Illustrative thought instrument

Longwave

An agent memory darkroom. Develop one episode over time and watch it separate into exact wording, durable gist, and uncertain residue.

Not neuroscience · not a benchmark · not a prediction of any real model

1The episode

Built-in synthetic episode

2Darkroom controls

Rehearsal intervalevery 15
How often the agent revisits the episode. Frequent rehearsal keeps exact wording sharp across many more epochs.
Semantic compressionmedium
How hard meaning is squeezed. High compression trades exact words for a longer-lived gist.
Interferencelow
Competing information crowding the episode. Interference dissolves fragments into residue.

3The developing tray

epoch 0 / 100 fresh exposure
Drag the timeline or tap a frame below to develop the episode forward.
epoch 0 — freshepoch 100 — long after

4Memory-design receipt

A compact record of your controls and what the illustration demonstrated — no score assigned.

How the illustration works

Longwave is a deterministic thought instrument. The same episode and the same control values always produce exactly the same development — there is no randomness in the outcome and nothing is saved or sent.

The episode is split into fragments. Each fragment starts as fully legible exact wording. As epochs pass, each fragment loses focus at a rate set by the controls:

Rehearsal interval sets how fast focus fades — raise it and exact wording stays legible across many more epochs of the contact sheet.
Semantic compression lets high-salience fragments survive as durable gist long after their exact wording is gone.
Interference speeds fading and pushes fragments into uncertain residue.

Filler fragments fade first; meaning-carrying fragments persist as gist; the rest becomes residue. It illustrates a familiar idea — we keep the shape of a thing long after the exact words dissolve — and is not a claim about any real system, brain, or benchmark.