The Dead Reckoning Problem: How to Make Good Decisions Without a Fixed Reference Point

Before GPS, sailors navigated open ocean using a technique called dead reckoning. You take your last known position, track your speed and heading, account for currents and wind drift, and calculate where you must be now. No landmarks. No satellites. Just accumulated inference, checked against whatever sparse signals the environment offers.

Photo by Ann H on Pexels.

It works remarkably well. It also compounds errors in ways that can kill you.

That tension is worth sitting with, because most serious decisions happen exactly this way. You have a last known position (some past moment when things were clear), a sense of direction, and a lot of accumulated drift you may or may not have accounted for. The question isn't whether you have perfect information. The question is how honestly you're tracking your own error accumulation.

Why Most Decisions Feel More Grounded Than They Are

Humans are terrible at noticing when their reference point has gone stale. Psychologists call this anchoring, but the dead reckoning version is subtler than the classic anchoring bias. Classic anchoring is about over-weighting an initial number. This is about over-trusting an initial model of the situation, even as the situation drifts away from it.

A manager who understood her organization's dynamics well in 2019 may still be navigating from that map. A policy analyst whose mental model of a regulatory environment was accurate three years ago is making inferences from a position that no longer exists. They feel confident because their internal model is coherent. Coherence and accuracy are not the same thing.

The longer you've been in a domain, the more likely this is to happen. Experience builds fluency. Fluency builds speed. Speed reduces the frequency with which you stop to ask whether your reference point needs updating.

Error Accumulation Is Non-Linear

Here's what makes dead reckoning genuinely dangerous: errors don't just add up, they compound. A one-degree heading error over 60 miles of ocean puts you a mile off course. Over 600 miles, it's ten miles. The same input error, multiplied by distance traveled, produces wildly different outcomes depending on how far you've gone.

Decisions work the same way. A slightly wrong assumption baked into a six-week project plan is annoying. That same assumption baked into a five-year strategy is potentially catastrophic. Time and scale amplify drift.

This is why long-horizon decisions deserve more frequent reference-point checks, not fewer. The instinct is often the opposite: once you've committed to a long-term direction, you trust the initial reasoning and resist revisiting it. But that's exactly when accumulated error is most dangerous.

graph TD
    A[Last Known Position] --> B(Estimate Current Position)
    B --> C{Check Against Available Signals?}
    C -->|Yes| D[Correct for Drift]
    C -->|No| E[Accumulated Error Grows]
    D --> B
    E --> F[/Compounded Mismatch/]

What "Fixing Your Position" Actually Looks Like

Real navigators using dead reckoning don't just calculate; they also actively hunt for position fixes. A star sighting. A depth sounding that matches a known underwater feature. A glimpse of coastline. Anything external that lets them compare their estimated position to reality and correct.

In decision-making, position fixes are the equivalent of honest feedback loops. Not the feedback you generate internally by reviewing your own reasoning, which just confirms your current model. External signals: actual outcomes, data that doesn't fit your expectations, direct input from people with different vantage points.

The critical skill is learning to treat disconfirming signals as gifts rather than noise. A sailor who ignores a depth reading that doesn't match the chart isn't being confident. They're being reckless.

Building in Drift Tolerance

Some navigators build in a deliberate bias when they're uncertain about currents. Instead of steering for the exact destination, they aim slightly upwind or upstream of it, so that when drift occurs (and it will), they're pushed toward the target rather than away from it. In decision terms: when you're uncertain about how your model might be wrong, structure choices so that common error modes push outcomes toward acceptable rather than catastrophic.

This is different from hedging everything. It's about knowing which direction you're likely to be wrong and building some tolerance for that specific failure mode into your plan.

Perfect calibration is never available. What skilled navigators develop over time is an accurate sense of their own instrument error: how much their estimates drift, under what conditions, and in which direction. That self-knowledge doesn't eliminate uncertainty. It makes them honest about the margin they're operating within.

Knowing your error budget is a form of competence. Pretending you don't have one is how ships end up on rocks.