The Triangular Trade: What Navigation Can Teach Us About Data Redundancy

I’ve been reading about the age of sail, about ships navigating vast, featureless oceans with little more than a compass, a sextant, and a healthy dose of fear. One detail, in particular, stuck with me, not for its historical significance but for its startling relevance to the work we do every day. It’s the concept of the ‘triangle of error’—a navigator’s method of reckoning position when celestial observations are unreliable. This practice offers a profound analogy for how we should think about redundancy in our systems.

When a navigator takes a sighting on the sun or a star, they calculate a single line of position. But a single line on a chart is not a location; it’s merely a suggestion. You are somewhere along that line. To fix their position with confidence, a navigator needs at least two such lines, and where they intersect is the assumed location. The wise navigator, however, seeks a third. Three sightings create a small triangle on the chart, not a perfect point. This triangle isn't a sign of failure; it is the map of certainty. It defines the area of uncertainty, and within it lies the truth.

The False Point of Perfect Intersection

We, as engineers, are obsessed with a single point of truth. Our monitoring tells us a server is down. Our logs show an error. We pinpoint the ‘root cause’ and declare the case closed. This is like trusting a single line of position on the open sea. It feels definitive, but it is dangerously brittle. A single source of information, no matter how authoritative it seems, is just one perspective. It could be a misconfiguration in the monitoring agent, a network blip, or a flawed log entry. Relying on it is a gamble.

The navigator’s triangle teaches us to embrace the ambiguity inherent in any measurement. Instead of seeking one perfect answer, we should triangulate from multiple, independent sources. Does the application metric for failed requests spike at the same time as the load balancer health check fails and the database connection pool empties? The overlapping area of these three signals—the triangle of error for our outage—gives us a far more reliable picture than any one of them alone.

This principle extends beautifully to data backups, which is where the analogy truly shines. Many of us operate under the assumption that a backup is a binary state: it either exists or it doesn't. But a backup, like a celestial sighting, is only as good as its verifiability. Having a backup on disk, a second on an off-site tape, and a third snapshot in an immutable cloud object store is our ‘triangle of error’ for data integrity. The failure of one medium, the corruption of one dataset, or the accidental deletion of one repository is just a single line on the chart. It’s a problem, but it doesn’t mean we are lost. The other two points of our triangle keep us firmly located in the realm of recoverability.

We chase precision in our digital worlds, but the wisdom from an era of wooden ships and iron men reminds us that true reliability isn't born from a single point of perfect data. It emerges from the careful overlap of multiple, imperfect perspectives. Our systems are not points on a grid; they are dynamic, complex entities. By adopting the navigator's mindset, we learn to navigate their complexity not by seeking a single, deceptive source of truth, but by mapping the contours of our confidence. The triangle of error isn't a sign of weakness. It is the very shape of resilience.

Notes & further reading

A few pages I came back to while writing this: