Context Cartography: Toward Structured Governance of Contextual Space in Large Language Model Systems

2026-03-20

AI Summary

This work addresses the performance bottlenecks—such as the “lost-in-the-middle” phenomenon—that arise from the indiscriminate expansion of context windows in current large language models, which overlook structural gradients, saliency asymmetry, and entropy accumulation within long contexts. To this end, we propose Context Cartography, a novel framework that formally partitions the context space into three regions: Black Fog (unobserved), Gray Fog (memory storage), and the Visible Field (active reasoning). We introduce seven cartographic operators to govern information flow across these regions. Grounded in the geometry of attentional saliency and information-theoretic partitioning, our framework systematically uncovers structural principles underlying effective context management. Cross-system analyses of Claude Code, Letta, MemOS, and OpenViking demonstrate the convergence of these operators in real-world implementations and provide testable ablation hypotheses alongside diagnostic benchmarks.

Technology Category

Planning, Routing, and Scheduling: Planning with Language ModelsMachine Learning: Large Multimodal Models (LMMs)Computer Vision: Language and Vision

Application Category

Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphsSemantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMsSearch and Retrieval-Augmented AI: Personalized, context-aware and across-device search

Abstract

The prevailing approach to improving large language model (LLM) reasoning has centered on expanding context windows, implicitly assuming that more tokens yield better performance. However, empirical evidence - including the "lost in the middle" effect and long-distance relational degradation - demonstrates that contextual space exhibits structural gradients, salience asymmetries, and entropy accumulation under transformer architectures. We introduce Context Cartography, a formal framework for the deliberate governance of contextual space. We define a tripartite zonal model partitioning the informational universe into black fog (unobserved), gray fog (stored memory), and the visible field (active reasoning surface), and formalize seven cartographic operators - reconnaissance, selection, simplification, aggregation, projection, displacement, and layering - as transformations governing information transitions between and within zones. The operators are derived from a systematic coverage analysis of all non-trivial zone transformations and are organized by transformation type (what the operator does) and zone scope (where it applies). We ground the framework in the salience geometry of transformer attention, characterizing cartographic operators as necessary compensations for linear prefix memory, append-only state, and entropy accumulation under expanding context. An analysis of four contemporary systems (Claude Code, Letta, MemOS, and OpenViking) provides interpretive evidence that these operators are converging independently across the industry. We derive testable predictions from the framework - including operator-specific ablation hypotheses - and propose a diagnostic benchmark for empirical validation.

Problem

• context governance
• contextual space
• large language models
• attention mechanisms
• information salience

Innovation

• Context Cartography
• contextual space governance
• cartographic operators
• transformer attention salience
• zonal information model