Decisions.

The framework

1. 01

   Learn the sound of it — the past arguing about the future.

   It rarely arrives labelled as a fallacy. It arrives as a reasonable sentence in a meeting: we have come too far to stop now; we have eighteen months in this already; let's not waste what we have spent; one more push and it turns. Every one of them argues from the past into the future, which is the single direction the logic cannot run. Teach yourself and your team to hear the grammar rather than the tone, because the tone is usually sensible and the grammar is always broken. Any case for continuing that leans on what has already gone is not an argument. It is an accounting of regret.

2. 02

   Rebuild the decision as if you inherited it this morning.

   Strip the history off it. You have just been handed this exact position — this team, this remaining scope, this market — with no invoices behind it and no author in the room. What would you spend from here, and what would it return? That is the entire decision. Count the forward cost in full: not only the remaining budget, but the people it occupies and the work it displaces. And watch the opposite error, because some ventures are genuinely front-loaded and early spend does buy later returns. The question is never whether it has cost a lot. It is whether the remaining plan clears the bar you would set for something new.

3. 03

   Authorship escalates — the spender shouldn't be the judge.

   Sunk cost is amplified by who made the original call. Barry Staw, the organisational psychologist, published Knee-deep in the Big Muddy in 1976, a study in which people who had personally made a failing investment decision then put substantially more money into it than people handed the identical failing position by someone else. The lesson is structural rather than moral, and it is not a comment on anyone's character. Do not ask the champion of a project to write its verdict. Give the continue-or-stop call to someone with no authorship, brief them on the forward case only, and let the champion argue rather than decide.

4. 04

   Make stopping cheap, or the limping never ends.

   The reason sunk costs hold is rarely arithmetic. It is that reversal costs status, and if the only person who suffers for a failed project is the one who finally says stop, everybody learns to keep it moving until it becomes somebody else's problem. So change the accounting around the decision rather than the decision itself. Announce closures with the figure written off in plain numbers and the learning stated beside it, credit the person who made the call by name, and put a scheduled re-decision on anything long-running, so that continuing becomes an active choice each quarter instead of the default that requires nobody's signature.