Loading video...

Video Failed to Load

Go Home

Steve Ballmer reveals the interview test Microsoft used to separate problem-solvers from gamblers: "I'm thinking of a number between 1 and 100. First guess, I give you $5. Then $4, $3, $2, $1. After that, you pay me." "There are far more numbers on which you lose than win."

4,381,423 views • 4 months ago •via X (Twitter)

34 Comments

Vivek Girotra's profile picture
Vivek Girotra4 months ago

ok the video clarifies that he tells you high/low after the answer which makes this a binary search solution

John Hatcher's profile picture
John Hatcher4 months ago

The stock lost 45% of its value under his leadership. He’s an abject retard and no one should listen to him for any reason.

Doug Heinz's profile picture
Doug Heinz4 months ago

I'm forever curious by the severe lack of awareness of these chicken-shit, dumb-asses who do things like this thinking they're clever in a job interview. Ballmer is Exhibit A of this. And also, the LinkedIn influencer barnacles who those who think this is some kind of magic wizardry.

David Cooper's profile picture
David Cooper4 months ago

"Probabilistically you will lose money." "Why" Me in the background, "How do you get in a position in life to interview Steve Ballmer but you don't understand middle school math?"

Gilles Gouaillardet's profile picture
Gilles Gouaillardet4 months ago

I might play, but on the condition he writes down the number before starting the number.

Danby's profile picture
Danby4 months ago

Binary search. Gets you in the range quickly, but dialing in the final guess stacks towards the end and results negative.

John Leroy's profile picture
John Leroy4 months ago

"You will lose money" "Why?" I am guessing she claims to be educated, because she has a degree(s) in something.

boxxa's profile picture
boxxa4 months ago

It sounds stupid on the surface but there is no reward to getting the number right. Guess once, you are wrong, take your $5 and stop.

Harsh Murari 🇺🇸's profile picture
Harsh Murari 🇺🇸4 months ago

I’d actually play. The odds of winning are: 5/100+4/50+3/25+2/12.5+1/6.25= .57

Glitch Truth's profile picture
Glitch Truth4 months ago

Binary search caps at 7 guesses for 1-100. Best case +$5, worst case -$3, EV slightly negative. Ballmer wasn't testing math, he was filtering for who keeps playing when the cost flips red. Microsoft was hiring traders, not engineers.

steve's profile picture
steve4 months ago

I know guys like Ballmer well enough to know the number is 69.

0b1001011's profile picture
0b10010114 months ago

>"Software company" guy telling lady performing a textbook Binary Search is a bad idea I have just learned something about the current state of Windows.

james labastida's profile picture
james labastida4 months ago

whole EV calculation assumes uniform random selection. it’s not probability it’s game theory. it’s unverifiable. a real quant problem has specified distribution. ballmer’s house edge is undefined since he picks the number.

Eldar Aliev, EA, USTCP's profile picture
Eldar Aliev, EA, USTCP4 months ago

The answer is I don’t want to play

Ben Burban's profile picture
Ben Burban4 months ago

Problem with old heads like this is even if you guessed his number first try, he gonna change his number just to make his point 😂

Jeff Landis's profile picture
Jeff Landis4 months ago

Make one guess, collect $5 then stop. He never said you had to guess correctly.

Rod Farva's profile picture
Rod Farva4 months ago

Idiots like this who got rich because they were positioned correctly and got lucky bother me. Then they think they are geniuses.

Michael V 💎's profile picture
Michael V 💎4 months ago

Just tell me after every guess if the number is higher or lower than my last guess. It's as simple as that.

Kid Vader's profile picture
Kid Vader4 months ago

All those tests are bullshit all you learnt was Ballmer does not know if one number is higher or lower than another number. He got it wrong numerous times.

chris's profile picture
chris4 months ago

if only he’d known what the number was before he panicked and made it up at the end. Clown.

Frank Ansah's profile picture
Frank Ansah4 months ago

Here are some questions that help you solve the problem. 1. Is the number even or odd?The answer to the question eliminates 50 wrong answers. 2. Is the number greater than or less than 50? That reduces the options to 25. 1/2

Matt @ Meritum's profile picture
Matt @ Meritum4 months ago

Do binary search for 3 rounds, then always assume he would think adversarially and go to margins of probabiliyy space (i.e. no 57/58). If you still havent hit by last free guess, stop playing. He never stipulated you have to finish the game.

Craig_Shiff's profile picture
Craig_Shiff4 months ago

He clearly changed the number to prove his point, hence getting confused if it was higher or lower

danbianconyc 🇺🇦🇬🇪🇺🇸's profile picture
danbianconyc 🇺🇦🇬🇪🇺🇸4 months ago

How does someone electing the $5 or some amount higher than $0 mean they are a problem solver?

