02. 🎲 Decisions under Uncertainty

Started on the Sunday after the go or no-go, with the heuristics paper still open on the table. This is the map for what a number can and cannot tell me. It is one flat list because the notes do not sort into neat groups: every one of them is the same suspicion wearing a different coat.

🧭 The Atlas

• 💎 A number without its base rate is a story: ask how often this happens anyway. The definition lives at 🔤 Base rate.
• 💎 My launch estimates ignore every launch I have seen: the base rate, applied to me, in February.
• 💎 Late projects are late by a multiple and not by a margin: why padding by a week never worked.
• 💎 A significant result is not the same as a true one: the prior odds do the heavy lifting.
• 💎 An interval is more honest than a point: range first, middle second.
• 💎 A chart is an argument and can be cross-examined: five questions, asked out loud in meetings now.
• 💎 Evidence persuades when it compares: two water companies beat one famous map.
• 💎 Extreme results drift back toward the average on their own: the carrier with one terrible month.
• 💎 I fear a loss more than I want the same gain: a seed, and mostly about the fund.
• 💎 Good enough is a rational stopping rule: the note that lets me stop reading and decide.

🔗 Connected MOCs

• 01. 🔁 Systems & Feedback: where the delays come from that make the estimates wrong.
• 06. ⏳ Time, Money & Enough: good enough and enough turned out to be neighbours.
• 00. Statistics for Product Decisions I Hub: the course most of this was learned in.

❓ Open question

If I know my estimates run late by a multiple, why do I still give the optimistic one first and correct it afterwards? Knowing the bias has changed what I write in the memo. It has not yet changed the number that comes out of my mouth in the room.