✶Explainer04:00
A decision is choosing which futures stay available to you
Duke defines a decision as a choice about which path to take, where the future branches into many possibilities. Choosing one road makes certain futures available and eliminates others. Critically, staying on your current path is also a decision, so the difference between commission and omission is just whether you actively enact change.
- The future branches; a decision picks which branch you make available
- Choosing one option rejects the futures the other options would have opened
- Not changing anything is still a decision (omission vs commission)
- Every choice carries the cost of the paths it forecloses
“staying on the path that you're already in is also a decision”
#decision-making#mental-models#optionality
✶Explainer27:00
The reversed-handlebar bike: why you must keep checking your gut
Duke uses learning to ride a bike as an analogy for how deliberate practice becomes automatic gut behavior. But if someone hands you a bike whose handlebars steer backwards, your gut makes you fall. Environments and markets change the same way, so even a well-honed gut needs periodic re-examination, and sometimes a better method exists even when nothing has changed.
- Deliberate skill eventually becomes effortless gut behavior, like cycling
- When the world changes the rules, gut habits suddenly fail
- Markets and environments shift, so gut calls need re-checking
- A better strategy may exist even when the world hasn't changed
“what happens when someone hands you a bike where now when you turn the handles it goes to the right it goes left”
#intuition#learning#adaptation#mental-models
✶Explainer42:30
We misremember what we knew at the time we decided
Beyond over-indexing on outcomes, Duke notes people rewrite their own past knowledge, insisting 'I knew that would happen.' She recounts pitching the Clinton analysis to a newspaper editor who refused it because he 'already knew,' despite never having published it. This corrupted memory of prior beliefs makes it nearly impossible to build fidelity into a decision-review process.
- Hindsight makes people believe they foresaw outcomes they didn't
- An editor claimed prior knowledge yet had published nothing
- You can't review a decision honestly if you can't recall what you actually believed
- This is why recording forecasts at decision time matters
“literally they're misremembering what their state of knowledge is”
#hindsight-bias#memory#decision-review
✶Explainer48:00
When to move fast: options expire and information is the reward
Duke explains that going slow has two hidden costs. Options expire: waiting to ramp up PPE production, or to invest before a funding round closes, can remove a good solution entirely. And moving fast on many small, reversible actions generates information about how the world and you respond, building better models for the genuinely big decisions. The core rule: if the downside isn't that bad, go fast.
- Slowness can let a good option disappear entirely, as with early COVID PPE
- Some preferences can only be learned by trying, not by researching
- Fast, cheap actions build better models of the world for big decisions
- The governing question is simply: how bad is the bad?
“if the bad isn't that bad it doesn't really matter you should probably just like move fast and break things”
#speed#opportunity-cost#optionality#decision-making
✶Explainer74:00
Why 'a real possibility' is a useless, unaccountable phrase
Duke attacks vague probability language. A Mauboussin survey found people hearing 'a real possibility' interpret it as anywhere from 20% to 80%. Such words fake agreement and let people weasel out of accountability later, since the target was never pinned down. She insists on numeric probabilities with lower and upper bounds, and on defining loaded words like 'spectacular' before treating them as shared.
- 'A real possibility' means 20% to 80% to different listeners
- Vague terms create the illusion of agreement that isn't there
- Weasel words let people dodge accountability for their forecasts
- Demand numeric probabilities and bounded ranges instead
“somewhere between 20 and 80 of the time ... it's like they're totally useless words”
#communication#probability#forecasting#precision