Is today your lucky day? Probably not

The human brain is really bad at understanding randomness. It's even worse at creating it.

For example: in a game of Rock Paper Scissors, a man will almost always throw out a rock as his first play. And for some reason, people predictably play scissors about 5 percent less often than rock or paper. Why? No one knows — but knowing this might win you a couple of bucks.

That's the subject of a new book, Rock Breaks Scissors: A Practical Guide to Outguessing and Outwitting Almost Everybody. Author William Poundstone explains the way randomness really works, and illustrates how we can use this understanding to our advantage.

Poundstone says when people think they're creating a random pattern, they're actually behaving predictably. And when they make a bet based on a perceived pattern or a hot streak — whether at a basketball game or at the roulette table — they are just as likely to lose as to win.

Poundstone says his interest in the brain’s probability problem goes back to the dawn of the computer age, when he “heard stories about a 'mind-reading machine' that was built at Bell Labs in the 1950s.”

“You played a simple guessing game,” Poundstone explains. “You chose heads or tails, and the computer would try to predict which one you were going to choose. If the computer guessed correctly, it won a point; if you fooled the computer, you won a point.” The computer would begin to see what patterns the player had, and it invariably won.

“This thing became a sensation at Bell Labs,” Poundstone says, “because at the time it was the center of the tech world and Nobel Prize-winning scientists [were] coming to visit. They would be … challenged to play the machine and none of them could beat it.”

That's far from the only example of "random" patterns that are easily recognizable. Poundstone tells the story of mathematician Theodore Hill, who gave his students an unusual assignment: go home and toss a coin 200 times and record whether the coin comes up heads or tails. Then he added a twist: he gave the students the option of not tossing the coin and making up their own data.

“At the next class, Hill would look at the papers and he could tell who had the real random data, and who had made up what they thought was random data,” Poundstone say. “One of the things he found is that there just weren't enough long streaks of the same heads or tails when people were making up what they thought was random data.”

The same is true for teachers who create multiple choice tests. Poundstone looked at about a hundred tests from a variety of disciplines and found the “exact same pseudo-random patterns” that he sees in any human-created context.

For example, on a multiple-choice test with four choices, the odds of a given answer being correct twice in a row should be about one in four. “In other words, if the answer is B this time, you’d expect a one-in-four chance of it being B the next time,” Poundstone explains. “But actually it's a lot less than that, because people just don't want to repeat the same answer twice in a row.”

In fact, on most tests, if you look at the answer key, “it’s like they're playing hopscotch,” Poundstone says. “They just don’t like to repeat.”

By the way, he adds, if you must guess on a four-choice question, the most likely answer is B; on five-choice questions, the most likely answer is E.

Poundstone has a couple of other tips about tests. On True-False tests, True is the correct answer 56% of the time (probably because its just easier to construct a question that has True for an answer, Poundstone speculates). If a question has the choice of None of the Above or All of the Above, this answer will be right more than half the time — when statistically it should be right only about 20 percent of the time on a five-choice test.

Now run right out and buy those lottery tickets — because today’s your lucky day!

This article is based on an interview that originally aired on PRI's Science Friday with Ira Flatow.

The human brain is really bad at understanding randomness. It's even worse at creating it.

For example: in a game of Rock Paper Scissors, a man will almost always throw out a rock as his first play. And for some reason, people predictably play scissors about 5 percent less often than rock or paper. Why? No one knows — but knowing this might win you a couple of bucks.

That's the subject of a new book, Rock Breaks Scissors: A Practical Guide to Outguessing and Outwitting Almost Everybody. Author William Poundstone explains the way randomness really works, and illustrates how we can use this understanding to our advantage.

Poundstone says when people think they're creating a random pattern, they're actually behaving predictably. And when they make a bet based on a perceived pattern or a hot streak — whether at a basketball game or at the roulette table — they are just as likely to lose as to win.

Poundstone says his interest in the brain’s probability problem goes back to the dawn of the computer age, when he “heard stories about a 'mind-reading machine' that was built at Bell Labs in the 1950s.”

“You played a simple guessing game,” Poundstone explains. “You chose heads or tails, and the computer would try to predict which one you were going to choose. If the computer guessed correctly, it won a point; if you fooled the computer, you won a point.” The computer would begin to see what patterns the player had, and it invariably won.

“This thing became a sensation at Bell Labs,” Poundstone says, “because at the time it was the center of the tech world and Nobel Prize-winning scientists [were] coming to visit. They would be … challenged to play the machine and none of them could beat it.”

That's far from the only example of "random" patterns that are easily recognizable. Poundstone tells the story of mathematician Theodore Hill, who gave his students an unusual assignment: go home and toss a coin 200 times and record whether the coin comes up heads or tails. Then he added a twist: he gave the students the option of not tossing the coin and making up their own data.

“At the next class, Hill would look at the papers and he could tell who had the real random data, and who had made up what they thought was random data,” Poundstone say. “One of the things he found is that there just weren't enough long streaks of the same heads or tails when people were making up what they thought was random data.”

The same is true for teachers who create multiple choice tests. Poundstone looked at about a hundred tests from a variety of disciplines and found the “exact same pseudo-random patterns” that he sees in any human-created context.

For example, on a multiple-choice test with four choices, the odds of a given answer being correct twice in a row should be about one in four. “In other words, if the answer is B this time, you’d expect a one-in-four chance of it being B the next time,” Poundstone explains. “But actually it's a lot less than that, because people just don't want to repeat the same answer twice in a row.”

In fact, on most tests, if you look at the answer key, “it’s like they're playing hopscotch,” Poundstone says. “They just don’t like to repeat.”

By the way, he adds, if you must guess on a four-choice question, the most likely answer is B; on five-choice questions, the most likely answer is E.

Poundstone has a couple of other tips about tests. On True-False tests, True is the correct answer 56% of the time (probably because its just easier to construct a question that has True for an answer, Poundstone speculates). If a question has the choice of None of the Above or All of the Above, this answer will be right more than half the time — when statistically it should be right only about 20 percent of the time on a five-choice test.

Now run right out and buy those lottery tickets — because today’s your lucky day!

This article is based on an interview that originally aired on PRI's Science Friday with Ira Flatow.