Code Is for Summoning

A design defines an object. A program reveals a family.

As we witness frontier models producing images and videos of staggering fidelity entirely with code, it is useful to examine the medium of code more closely.

Code as a visual medium is often reduced to an alternative means of creating images and videos, simply because these formats are the only way social media platforms can safely distribute content. This framing vastly underestimates the potential of visual programs.

Code is the only medium we have access to that is capable of summoning infinite forms into material reality.

To illustrate this, consider the following program: https://www.shadertoy.com/view/3tsyzl.

This is a distance-estimated field to a quaternion Julia set, paired with a raymarching-based renderer to sample the distance field and display it on a screen. For more on Julia sets, fractal distance estimation, and raymarching, see https://iquilezles.org/articles/distancefractals/.

In essence, this program first restates the mathematical function that gives rise to the Julia set, provides a means to sample its form into discrete chunks, and leverages this to render it in real time. This program quite literally makes a slice of the Julia set physical – when executed, it writes charge to pixel capacitors based on the underlying iterated polynomial. But driving a two-dimensional grid of pixels is an arbitrary move – the program could just as easily be set up to output G-code for a CNC machine or a 3D printer.

The critical point is that a program such as this one is the only practical means of making the Julia set material. It is not feasible to carve or sculpt an accurate Julia set by hand, there is nothing in nature we can scan or cast that is exactly a Julia set, and there is no procedure for deriving its form other than calculating a sample of iterations defined by the function over a discrete physical space, and doing so by hand would just be an analog retracing of the programmatic approach. There is simply no other way to “get to it.”

So we have a clear example in which code is the sole bridge between an infinite theoretical shape and material reality.

The immediate question, and the one that should most excite us, is: what else is hiding in this medium? What strange, wonderful things may occur if we impose structural constraints on the forms that programs open to us? Imagine the infinite cathedrals sequestered in this space, waiting to be found. We have limited our understanding of code to simply a means of driving pixels and feel a sense of loss as LLMs saturate our screens. We must realize that this is a tiny sliver of the capability of programs and turn our gaze to a new frontier for code.