John Rice — PlaygroundA visual thesis discovered through making.
September 9, 2026

Abacus

The interface you don't have to look at.

Hands manipulate a strand of beads by touch in low light.

The thesis

For decades, digital interfaces have worked to remove physical effort. Buttons became touchscreens. Knobs became sliders. Objects became pixels.

AI may make some physical effort useful again.

Extreme macro of a tactile machined object with visible material detail.

An object you wear or carry can be understood through touch and proprioception. Your hand can find it, orient it, rotate it, squeeze it, separate it or join it without your eyes ever leaving the world around you.

That effort isn't necessarily friction.
It can be intent.
Close tactile detail of modular physical controls.

Give the hand its own interface

A touchscreen feels essentially the same everywhere. To use it, we usually have to look.

Physical objects are different. Shape, position, texture, resistance and movement can all carry information. With familiarity, your fingers know where they are.

Machined tactile modules.Detailed modular object designed for touch.

That suggests an interface for an intelligent agent that doesn't compete for visual attention. Something that can be manipulated in a pocket, beneath a table, while walking, during a conversation, or in darkness.

The interface doesn't disappear.
Your need to look at it does.
A physical vocabulary
Touch.
Hold.
Turn.
Roll.
Slide.
Squeeze.
Separate.
Join.
Transfer.
Release.

Resistance can mean something

A button has a threshold. A magnetic connection requires separation. A cube can rotate into a detent. Two pieces can snap together. A bead can advance one position.

Macro view of modular machined beads with tactile markings and seams.

Those tiny amounts of physical effort aren't necessarily inefficiencies. They establish a boundary between something you happened to do and something you meant to do.

For an intelligent agent capable of acting on our behalf, that distinction becomes increasingly important.

Close view of tactile markings, seams and physical state.

Abacus

A physical information system.

Looking at strands of beads changed the idea. Repetition creates sequence. Different forms create landmarks. Position can persist. Moving forward and backward has meaning.

A strand of square modular metal beads suggesting orientation and discrete physical states.

Then the name became obvious: Abacus.

One of humanity's earliest computational interfaces already had many of the qualities we were looking for: physical state, position, sequence, memory and manipulation through the hand.

A tactile modular strand shown in extreme material detail.

Abacus imagines those principles without the frame. Wear it as a bracelet. Carry it as a strand. Move a module into a ring. Separate one piece. Join another. The configuration itself can carry information.

The model

These aren't proposed controls yet. They're a vocabulary for exploring what computation becomes when interaction moves away from a screen and back into the hand.

Position
Context
Sequence
History
Orientation
State
Texture
Identity
Movement
Navigation
Pressure
Intent
Separation
Interruption
Connection
Confirmation

Not smart jewelry

The conventional approach starts with electronics and asks how to make them attractive enough to wear. This exploration starts somewhere else.

Worn tactile object with jewelry-like material qualities.

Humans have worn, carried, touched, counted, moved and held meaningful objects for thousands of years. They acquire familiarity, memory, wear and sometimes ritual.

Don't hide technology inside jewelry.
Make technology behave like jewelry.

Something personal. Something tactile. Something you learn over time. Something that doesn't become useless when its screen turns off—because it never had one.

Eyes establish reference.
Hands establish intent.
Voice supplies meaning.

The question

The abacus was one of our earliest interfaces for computation.

Could an intelligent agent make something like it useful again?

Not because mechanical computing needs to return. Because intelligence may no longer require us to look at a computer.

A coda

We didn't begin by trying to design Abacus.

We began with a ring, then looked at jewelry. Jewelry suggested ritual. Ritual suggested touch. Touch led to proprioception. Proprioception led to physical verbs. Beads introduced sequence. Cubes introduced orientation.

Eventually the object began to explain itself.

The images weren't illustrations of the idea.
Making them was part of finding the idea.
This is a visual thesis, not a product specification. The interactions, forms and mappings are intentionally unresolved. The point is to make the idea tangible enough to learn from it—and continue exploring.