Theodore Beers's profile picture
Theodore Beers4 months ago

I haven’t watched the video – just read the quote – but aren’t there versions of this question where the idea is that you can guess, get an over/under answer, and use binary search? Takes 6 guesses on average, worst case 7. Starting at $5 is not attractive.

Keith W. Boone's profile picture
Keith W. Boone4 months ago

If he sticks to pick numbers that are hard to get with naive binary search I win with a Nash equilibrium of about a buck and a quarter in my favor. If we both go for an optimal strategy then I lose somewhere between two dimes and a quarter

John Fox's profile picture
John Fox4 months ago

clearly demonstrating that he never ever was interviewed to get a job. He was given a job by his roommate in college and that’s it.

Cyril Bèle's profile picture
Cyril Bèle4 months ago

If I have the info of higher and lower for sure I play, I will take at most 7 guess to get the number so in this case I lose 1 dollar but most of the time it won t reach the 7th turn and I will get money

Alex's profile picture
Alex4 months ago

Motherfucker need to write it down THEN maybe we play

Jim Osman's profile picture
Jim Osman4 months ago

Probability thinking quickly separates investors from gamblers.

Shawn D.'s profile picture
Shawn D.4 months ago

This is basically a probability version of “don’t confuse activity with strategy” 🔥♟️

Contrarian Matt's profile picture
Contrarian Matt4 months ago

Whether to play or not depends on whether the game is adversarial or if the number is random. If the game is adversarial and Steve is intelligent, he will avoid numbers that lie in the path of the first 5 iterations of the binary search (guessing in the middle of each range until the range goes to 1) In this case we can do an see how it plays out with an example. Guess 1 50 - wrong - lower Guess 2 25 - wrong - lower Guess 3 12 - wrong - higher Guess 4 18 - wrong - lower The search space is 13 - 17 (5 numbers) We can see we shouldn't have played because we have a 1/5 chance of a $1 gain and a 4/5 chance of some loss The if we carry on with binary search Guess 5 15 - wrong - lower Answer is now 13 or 14 Depending on whether we guess right, we will owe Steve $1 or $2 This means we will LOSE $1.50 on average If the number is randomly generated, this changes the odds somewhat as we might hit lucky on one of our guesses whilst doing the binary search. We can see how much we get on average by multiplying the amount I'll get each time (i.e. $5, $4, $3 etc.) by the chance of randomly stumbling on it, and see if it add up to more than $1.50 $5 / 100 = $0.05 $4 / 50 = $0.08 $3 / 25 = $0.12 $2 /12.5 = $0.16 $1 /6.25= $0.16 Total: $0.57 Still not enough to outweigh the risk of losing $1-$2 Of course, you could just tell Steve you like a gamble and its worth losing a few dollars for the fun of it.

arpeggi's profile picture
arpeggi4 months ago

It's just binary search

Al's profile picture
Al4 months ago

He obviously was just changing his number to prove a point. He realized lower than 57 means she'd win next guess, then changed it.

Related Videos

Steve Ballmer has a 30-second test that separates good operators from great ones: The former Microsoft CEO uses a simple game in interviews. He's thinking of a number between 1 and 100, and you have to guess it. "I'm thinking of a number between 1 and 100. You can guess. After each guess, I'll tell you whether high or low." Then he lays out the economics. Your first guess pays five dollars, then four, three, two, one, zero. And after that the cash flow reverses and you start paying him: a buck, then two, then three. Every additional move destroys value. "And the question is, do you want to play or not?" When the interviewer jumps straight in to play, Ballmer stops him: "I'm not trying to get you to do it on camera. It's actually quite a hard problem." They run it anyway. The interviewer guesses 50 (too low), 75 (too high), and zeroes in on the number in seven guesses. The outcome? "You owe me a buck." The point was never the guessing. It was the decision before the guessing. Ballmer's read on the interviewer: "I learned that... you need to step back... and really ask whether you should expect to win or lose money on this thing." Run the numbers and the deal is bad: "There are far more numbers on which you lose than numbers on which you win even with your strategy. And number two, I can pick numbers specifically that are hard for you to get." That's the whole lesson in a sentence. The losing outcomes outnumber the winning ones, and the counterparty controls the setup. No amount of in-the-moment cleverness fixes a negative-expected-value bet. The right move is to decline it. Ballmer still remembers the one candidate who saw it instantly. An interview at Purdue: "He said, 'Well, here's the answer.' Wrote it down and said, 'This is the expected value of the game.'" He didn't play. He priced it.

Business Nerd

15,252 views • 3 months ago