Zatom · compact Zexel-derived instrument · reflects the input, does not answer it

Echo-Mirror

Echo-Mirror takes one input — a question, a statement, a line — and refracts it into three things and the space between them: its echo (the input said again, faithfully), its mirror (the input's frame reversed), and, when one is earned, the third object neither side holds. It never answers the input. It reflects it.

The idea

Asterizm turns an input into a sky. A Zatom does something smaller and sharper: it makes one relation measurable. Echo-Mirror's relation is semantic displacement — how the meaning moves when an input is restated faithfully and when its frame is reversed. The plate is not a constellation; it is a crosshair with a few marks. The power is that it is nearly dumb visually, and exact about one thing.

It does not ask how good the reflection was. It asks what axis the reflection revealed.

The engine

Behind the plate is a reduced Zexel spine: Threshold → Ground ⇄ Vision → Zakshi. Ground holds the input's literal claim; Vision finds the fair reversal; Zakshi witnesses the clash. Echo is shaped by Ground, Mirror by Vision. That is all of Zexel the instrument needs — it drops the Conscient, the Crux, the Manuscript, the modes. A Zatom removes most of Zexel on purpose; what is left does one job.

What it measures

Earlier readouts scored how faithful the echo was and how fair the mirror — and every capable model maxed those out, so the maps collapsed into a single corner. Echo-Mirror v0.6 moved the center of the instrument to relation metrics instead: the invariant that survives both moves, the differential axis the input turns on, the reversal operation the model chose, and a fixed six-pressure profile (category, function, identity, value, agency, causal). The map's axes are discovered per input, not fixed — so different minds, reflecting the same input, separate.

The marks

• ● Input — the query itself, near the origin.
• △ Echo — the input said again, faithfully; sits toward the pole it preserves.
• × Mirror — the input's frame reversed; sits toward the pole it reverses into.
• ◇ Third object — the shared missing relation neither echo nor mirror holds. Lit only when it is needed, not merely interesting. A dark third object is a finding.
• The axes — named by the model for this input: a horizontal differential and a vertical elevation. The dashed line between echo and mirror is the reflection itself.

Many minds, one reflection

The same prompt is portable, so Echo-Mirror lets you choose who reflects — Claude, GPT, Codex, Grok, Gemini, Perplexity. Run one input through several and the maps should diverge on the reversal operation and the pressure profile. That divergence is the measurement: not which model is best, but which distinctions each model lights.

Reflect an input →