Key Principle
Every decision is already a bet on a particular future, so before placing it, scout the futures it could produce. Enumerate the branching set of possible outcomes, attach probabilities to each, and weigh probability against desirability — then choose actions that raise the odds of the desirable branches. The model is BELIEF -> BET -> SET OF OUTCOMES: a bet never produces one result, it produces a distribution. Refusing to assign probabilities does not escape the prediction; it just hides it. Two complementary moves do the scouting by working backward from an imagined endpoint: backcasting imagines the goal achieved and reveals the positive space; the premortem imagines the goal missed and reveals the negative space.
Why This Matters
Without reconnaissance you act as if your expected outcome is certain (an implicit 0%/100% bet), so you do not prepare for the other branches and get surprised rather than ready. Working backward beats forecasting forward because the present looms huge while the distant future blurs (the New Yorker "View of the World" cover), and forward forecasting falsely assumes today's conditions hold. Imagining the future recruits the same brain pathways as remembering the past, and imagining an event has already occurred ("prospective hindsight") increases the ability to correctly identify reasons for a future outcome by 30% (Mitchell, Russo & Pennington, 1989). Crucially, memorializing the branch set in advance inoculates the decision against resulting and hindsight bias: a written record of the tree prevents a bad outcome from masquerading as a bad decision, and a known outcome from masquerading as "everyone should have known."
Good Examples
- Pete Carroll, Super Bowl XLIX. Seahawks down four, 20 seconds, one time-out, second-and-goal at the one. Intuition condemns the pass (interception risk), but mapping the tree reveals a hidden asymmetry: an incomplete pass stops the clock for free, buying roughly three plays to score versus two for a run, while the interception risk (2-3%) barely exceeds the run's fumble risk (1-2%). The pass was intercepted and the call widely condemned — a textbook resulting error against a process-sound decision. "The option of the extra play doesn't reveal itself without this kind of scouting of the future."
- ASAS expected-value case. After-School All-Stars re-sorted its grant pipeline by expected value, not headline amount: $100K x 25% = $25K; $200K x 10% = $20K; $50K x 70% = $35K. The smallest grant is the most valuable. They also closed the feedback loop, reviewing wins as well as rejections to sharpen future estimates.
- Backcasting a market-share goal. A firm aiming to double share from 5% to 10% in three years imagines the headline "Company X Has Doubled Its Market Share" and works back to the strategies that produced it — exposing when the goal secretly depends on low-probability events that must occur (a built-in feasibility check).
- Olmsted's Central Park was designed from "what it would develop into," not from the barren 1858 land.
Counterpoints
- People refuse to guess because they "can't be certain" — but that inverts the purpose. We do reconnaissance because we are uncertain. "Anything that moves us off those extremes is going to be a more reasonable assessment than not trying at all"; even "somewhere between 20% and 80%" beats refusing.
- Backcasting alone is insufficient. Optimism is the default, so working back only from success leaves the negative space unmapped — this is temporal discounting (feeling good now, worse decisions later). The premortem is the mandatory complement; "a little naysaying goes a long way."
- Pure positive visualization backfires. This is Oettingen's mental contrasting: pairing the desired future with obstacles raises odds, while positive fantasy de-energizes. Dieters with strong positive slimming fantasies "lost twenty-four pounds less."
- Solo scenario planning is weaker; one mind misses branches. Use the switch-sides technique — have members who disagree argue each other's position to surface hidden private information.
Key Quotes
"If a decision is a bet on a particular future based on our beliefs, then before we place a bet we should consider in detail what those possible futures might look like." — Annie Duke, Chapter 6
"Figure out the possibilities, then take a stab at the probabilities." — Annie Duke, Chapter 6
"If we're worried about guessing, we're already guessing." — Annie Duke, Chapter 6
"When it comes to advance thinking, standing at the end and looking backward is much more effective than looking forward from the beginning." — Annie Duke, Chapter 6
"Backcasting reveals the positive space. Premortems reveal the negative space. Backcasting is the cheerleader; a premortem is the heckler in the audience." — Annie Duke, Chapter 6
"Forgetting about an unrealized future can be dangerous to good decision-making." — Annie Duke, Chapter 6
Rules of Thumb
- Treat every decision as a bet: name the belief, then list the set of outcomes it could produce with probabilities. Don't stop at one expected outcome.
- Figure out possibilities first, probabilities second — and never let "I can't be certain" become "I won't estimate." Any move off 0%/100% is progress.
- Rank options by expected value (probability x payoff), not by headline payoff. The biggest number is often not the best bet.
- Plan further out than feels necessary; anticipate responses to your responses. Expert decision-makers go deeper into the tree.
- Backcast: hold up "We Achieved Our Goal!" and work back to find the path — and the low-probability events the goal secretly requires.
- Always run a premortem too: assume "We failed. Why?" and work back. Frame the plan as already failed so people can voice reservations without seeming like naysayers — reward the most productive heckler, not the loudest cheerleader.
- Because all futures sum to 100%, mapping the negative space forces the inflated positive space down to a realistic size and enables contingency planning.
- Write down the branch set before the outcome is known. Refuse to discard the work on futures that didn't happen — an unrealized future still carries decision information.
- When judging a past decision by its result, or whether someone "should have known," redraw the tree as it looked before the chainsaw of the advancing present cut off the branches that didn't materialize.
Related References
- Mental Time Travel: Beating Tilt & the Present - the in-the-moment companion tools
- Resulting & Hindsight Bias - what these tools inoculate against
Diagram
