How Many Steps Do You Really Look Ahead?

The Chess Question

In chess, the depth of lookahead is measurable. Strong players calculate forced lines to eight or ten moves. Grandmasters can go deeper in tactical positions. Computers go deeper still.

Outside chess, the equivalent question is harder to answer, but equally important. When you decide to take a job, start a project, make an investment, or have a difficult conversation - how many steps ahead are you actually thinking?

Most people, if asked, would say three or four. In practice, the evidence suggests the answer is closer to one or two.

The Planning Fallacy as a Lookahead Problem

The planning fallacy - the tendency to underestimate how long and how much projects take - is partly a failure of lookahead. Plans that appear comprehensive at the first and second level of consequence look increasingly thin as they encounter the actual complexity of execution.

The first step seems clear. The second step follows from the first. The third step encounters the first complication that the plan did not anticipate. The fourth step is off-script entirely.

The plan was not wrong about the first step. It was wrong about what the first step would produce - what situation would result from taking it, and therefore what the second step would actually look like when you arrived at it.

Why Lookahead Degrades

Lookahead degrades because the world is not a game tree. In chess, the rules constrain the possibilities at each step - there are only so many legal moves. In real decisions, the number of possible consequences of each action is effectively unbounded, and many of the important ones are neither obvious in advance nor reliably predictable from the current state.

The deeper you look, the more your predictions depend on assumptions about the world that may not hold. The interaction effects multiply. The error bars widen. By the fifth or sixth step, you are predicting a world that depends on so many things going a particular way that the prediction is more imagination than analysis.

This is not a failure of intelligence. It is a structural feature of complex systems. The world is not obligated to produce the sequence of situations your plan requires.

What Good Lookahead Actually Looks Like

Good lookahead is not deep lookahead. It is broad lookahead at shallow depth, combined with robust adaptation for what happens beyond that depth.

This means: for the first two or three steps, plan specifically - identify the actual actions, the actual required conditions, the actual likely responses. For beyond that depth, don't plan specifically. Instead, build adaptive capacity: resources that are not committed, options that are preserved, decision points that are marked in advance so you know when new information should update the plan.

The question "how many steps am I actually looking ahead?" is useful not to make lookahead deeper but to make it more honest - to know where your plan transitions from analysis to assumption, and to treat the assumption-based portion differently from the analysis-based portion.

Knowing the limit of your lookahead is the beginning of planning around it.