Decisions.

The framework

1. 01

   Square for your move, circle for the world's.

   Draw the decision left to right. A square marks every point where you choose; a circle marks every point where something outside you resolves — a customer answers, a rival moves, a regulator rules. The discipline is entirely in the sorting, because most people draw circles as squares and squares as circles: they treat a competitor's response as something they can plan away, and treat their own next move as weather. This is also what separates the tree from the scenario planning of Day 63. Scenarios rehearse futures you do not control; the tree interleaves those futures with the moves you do.

2. 02

   Two moves out, then stop: four moves deep and the tree is wallpaper.

   Almost all the value sits in the second layer: writing down what you would do under each reply, before the reply arrives. Past that, precision decays faster than the branches multiply — every contingency, forty endpoints, a week of work, rigorous-looking, read exactly once. So cap it deliberately. Two or three options at move one, two or three replies to each, and one written decision under every reply. If a branch later deserves a tree of its own, it will still be sitting there when you reach it.

3. 03

   Value the ends, then roll backwards — the real argument lives at the far right.

   This is Day 56's arithmetic, run at every circle rather than once at the end. A tree is read right to left. Put a number on each endpoint in one unit you actually care about — money, months, market position — then work leftwards, folding each circle into the rough worth of everything beyond it and each square into the branch you would truly take. What this exposes is where the disagreement sits. Teams argue for an hour about move one when the entire dispute is a single endpoint two people value differently by a factor of five. Mark every endpoint solid or guessed, and take the argument to the guessed one.

4. 04

   Read the tree for option value, not just for the highest number.

   The tree's most useful output is rarely the highest-value path; it is the picture of which first move leaves the most branches alive two steps later. A slightly worse expected outcome that keeps three options open often beats a better one that collapses to a single road, because your estimates are wrong and open branches are what let you be wrong cheaply. Hunt for the one-way branch: the hire, the contract, the public commitment that deletes the rest of the diagram. That is Day 64's door, seen from inside the tree — check you are walking through it for a reason worth the deletion. Then look left of everything: the small test that converts a guessed endpoint into a known one is usually the best move on the page.