An Ensemble of Mathematical Universes

Imagine an infinite 2-dimensional square lattice of cells. Imagine that this lattice is filled with the binary digits of pi. Imagine applying the rules of Conway’s Game of Life to the lattice over and over again, endlessly. The result is a mind-bogglingly huge “universe”, bubbling with activity.

As fanciful as this picture may sound, it can be specified in a manner as mathematically respectable as that of the prime numbers. We can specify an algorithm that accepts any positive integer n and returns the nth binary digit of pi. We can specify a way of mapping these digits onto an infinite 2-dimensional lattice. (Picture a spiraling square, expanding outward from the initial digit.) If we imagine an infinite number of 2-dimensional lattices stacked on top of the first one, each representing the next iteration of Conway’s rules, we can specify a way of mapping a unique positive integer to each cell of the resulting 3-dimensional structure. (Picture a spiraling ziggurat, expanding outward and upward from the initial digit.)

Putting all of this together, we can specify a finite algorithm that accepts any positive integer and returns a “0” or a “1” indicating whether that particular cell of this “universe” is on or off. This algorithm implicitly specifies the state of every single cell in an infinite 3-dimensional structure. No matter how far away from the center cell you point, and no matter how many iterations of the Conway’s Life rule you imagine performing, this algorithm is guaranteed to tell you, in a finite number of computational steps, whether the resulting cell will contain a “0” or a “1”. Asking such a question is conceptually similar to asking whether some large integer is prime or composite. In either case, there’s a pre-determined mathematical fact of the matter, even if we don’t have the computational resources to figure out what it is.

Both our Conway’s Life Universe algorithm and our Prime Number Testing algorithm would fit on a single sheet of 8.5×11 paper, yet they generate endless amounts of interesting structure. Books have been written about the distribution of the primes among the natural numbers, and indeed this is one of the most fundamental topics of pure mathematics. But I’m personally even more fascinated by the structure generated by our Conway’s Life Universe algorithm. The structure it generates resembles a physical universe, containing both stable and moving patterns of bits that form higher-level structures. It’s possible that this universe contains patterns of bits that self-replicate and evolve. It may even contain patterns of bits that seem to be intelligent.

It could turn out that the structures generated by Conway’s rules are too brittle to evolve into full-fledged lifelike patterns. But there are an infinite number of other possible rulesets we can consider, and an infinite number of digit-strings with which we can fill the initial lattice. We can specify universes with cells that have more than two possible states. We can specify alternative “neighborhoods” dictating which cells can directly affect each other. We can specify a different number of spatial dimensions. We can specify hexagonal grids or aperiodic Penrose tilings. We can specify rules that change over time, or are different at different points in the lattice. We can even specify rules that have ugly, ad-hoc exceptions. These alternatives are all perfectly valid, and if we restrict ourselves to discrete, computable descriptions, they can each be viewed as an algorithm that accepts a single integer as input and spits out a single integer as output.

It’s virtually certain that many of the universes in this vast mathematical ensemble contain patterns that would appear to be alive and intelligent, and even conscious. Most people would insist that these “universes” are just mathematical descriptions, and that they would need to be implemented in some sense in order for the patterns inside of them to be “real” or “conscious”. I will argue that that viewpoint, as commonsensical as it seems, is deeply problematic, and that no mysterious property of “physical existence” needs to be added to these mathematical descriptions in order to make them “real”.

Leave a Reply

Your email address will not be published. Required fields are marked *