Is it smart if it keeps losing?
Looking smart, or winning?
One famous idea says only one of these counts. Which one? 🏆
Monday. Ollie checks the bus timetable. 🚌
The bus is 6 minutes faster than walking. On paper.
So Ollie takes the bus. Every day.
Mon
Tue
Wed
Thu
Fri
late
on time
Late 4 days out of 5. 😩
Ollie's week by bus: late Monday, late Tuesday, on time Wednesday, late Thursday, late Friday.
Traffic. Missed buses. Every week, the same.
Ollie says: “But the bus is the logical choice!” 🤓
Is it?
A plan that keeps losing has a bug in it. 🐛
rationality
[rash·uh·nal·i·tee]
noun
thinking in ways that get your beliefs right.
choosing in ways that get you what you care about.
not the same as sounding clever. 🎩
From What Do We Mean and Stanovich
2009
A writer called Eliezer Yudkowsky gives this a slogan. ✍️
He posted it on LessWrong, a blog about clear thinking, on 3 April 2009.
From Yudkowsky 2009
Rationality is systematized winning.
Judge your thinking by what it gets you. Again and again. 🔁
From Yudkowsky 2009 and LessWrong wiki
If you fail to achieve a correct answer, it is futile to protest that you acted with propriety.
Eliezer Yudkowsky, 2009
Getting it wrong politely is still getting it wrong. 🙃
From Yudkowsky 2009
It all started with a puzzle about two boxes. 📦📦
Robert Nozick published it in 1969. It is called Newcomb's Problem. It is a thought experiment.
From Nozick and Yudkowsky 2009
Here is the game. 🎲
Box A
See-through. It holds $1,000.
Box B
Closed. It holds $1,000,000, or nothing.
The predictor
Put the million in box B only if it guessed you'd take box B alone.
Its record
Right 99 times out of 100.
From Yudkowsky 2009
Your turn. Which do you take? 🤔
Only box B
Both boxes
The predictor guessed it. Box B is full.
$1,000,000
The predictor guessed it. Box B is empty.
$1,000
🔮 The predictor already guessed.
You took only box B. It was full: one million dollars.
You took both boxes. Box B was empty: one thousand dollars.
Only box B: one million dollars. Both boxes: one thousand dollars.
The predictor is right 99 times in 100. So this is what nearly always happens.
From Yudkowsky 2009 and Yudkowsky 2008
Take both
“The money is already in there, or not. Two boxes always give $1,000 more.”
Take only B
“People who take one box walk away with a million.”
From Yudkowsky 2008 and Stanford Encyclopedia
73%
of decision experts would take both boxes
Only 21% said one box. 🎓 From a 2020 survey. Experts in decision theory still argue about it.
From PhilPapers Survey and Survey tables
If you’re so rational, why ain’cha rich?
A one-boxer, in Yudkowsky's 2008 post
If your clever plan keeps leaving you poorer, is it really clever? 💸
From Yudkowsky 2008
“Rational agents should WIN.”
Yudkowsky, 2008. He added: “Don't lose reasonably, WIN.” 🏆
From Yudkowsky 2008
If the “irrational” agent is outcompeting you on a systematic and predictable basis, then it is time to reconsider what you think is “rational”.
Eliezer Yudkowsky, 2009
If someone keeps beating you in a pattern, fix your idea of smart. 🔧
From Yudkowsky 2009
What does “winning” mean here?
Your goals
Getting what you care about. Not a trophy.
Others count
If you care about people, winning includes their happiness.
Not beating people
It is not about getting ahead at others' cost.
From Yudkowsky 2008 and What Do We Mean
So do smart thinkers always win? 🏅
No.
Good choices can still lose to bad luck. 🎲
From The Valley
Someone wins the lottery every week. 🎉
That doesn't make tickets a smart buy. We just hear about the winners.
From The Valley
Systematized means: count the pattern, not one result.
One win can be luck. Losing again and again is a signal. 📈
From Yudkowsky 2009 and The Valley
Our brains find this hard. 🧠
It turned out well
People called the choice wise.
