one-way communication also lets us measure how well a deployed context compaction algorithm performs on a query relative to the optimal strategy. As an example, we present a case study that evaluates Anthropic's context compaction endpoint on set membership queries.