Can $π$ generate itself? A Monte Carlo analysis of 314 trillion digits

Imagine you are trying to guess the shape of a circle, but you don't have a ruler, a compass, or even a piece of paper. All you have is a bag of marbles and a square box. This is the essence of a famous trick in mathematics called a "Monte Carlo" simulation. The idea is simple: if you throw thousands of marbles randomly into a square box that contains a quarter-circle drawn inside it, the ratio of marbles that land inside the curve to the total number of marbles will tell you the value of $\pi$ (the number that relates a circle's diameter to its circumference). The more marbles you throw, the closer your guess gets to the real answer. Usually, to do this, you need a machine to generate truly random numbers to decide where each marble lands. But what if you didn't have a random number generator? What if you tried to use the digits of $\pi$ itself—the very number you are trying to find—as the source of your "random" marbles? It sounds like a paradox, like trying to lift yourself up by your own bootstraps, but that is exactly the strange question a team of scientists decided to test.

This paper, written by researchers from the Gran Sasso National Laboratories in Italy, explores a mind-bending idea using a massive dataset: the first 314 trillion decimal digits of $\pi$. To understand the stakes, we need to know a little about the nature of $\pi$. It is a deterministic number, meaning its digits are not random at all; they are fixed and calculated by a strict mathematical rule. However, for centuries, mathematicians have suspected that these digits look random. They seem to be "normal," a fancy word meaning that every possible combination of digits (like "123" or "999") appears with the same frequency, just like if you were rolling a fair die forever. If $\pi$ is truly normal, its digits should be good enough to act as a source of randomness for computer simulations. The researchers wanted to see if they could use the digits of $\pi$ to run a Monte Carlo simulation that would successfully "re-discover" $\pi$ without using any outside random numbers.

The team took the record-breaking dataset of 314 trillion digits, which was computed in late 2025, and treated it like a giant stream of numbers. They chopped this stream into tiny blocks of 14 digits at a time. The first 7 digits of a block became the "x" coordinate of a point, and the next 7 digits became the "y" coordinate. Imagine these coordinates as the landing spot of a dart thrown at a square target. By doing this for every possible block, they created about 22.4 trillion "darts" (or data points) to throw at their virtual target. They then counted how many of these points landed inside the imaginary quarter-circle.

The result was a stunning success. When they crunched the numbers, the simulation produced an estimate of $\pi$ equal to 3.141593. This matches the first seven decimal places of the actual value of $\pi$ perfectly, right within the margin of error expected for a simulation of this size. The authors note that this doesn't prove that $\pi$ is mathematically "normal" or that its digits are truly independent of one another (since they are generated by a fixed rule, they can't be truly independent). However, the experiment provides strong empirical evidence that, for all practical purposes, the digits of $\pi$ behave so much like random numbers that they can be used to generate a Monte Carlo simulation that successfully recovers the number itself.

In simpler terms, the paper shows that $\pi$ can, in a way, "generate itself." It's as if you took a dictionary, cut out every letter, and used those letters to write a new book that somehow contained the exact same dictionary you started with. The researchers found that the digits of $\pi$ are so statistically chaotic and well-distributed that they can serve as their own source of randomness. This doesn't mean the digits are magic or that they create new information; rather, it suggests that the complexity of $\pi$ is so high that it hides its own deterministic nature, acting as a perfect random environment for itself. The study confirms that even though $\pi$ is a fixed, calculated number, its digits are a reliable tool for simulating randomness, at least up to the scale of 314 trillion digits.