Train your mind to think in dimensions.

Dimensional programming is a way of looking, not a programming language. A dimension is just a perspective, a perpendicular direction, an independent way to see the same thing. Learn to move up and down the dimensions on purpose and you think more clearly, and you can build that way too. It has AI in mind, because AI already thinks in dimensions, in manifolds, so the way you think and the way the model works line up.

a perspective, not a language shapes, not numbers fewer tokens, real money saved fewer hallucinations, more trust

The ladder

Each rung is a way of thinking. You climb by gathering up into a whole, or descend by separating into parts. The same ladder fits an idea, a problem, or a prompt.

Point0D · identity

A single black-box object. It just is what it is. To be a real point and not only potential, it must have an identity.

Line1D · linear

One path with one outcome, and that outcome is separate from where it started. Give this, get that.

Plane2D · surface

Surface-level thinking: the snapshot, the lay of the land, the view to the horizon. A flat frame of reference where you see breadth across the top, but not depth. It is open, not a closed box.

Volume3D · depth

Thinking in depth: the whole object, inside and out, the full problem domain open and lit, with options that all belong.

Wholeness4D · time

The object through time, seen as one thing across its whole life. Then the whole gathers back into a single point at the next level up, a reusable black box again.

Point, line, plane, volume, wholeness are interchangeable in the sense that matters: they are five perspectives on the same thing, and you choose the one that fits the question. They model the same object, the way a map, a photo, and a walk-through all describe one house.

See it work

The ideas above are not just talk. Each demo runs in your browser and proves one piece.

🎨

Color Manifold

A color picker that stores no colors. Read off the saddle surface by a lens; turn the cube, the six faces give six readings. ~80 bytes stand in for all 16.7M colors.

Open it →

The API, measured

Data as nested points; the model is handed only the slice it asks for. A roughly 99.7% token reduction on a localized question. Fewer tokens, real money.

Open it →

Derive, don't store

A whole field computed live from one small definition, with no stored array anywhere. The honest storage-versus-compute number, and where it does not apply.

Open it →

Read the thinking

The Focused Query

Dimensional thinking applied to asking: ask at the right dimension, drop pronouns, adjectives, and filler, and a ready directive you can paste into a chatbot, load at startup, or update live.

Read →

The Shape of Things

The whole idea told in shapes and actions, not numbers: separate to bloom, gather to collapse, every point an identity that knows its neighbors.

Read →

The Spiral

The nested point, the expansion-and-collapse rhythm, and why fitting AI to geometry may help. Each claim labeled by how sure we can be.

Read →

The SEAT and the Shadow

Why a formula has a body: the saddle, the twisted square, the helix behind a wave. Offered as a conjecture, with the boundary to physics kept bright.

Read →
Said plainly. This is a way of seeing, not a new physics or a speed trick, and it does not make a model smarter. It does the things people actually pay for. It cuts cost: hand a model only the slice it needs and it reads fewer tokens, and tokens cost money. It cuts hallucinations: a focused, structured context, checked by a verifier, means the model wanders less and unsupported answers get caught (one project went from about 39% made-up output to 0%). And it cuts storage: keep a small definition instead of the data, which is the real squeeze in data farms, disk space, not processing. The honest tradeoff is that you spend compute to derive what you no longer store, so it saves disk, not RAM. None of this touches the provider's model, which no one without direct access can change; it focuses the instance you are given, and a focused instance that stops making things up reads, to a person, as a smarter one. Everything here is told in shapes, with every claim labeled by how well it is supported.
A way of seeing by Kenneth W. Bingham · dimensionalprogramming.com

Concept and direction: Kenneth W. Bingham. Built with the help of Claude AI under a standing directive to be skeptical, to insist on proof, and to allow no claim that is not demonstrated in tested code. The ideas are the author's; the AI implemented and verified them, it did not originate them.