4.6/5 - (57 votes)

A fair coin, a casino and a game whose average payout is infinite: in 1713, a letter written to a French mathematician set a trap that probability theory has been arguing over ever since.

The game that should be priced at infinity

The rules are simple. A casino flips a fair coin until it lands on heads. The prize starts at 2 dollars and doubles every time tails appears: 2 dollars if heads comes first, 4 if tails comes once, 8 if twice, and so on.

Now ask what a fair entry price would be. Heads on the first flip has a probability of one half and pays 2. Heads on the second has a quarter and pays 4. Every round adds the same contribution to the average, and the rounds never stop. The expected winnings grow without bound, so on paper the game is worth an infinite amount of money.

The letter that started it

The problem was invented by Nicolas Bernoulli, who described it in a letter to Pierre Raymond de Montmort on September 9, 1713. It carries the name of his cousin Daniel, who analysed it in 1738 in the journal of the Imperial Academy of Science of Saint Petersburg, where he had once lived.

Daniel put his finger on the problem: “Any fairly reasonable man would sell his chance, with great pleasure, for twenty ducats.” Centuries later, Ian Hacking made the same point: “Few of us would pay even $25 to enter such a game.”

First fix: money is not worth the same to everyone

Daniel’s answer was that the value of a gain depends on who receives it. In his words, “a gain of one thousand ducats is more significant to the pauper than to a rich man though both gain the same amount.”

He modelled this with a logarithm, so each extra dollar counts for less as you get richer. With that rule, the average stops being infinite. A millionaire should pay up to about 20.88 dollars, while someone with 1,000 dollars should stop at about 10.95.

See also  Intelligence guessed 1,400 German tanks a month, math said 246

Gabriel Cramer, a mathematician from the Republic of Geneva, had reached part of the same idea in 1728, writing that “the mathematicians estimate money in proportion to its quantity, and men of good sense in proportion to the usage that they may make of it.”

Second fix: no casino has infinite money

The infinite average assumes a banker who can pay any prize. In 1754, Alexis Fontaine des Bertins pointed out that no backer has unlimited resources. With a finite bankroll, the game’s value grows only slowly.

In 1777, the Comte de Buffon worked out that after 29 rounds there would not be enough money in the Kingdom of France to cover the bet. Cap the casino’s bankroll, and the fair price drops to a modest figure.

Third fix: ignore the very unlikely

Buffon had another idea. Sensible people ignore events that are unlikely enough. He noted that a 56-year-old man does not worry about dying in the next 24 hours, a risk of about 1 in 10,189 by the mortality tables of the day.

Treat any event rarer than 1 in 10,000 as impossible, and the game is worth only about 13 dollars. Nicolas Bernoulli himself had suggested that people neglect unlikely events, an idea that returned centuries later in the prospect theory of Daniel Kahneman and Amos Tversky.

A paradox that refuses to die

None of these fixes is final. Since the game can be rewritten with faster-growing prizes, the paradox comes back for any unbounded utility function, as Menger first pointed out. The economist Paul Samuelson offered a different way out: if the game had infinite value to the player, it would have infinite cost to the casino, so nobody would ever offer it.

See also  The study that debunked hot streaks in 1985 had a hidden flaw

The same flaw haunts the old betting trick of doubling your stake after every loss. It only works if your bankroll never runs out, which is the same impossible assumption. If you enjoy the maths of gambling, the way a casino can lose on one spin but never over a year shows the other side of the same coin.

The lesson is bigger than a coin game. An average can be perfectly correct and still be a poor guide to what a real person should do. And the family behind the puzzle is the one that also gave maths one of its most famous numbers.