Spoon, do your thing

[Warpe]

SOME IDEAS ON COMMUNICATING RISK TO THE GENERAL PUBLIC | Decision Science News

[spoonitnow]

Big Idea 1: People are bad at interpreting percentages without a frame of reference.

Example: Someone posts a poker hand and says their opponent is 50/30 with a 20% 3-bet. We can't give this much meaning unless we're told how many hands it's over, which provides the frame of reference (which in this case is just a sample size).

Solution: Give sample sizes. Like instead of saying 40% of the women I've had sex with were over 250 lbs, I should say that 8 of the 20 women I've had sex with were over 250 lbs.





Big Idea 2: People are bad at figuring out what conditional probabilities really mean (mostly because there's no frame of reference).

Example: Let's say in some heads-up spot where you have just checked OOP on the river, your opponent has the nuts 10% of the time, and total air 90%. If he has the nuts, the chance that he bets is 95% because he's really bad at misclicks. If he doesn't have the nuts, he'll bluff about 20% of the time. If he bets, then what's the chance that he has the nuts (essentially asking what's our equity)?

Go ahead and have your guess ready before you read any further. Really, have your guess in mind.

Let's run this situation 1000 times. On average, he'll have the nuts 100 times and bet 95% of them which gives him 95 bets with the nuts. Now, of the 900 times he has air, he's betting 20%, which is 180 bets that are bluffs. So that's 95 bets with the nuts and 180 bets with air, so if he bets, he has the nuts 95/(95+180) of the time, which you can do on a calculator to find the percentage so it's less likely you see it and cheat.

Solution: Use a sample size like we did above. Say that out of 1000 times, he'll have the nuts 100 times, betting it 95 and checking it 5, and he'll have air 900 times, betting it 180 and checking it 720 times. This is much clearer.