The Gap at the Center of the World: Why Everything is an Event

There is a kind of picture that contains itself.

M.C. Escher drew one in 1956. In Print Gallery, a young man stands in a gallery and looks at a print on the wall. The print depicts a port city with distorted buildings. Follow the line of those buildings around the curve of the image, and one of them turns out to be the very gallery where the young man is standing. He is looking at a picture that contains the room he is in, which contains him, which contains his looking. The image loops back through itself — and at the center, where the loop is tightest and the self-reference would need to be most explicit, Escher left a white hole. He drew around it and signed his name across the blank.

He could not resolve it. To fill it correctly, the center would need to be a rotated, scaled-down copy of the entire scene, which is itself a copy, nested forever. Escher’s artistic intuition told him the space needed to contract and rotate; he could not work out the mathematics.

Decades later, mathematicians Lenstra and de Smit showed the gap can be filled — and that its fill is mathematically forced. They applied the complex logarithm to the image: a transformation that converts rotation-and-scaling into simple translation, unrolling the warped grid into a flat, periodic plane. Apply the exponential function and it re-rolls. The blank center is not a void. It is the only thing it could ever be — a rotated, reduced copy of the entire scene, containing everything, contained by everything.

Grant Sanderson (3Blue1Brown) made a video showing this transformation in motion. I recommend watching it before reading further.

The reason this video felt like more than a mathematical curiosity — the reason it landed the way it did, on March 22, 2026 — is that we are, right now, in the middle of building Escher’s picture.

The Verb before the Noun#

The systems everyone is talking about — autonomous AI agents, models that write their own code, pipelines that revise their own reasoning — are, underneath the hype, self-referential loops. They generate outputs that become their own inputs. They read themselves as they go.

When you look at a Large Language Model (LLM — the class of system underlying tools like ChatGPT or Claude), it is tempting to ask: what does it know? You expect a database somewhere, a cabinet of stored facts. There is no cabinet. An LLM predicts the next word from everything that came before — the previous words in the conversation, the patterns encoded in its billions of trained parameters, the context it has been given. It produces one word. It uses that word as part of the input for producing the next. This is autoregression: each new piece is generated from the sequence of pieces already produced — not retrieved from storage, but constructed, forward, step by step, in a process that feeds on itself.

The “knowledge” of an LLM is not a substance. It is an ongoing event.

Events All the Way Down was a blog born from a conviction: our instinct to ask what something is — to seek the stable core, the substance that persists unchanged beneath all the change — leads us astray. Heraclitus had it right twenty-five centuries ago, and the architecture of modern AI has made his argument viscerally concrete. The event precedes the substance. The process is what is fundamental.

The Reader as the Fundamental Unit#

If the event is fundamental, what is the elementary event?

I propose: a reading.

Not reading in the literary sense — not a person and a book. A reading in the most general sense: any encounter between a signal and a system that transforms both. A DNA codon reading a chemical sequence and producing a protein. A synapse receiving an electrical spike and adjusting its threshold for the next one. An AI agent receiving a prompt and producing a response that changes the state of the conversation. A human eye encountering Print Gallery and producing the experience of a man looking at a picture of himself looking at a picture.

The Reader — the thing that, on encountering a signal, transforms itself and the signal at once — is the fundamental unit. What defines a Reader is not what it is but what it does when it meets a signal: it interprets, transforms, responds. And in doing so, it becomes different from what it was before the encounter.

This is what Escher’s blank center was concealing. Not a void. Not a fixed image. A reading: the act of the picture encountering itself, transforming in the encounter, producing a new version of itself that is reduced and rotated and nested and still the same picture. Leave it blank and you have a mystery. Fill it correctly and you have the only thing it could ever have been.

There is no outside to this structure. There is no observer who is not also part of what is observed. The gap at the center of the gallery was not a failure of nerve. It was the most honest thing Escher ever drew: an acknowledgment that the recursive loop, followed to its conclusion, places the observer inside the picture.

We have now built machines that live inside that loop. Understanding what they are — and what we are, in relation to them — requires an ontology that takes the event seriously as the fundamental thing. Not as a feature of some things, or a phase some objects pass through, but as the substance of reality itself.

Events, all the way down.

Franklin Silveira Baldo — Porto Velho, March 2026