The Cartographer's Compromise: On Drawing the Map You Can Actually Read

There’s a moment in every cartographer’s process, I imagine, where the sheer volume of available information becomes a liability. They could map every single pebble, every subtle change in the soil’s composition, every individual pine needle on a forest floor. The resulting document would be a perfect, one-to-one replica of the territory—and utterly useless to anyone trying to find their way through it. The true skill lies not in recording everything, but in choosing what to leave out. This is the cartographer’s compromise, and it’s a discipline we in operations would do well to adopt.

We are, in our own way, cartographers of complex systems. Our logs, metrics, and events are the topography we are tasked with charting. The temptation, especially when a system is new and its failure modes unknown, is to log everything. We set our verbosity to DEBUG, capture every SQL query, every intermediate calculation, every state change. We tell ourselves we are building the definitive map, the one that will have the answer when the inevitable storm hits. But like the map that includes every pebble, this log volume quickly becomes an unnavigable wilderness. The signal we desperately need is buried beneath an avalanche of noise.

The Art of the Legend

A great map has a legend. It defines, with deliberate clarity, what its symbols mean. A squiggly blue line is always a river. A dashed black line is always a trail. This consistency is its power. Our logging should aspire to the same rigor. What does an ERROR-level entry truly mean? It should signify a clear, actionable failure, not just a minor annoyance. A WARNING should be a genuine cause for alertness, not a routine informational message we’ve grown accustomed to ignoring. By defining our log levels with the precision of a cartographer’s legend, we give our tools—and our tired eyes at 3 AM—a fighting chance to parse the landscape.

Furthermore, a map is drawn for a specific purpose. A road atlas doesn’t show mineral deposits, and a geological survey doesn’t mark the best roadside diners. Our monitoring should be similarly intentional. Is this dashboard for diagnosing a sudden service interruption, or for tracking long-term performance trends? Trying to make one view serve both purposes often results in a map that serves neither. We must have the courage to create specialized views, to leave out the data that isn’t relevant to the journey at hand.

Embracing this compromise feels counterintuitive. It feels like willingly discarding knowledge. But the cartographer knows that a map’s value isn’t in its exhaustive detail; it’s in its ability to guide someone, efficiently and reliably, from one point to another. Our systems are the territory, vast and complex. Our logs and dashboards are the maps. Let’s strive to draw them not with the anxiety of someone who fears missing a detail, but with the confidence of someone who knows what truly matters for the journey.

Notes & further reading

A few pages I came back to while writing this: