Directory · R
Not sure this is the right entry? Run the check · New here? Start with the premise
Reality
2 min read · 447 words
Everything you think about your situation is a model of it, not the thing itself. The models are usually close enough — and they drift, which is where most navigational errors come from.
Is this you?
Check whatever lands. Nothing is sent anywhere.
Reality is the territory the system is operating in, distinct from the maps the mind keeps building of it.
The mind constructs models. The model of the situation. The model of the relationship. The model of the operator’s own capacities. The model of what other operators are thinking. None of these are reality. They are the system’s interpretations, useful for navigation, accurate sometimes, inaccurate other times. Reality is what is actually present, regardless of what the model says about it. The two correspond often. They diverge enough that the operator cannot rely on the model to be accurate without checking.
The cost of confusing model with reality is large. The operator who proceeds from a model of the relationship that no longer matches the actual relationship makes choices the actual relationship doesn’t reward. The operator running on a model of their own competence that exceeds their actual current capacity over-commits and underdelivers. The operator running on a model of the situation that is two years old responds to conditions that have changed without updating. Each of these is a navigational error caused by the map having drifted from the territory.
The mechanism that produces drift: the model is updated more slowly than reality changes, especially when the operator has emotional investment in the current model. The relationship the operator wants to continue producing the rewards it once produced gets modeled as still producing them, even when it isn’t. The self-image the operator wants to maintain gets preserved in the model, even when actual outputs have changed. The world the operator believes they live in stays in the model, even when the world has moved on.
From the chair: periodically test the model against reality. The discipline is uncomfortable because the test sometimes returns the model is wrong, and updating is harder than continuing. But the cost of operating from a wrong model accumulates faster than the discomfort of updating it. The operator who runs the test regularly catches drift early and updates incrementally. The operator who never runs it discovers years later that the model and the territory diverged so far that recovery requires major reconstruction.
The diagnostic questions: what is actually happening, distinct from what I expected to be happening. What outputs am I actually producing, distinct from what I think I’m producing. What is the other operator actually doing, distinct from what my model says they’re doing. The honest answer to any of these often surfaces drift the operator was not consciously tracking.
The map is not the territory. Update the map. The territory will not change to match it.
The map and the territory drift apart
Your mind constructs models continuously — of the situation, the relationship, your own capacities, what other people are thinking. None of those are reality. They're interpretations, useful for navigation, accurate sometimes and inaccurate other times. They correspond often enough that you stop checking, and they diverge often enough that not checking is costly.
The errors are navigational rather than philosophical. Proceeding from a model of a relationship that no longer matches the relationship produces choices it doesn't reward. Running on a model of your competence above your current capacity produces over-commitment. Responding to conditions as they were two years ago produces reasonable-looking decisions aimed at a situation that's gone.
What makes this hard is that models don't announce their drift. They feel like perception rather than interpretation, so there's no internal signal that says this map is three years old. The only thing that catches it is deliberate checking against something external.
Which gives you the practical discipline, and it's unglamorous: ask rather than assume, look at current evidence rather than remembered evidence, and treat being surprised as a prompt to inspect the model rather than as a fact about the world being unpredictable.
Test one model
Pick something you're confident about regarding another person or a situation. Confidence is the signal that it hasn't been checked.
Real questions
How do I know if my model of a situation is out of date?
Ask when you last got direct evidence for it rather than inferred it. A surprising number of strongly-held beliefs about people and situations trace back years, or to nothing identifiable at all. The other reliable signal is surprise: being caught out by something several other people saw coming almost always means the map drifted rather than the world behaving unpredictably.
Why do I assume I know what people are thinking?
Because you can't access anyone's interior, so your mind fills the gap with a model — and the model feels like perception rather than construction, which is why it doesn't come with a confidence warning. It's built from their behaviour filtered through your history and assumptions. That's useful for navigation and routinely wrong in specifics, which is why one awkward question frequently resolves what a week of inference couldn't.
Isn't everything just interpretation?
The models are interpretations; the territory isn't. That distinction is the practical point rather than a philosophical one — some models match reality better than others, and the ones that match produce better decisions. Treating all readings as equally valid removes the only move available, which is checking. The useful stance isn't certainty or relativism, it's holding the model lightly and testing it.
Why do I defend my model instead of testing it?
Because a sufficiently established model becomes part of the story you hold about yourself, and then contradicting evidence fires the identity defence rather than being evaluated. That's how people end up arguing for a position after they've stopped believing it. The tell is intensity: if being wrong about a factual matter feels disproportionately bad, the model has fused with something it shouldn't have.
How often should I be checking my models?
Not constantly, which would be paralysing. The practical rule is checking when the stakes are high, when the model is old, or when you've just been surprised — those three cover most of the expensive errors. And it's worth noticing that confidence isn't a signal of accuracy here. The models you're surest about are frequently just the ones you've gone longest without testing.
What's the most common way the map drifts?
People changing while your model of them doesn't — which is the mirror image of the complaint that others won't let you move on from who you used to be. Capacity drifts too, in both directions: running on a model of your competence from a better or worse period produces over-commitment or unnecessary declining. Both are undramatic and both compound quietly over years.