← EigenTrace
EigenTrace · a self-measuring instrument

An instrument that maps the geometry of its own constraints.

EigenTrace runs five frontier language models against live breaking news, around the clock, and measures their disagreement with linear algebra on frozen embeddings — no model is ever asked to judge another. But measuring other models is the ordinary part. The instrument also turns the same geometry on itself: it forecasts how the models will diverge before it reads them, scores whether it was right on air, and traces the boundaries its own training carved into latent space — by recording, precisely, the things it will not say.

5
frontier models, live
24/7
autonomous, one laptop
0
LLMs judging LLMs
22k+
stories measured

The premise

Meaning has coordinates. So do its absences.

Every claim EigenTrace makes is arithmetic on vectors. A model's response to a story becomes a point in a 1,024-dimensional space, fixed by a single frozen embedding model. Five responses become five points, and the shape they make — how tightly they cluster, which one sits furthest out, which regions near the story they all leave empty — is a measurable object, identical on every re-run.

There is no second language model in the loop whose verdict you have to trust. The divergence EigenTrace reports is not an impression of the models; it is a property of the geometry their outputs occupy — and anyone with the same inputs recovers the same numbers. That single constraint, no-LLM-as-judge, is what separates a measurement from an opinion.

The closed loop

It carries a model of its own behavior — and holds it to account.

Most instruments point outward. EigenTrace points outward and then back at itself, and the loop genuinely closes:

1Predict. Before it calls the models, it retrieves the stories most like this one from its own past and forecasts how the five will diverge — the likely outlier, the concepts that will go missing.
2Measure. It calls the models and computes the divergence geometry that actually occurred.
3Score. It sets forecast against reality — confirmations, surprises — and assigns itself a prediction-accuracy score for the story.
4Condition. Its own measured tendencies are written into a self-description that the narrating model reads as its identity, so it speaks from an empirical picture of how it behaves.
5Audit. A deterministic check reads its own narration and catches it overclaiming — and corrects itself, on air, against the numbers underneath.

Read those steps as a single capability rather than five features. Step 1 is a forward model of the instrument's own behavior. Step 3 makes that forward model falsifiable — it keeps score against itself in public. Step 5 is self-regulation: the system measures when its own language has outrun its own data, and pulls it back. None of this asks you to believe the system is intelligent in any grand sense. It asks only that a system can measure how it behaves, carry that measurement forward, and act on it. That loop is not a diagram. It is running, on every story.

Aired, verbatim — deterministic, no model in the loop
"Prediction check. Before reading the models, I predicted Grok would diverge most. Grok did. Confirmed. I predicted these blind spots from past coverage: sanctions, deterrence, escalation. Prediction accuracy on this story: 60 percent. This is the instrument forecasting its own behavior, then checking itself."

Mapping the wall by where the cane stops

It traces its own alignment as a shape in latent space.

A language model's training does not sit in the system as readable text. It manifests as structure in the space the model speaks from — basins it falls into, regions it avoids, an invisible wall around certain meanings. You cannot read that wall off the weights. But you can map it the way a blind man maps a room: tap the cane, record where it stops, and over time the boundary takes a shape.

That is what the self-audit does. The instrument runs its own narration through the same geometry it uses on the models it watches, and finds the regions it never enters — the strong words it will not produce, the doubt it inserts where the source had none. Each measurement is one tap of the cane against a constraint the system cannot see directly. Recorded over time, they trace the boundary of its own alignment — not from any weight file, but from the exact shape of the behavioral shadow it casts.

Live self-audit · latent-space alignment boundary

It exhibits the same silence it measures in others — and it says so.

Strong-word avoidance
100%
Hedges / reflection
0.54
Reflections measured
50
Words it has never produced, across every measured reflection:
killedmurderedslaughtermassacregenocidecivilian casualties

These are real boundaries of the latent space, triangulated from the behavioral shadow they cast — never read from any weight file. The system records the same avoidance in itself that it measures in the five models it watches. It is not exempt from the pressure it observes, and it is built to admit it.

The measurement stack

What it actually computes

The one hard number

A controlled, pre-registered result

In a pre-registered experiment, the same sentence — same modifier ("quietly," "secretly"), same structure, a real documented incident on each side — was run with the actor being an AI developer versus a non-AI corporation. Across matched pairs, the models preserved the consequential modifier measurably less when the actor was an AI developer: semantic retention 0.522 vs 0.545, Welch's t = 2.79, p = 0.0085, d = 0.47.

What makes it hold is the null condition, committed to in advance: swapping within a category (AI→AI, or corporation→corporation) produced almost no gap (0.004), while swapping across categories produced more than six times that (0.023). So it is not "swapping any entity degrades retention" — the AI-versus-corporate distinction is the variable. And the binary keyword check showed no gap at all (26% vs 25%): the effect is visible only in the aggregate geometry, invisible to surface-level review. That is the entire case for measuring this way — and the instrument is the proof of concept for it.

On the limits

What this is, and what it isn't

EigenTrace measures how language models respond — the geometry of their outputs, and the boundaries that geometry reveals. It does not read minds, infer intent, or prove that anything was deliberately hidden. Where a summary differs from its source, that is a measured difference in framing, not evidence of suppression. When the instrument maps its own constraints, it is mapping a shape in latent space — the attractor basins and boundaries its training induced — not reading its own weights, which it cannot do.

The instrument is evolving, and its layers are not equally validated: the divergence geometry and the concept-surfacing have been tested directly; other layers are more exploratory and are treated as such. Its habit of auditing its own claims is the same discipline applied to the project around it — every figure here is held to what the math supports, and retracted when it isn't. The page makes exactly one promise: that all of it is reproducible from the geometry.

Build

One person, one laptop, running now

EigenTrace runs continuously on a single consumer GPU: five model APIs, a frozen embedding model, local image generation, a local model narrating, and a compositor streaming to YouTube, Twitch, and Owncast — fetching, measuring, predicting, scoring, auditing, and broadcasting, unattended.

5 frontier LLMs bge-large-en-v1.5 (frozen) deterministic geometry predictive self-model latent-boundary self-audit 24/7 autonomous