Your Data Is Made Powerful By Context (so stop destroying it already)

This article argues that the traditional three pillars model of observability (logs, metrics, traces) destroys the most valuable part of telemetry data: its context and relationships. It explains that data becomes exponentially more powerful with added context, enabling combinatorial insights. The author criticizes engineers for treating observability as a fire alarm rather than a continuous feedback loop, and warns that AI agents struggle with siloed data. The piece advocates for a unified approach to data collection to preserve relational seams, making it highly relevant to IT/technology, DevOps, and software engineering best practices.