Say someone finds your bank card, and the machine gives them three tries at your four-digit PIN. Each digit can be any of ten, so there are 10,000 possible codes.
If every code were equally likely, each guess would be right 1 time in 10,000, and three guesses, 3 times in 10,000.
How people really choose
Not at random. One study looked at the four-digit codes people chose as passwords, almost 3.4 million of them.
The picture scales them down to 1,000 people, one dot each.
1,000 people’s codes
Which three codes would you try first?
The 20 codes people chose most, in numerical order. Pick 3.
You picked the three codes people chose most. Nearly 1 person in 5 chose one of them, so your guesses would be right nearly 1 time in 5.
That’s more than 600 times as often as three guesses at equally likely codes.
Your three include 1234, the code people chose most: more than 1 person in 10 chose it. So your guesses would be right more than 1 time in 10.
That’s hundreds of times as often as three guesses at equally likely codes.
Your three would be right more than 1 time in 20, mostly thanks to 1111. But the code people chose most, 1234, isn’t among your three: more than 1 person in 10 chose it.
Even so, your three would be right hundreds of times as often as three guesses at equally likely codes.
Your three would be right fewer than 1 time in 25. Neither of the two codes people chose most, 1234 and 1111, is among them, and about 1 person in 6 chose one of those two.
Even so, your three would be right at least 25 times as often as three guesses at equally likely codes.
Where your total comes from
Each code’s share of the dots is the proportion of people who chose it, and it approximates the probability that a person chooses that code. That’s the Empirical Method.
One person has one code, so no dot can match two of your guesses. No two of your guesses can both be right: they’re mutually exclusive.
So the probability that one of them is right is the sum of their probabilities. That’s the Addition Rule for Mutually Exclusive Events.
What it adds up to
So could someone guess your PIN in three tries? More easily than 10,000 codes make it sound. The three codes people chose most, 1234, 1111 and 0000, would be right nearly 1 time in 5.
Ten thousand possible codes means 1 chance in 10,000 only if every code is equally likely, and people don’t choose that way.
