4.5/5 - (50 votes)

Pick any whole number. If it is even, cut it in half. If it is odd, triple it and add one. Keep going, and you will land on 1, every single time anyone has tried. Yet nobody on Earth can prove it always works.

This is the Collatz conjecture, a puzzle the German mathematician Lothar Collatz introduced in 1937, just two years after earning his doctorate. It fits in two sentences, and it remains one of the most famous unsolved problems in mathematics.

A game you can play on the back of a receipt

Try it with 12. Halve it to get 6, then 3. Three is odd, so triple it and add one: 10. Then 5, then 16, 8, 4, 2 and finally 1. Nine steps, done.

Once you reach 1, the rule simply loops you through 4 and 2 forever. That tiny loop is the only one anyone has ever found. The conjecture claims every starting number eventually falls into it, with no exceptions and no escape routes.

The number 27 refuses to go quietly

Small numbers usually behave. Then you try 27.

Instead of drifting down, it shoots upward. It takes 111 steps to reach 1, and along the way it climbs as high as 9,232, more than 340 times its starting value. Then it collapses all the way back down.

That wild up-and-down pattern earned these sequences a poetic nickname: hailstone numbers, because the values rise and fall like hailstones inside a cloud. Some mathematicians simply call them wondrous numbers.

And 27 is not even the record holder. Below 1,000, the number 871 needs 178 steps. Below one trillion, the champion, 989,345,275,647, needs 1,348.

See also  The extra zero at the end of a number is not just decoration

Checking sextillions of numbers still proves nothing

Here is where intuition goes wrong. Computers have tested every starting number up to roughly 2.36 sextillion, a 22-digit figure, according to a 2025 paper by David Barina. Every one of them ends at 1.

Surely that settles it? It does not. A single counterexample could be hiding far beyond anything a machine can reach. Mathematics has been burned before: the Pólya and Mertens conjectures both looked solid after enormous checks, and both turned out to be false.

There is also a tempting shortcut argument. Odd numbers grow, even numbers shrink, and on average each odd value in a sequence ends up about three quarters of the previous one. So the numbers should drift downward. It is a convincing hunch, but it quietly assumes each step behaves independently of the last, and nobody has ever proven that.

A puzzle with far too many names

As the problem spread, it collected aliases like a fugitive. Helmut Hasse called it the Syracuse problem in the 1950s, after a visit to Syracuse University. Others named it after Stanisław Ulam, Shizuo Kakutani or Bryan Thwaites. Many simply call it the 3x+1 problem.

The legendary Paul Erdős gave it perhaps its most famous verdict: “Mathematics may not be ready for such problems.” In 2010, Jeffrey Lagarias, one of its leading experts, went further and called it “an extraordinarily difficult problem, completely out of reach of present day mathematics.”

Getting closer without crossing the line

Progress has come in careful slices. In 1976, Riho Terras showed that almost every number eventually dips below its own starting value. Others have ruled out whole families of hidden loops, one by one, with new limits pushed as recently as 2023.

See also  You never needed long division to test if a number divides by 7

Then, in 2019, Terence Tao proved that almost all sequences fall to values that are tiny compared with where they started. Quanta Magazine wrote that Tao “came away with one of the most significant results on the Collatz conjecture in decades.” Yet the key word remains almost, and almost is not all.

Why a rule this simple might be unbeatable

In 1972, John Horton Conway showed that a natural generalization of the puzzle is undecidable, meaning no algorithm can ever settle every case. That does not doom the original question, but it shows how close this little game sits to the outer edge of what mathematics can know.

Like Goldbach’s conjecture, it has never been cracked, yet the effort has already paid off: chasing it has produced new techniques and a long list of partial results. Perhaps that is its real gift, a reminder that the deepest questions do not need complicated words, only a pencil, a number and a little patience.