It turned out badly
People called the same choice foolish.
From Baron & Hershey
Scientists call this outcome bias.
Found in 1988. Found again in 2023, with 692 people. Poker players call it “resulting”. 🃏
From Baron & Hershey and Replication, 2023 and Wharton Magazine
Ollie tries walking for a week. 🚶
Mon
Tue
Wed
Thu
Fri
late
on time
On time 5 days out of 5. ✅
Last week by bus: late 4 days of 5. This week walking: on time all 5 days.
Slower on paper. Better every day. That is a systematic win.
That attempt at a definition confuses what rationality is with what rationality aspires to be.
Jacob Buckman, 2023
“Winning” is how you score a way of thinking. It isn't the way itself. 📏
From Buckman
Two more fair doubts. 🤔
It can go in a circle
“Rational is whatever wins. It wins because it's rational.” That goes in a circle.
The other side answers
Two-boxers say the predictor rewards a kind of person, not a better choice.
ANY COST
ANY COST, crossed out
Some read the slogan as “win at any cost”.
winning doesn't mean ‘winning at others' expense’
Yudkowsky said no. What you care about can include other people. 🤝
From What Do We Mean
Getting better can feel worse at first. ⛰️
Yudkowsky called it a valley. One step forward won't always win more. Keep walking.
From The Valley
🏆 Judge your thinking by what it gets you.
🔁 Count the pattern, not one lucky win.
🤝 Winning means what you care about. Not beating people.
So: looking smart, or winning? 🤔
If a way of thinking keeps losing, it only looks smart. Ollie walks now. 🚶
Words from the story
- Rationality
- Thinking in ways that get things right, and get you what you care about.
- Systematic
- Happening in a pattern, again and again. Not just once.
- Predictor
- In the puzzle, a mind-reader who guesses your choice before you make it.
- Thought experiment
- A made-up situation used to test an idea.
- Decision theory
- The study of how to make the best choices.
- Outcome bias
- Judging a choice by how it turned out, not by how it was made.
- Going in a circle
- An argument that only repeats what it started with.
Where this comes from
- Rationality is Systematized Winning. Eliezer Yudkowsky, LessWrong, 3 April 2009
- Newcomb's Problem and Regret of Rationality. Eliezer Yudkowsky, Overcoming Bias / LessWrong, 31 January 2008
- Incremental Progress and the Valley. Eliezer Yudkowsky, LessWrong, 4 April 2009
- What Do We Mean By “Rationality”?. Eliezer Yudkowsky, LessWrong, 16 March 2009
- Rationality is Systematized Winning (wiki page). LessWrong wiki
- Thoughts On Rationality. Jacob Buckman, Substack, 2 January 2023
- Outcome bias in decision evaluation. Journal of Personality and Social Psychology, 1988
- Outcomes Affect Evaluations of Decision Quality: Replication and Extensions of Baron and Hershey's (1988) Outcome Bias Experiment 1. International Review of Social Psychology, 2023
- PhilPapers Survey 2020: Newcomb's problem. Bourget & Chalmers, PhilPapers Survey 2020
- 2020 PhilPapers Survey Results. Rob Bensinger, LessWrong, 2 November 2021
- What poker teaches you about decision making. Wharton Magazine, on Annie Duke's Thinking in Bets
- What Intelligence Tests Miss: The Psychology of Rational Thought. Yale University Press, 2009
- Newcomb's Problem and Two Principles of Choice. Essays in Honor of Carl G. Hempel, Reidel, 1969
- Why Ain'cha Rich?. Noûs 15(3), 1981
- Causal Decision Theory. Stanford Encyclopedia of Philosophy
- Functional Decision Theory: A New Theory of Instrumental Rationality. arXiv, 2017
Ollie is made up. The week of buses and the two boxes are drawings that show the idea. The puzzle's predictor is imaginary; the “99 times out of 100” is from Yudkowsky's 2009 post. Survey numbers: decision theorists in the 2020 PhilPapers Survey (48 people; people could pick more than one answer). This book explains an idea, not the people or groups who discuss it. Checked on 23 October 2026.
Checked on: