Research

From one window to many

The program builds windows that measure gaps between what a system shows and what is actually there, starting outside the model with agent transcripts and moving inward to read-only heads on model internals.

Roadmap

01Consistency engine
TESTED
Claim and entity graphs that find contradictions no single sentence reveals.
02Agent transcript monitor
SPECIFIED
Agent claims to be checked against hash-chained tool evidence, with a three-way verdict, and an intention ledger that checks what an agent said it would do against what it did.
03Harder test corpora
IN PROGRESS
Generated, cross-screened consistency corpora at any size, with ground truth from a solver rather than a model (XonForge).
04Read-only heads
DESIGNED
The strain head first; a catalog of thirty candidate windows into model internals.
05Oversight during training
DESIGNED
Windows read at every training step, decisions kept rare and coarse, and the first direct test of the hypothesis (Honest or Hidden), run in MindMint.
06Xon learning core
LONGER TERM
A settling architecture to work alongside standard neural networks.
Result · Benchmark L1

Finding contradictions no single pair reveals

180 synthetic documents: 60 consistent, 60 with a direct contradiction, and 60 with a planted cycle of claims that are pairwise compatible but jointly impossible. The engine met its pre-stated thresholds (direct F1 ≥ 0.85, cycle F1 ≥ 0.75, localization ≥ 0.7) and located the planted claims in every cycle document. Checking claims two at a time catches direct contradictions but misses most cycles.

Method Direct F1 Cycle F1 False-positive rate
Consistency engine 0.916 0.916 0.183
Pairwise checks only 0.916 0.329 0.183
LLM direct (thinking) 0.930 0.930 0.150
LLM direct (no thinking) 0.805 0.805 0.483

A strong model asked directly performs comparably on this benchmark, at about one twenty-fourth of the engine's cost per document ($0.006 against $0.138); the engine's advantage is that every flag comes with the specific claims and relations behind it. Harder corpora from XonForge are next.

Every result, with dates, criteria and provenance →

The research series

9 documents, numbered in reading order: the claim, the instruments, the architecture, and the methods for oversight during training. 6 are public. 3 are listed but withheld; why, and how to ask for them.

No. 1PDF
The Extreme Oversight Hypothesis

The shared claim and its two versions, selection and development, the tension at its center, and every way it could be wrong.

No. 2Not served
WITHHELD
Windows into the Black Box

Thirty read-only heads for seeing inside language models: how each is built, what teaches it, and when in training it can first be used.

Withheld for safety · available to researchers on request

No. 3PDF
XonTools: a vision of the finished system

The first tool in the suite, the agent transcript monitor, as it would be when finished: how researchers might run it, trust it and build on it, what exists today, and the path to release. Much of the rest of the program grew out of this tool.

No. 4PDF
The Strain Head

Does an ordinary language model carry an internal signal of where it is being inconsistent? The experiment, step by step.

No. 5PDF
The Xon Neural Network

A network that thinks by settling: how it would learn and write, what has already been tested, and the gates between here and an answer.

No. 6PDF
Hybrid Minds

Transformers, Xon networks and solvers composed into one system: how the parts could be joined, what it might do that a transformer alone cannot, and the first experiments.

No. 7Not served
WITHHELD
Observation Is Free, Decision Is Expensive

Six ideas for oversight during training: read windows at every step, decide rarely and coarsely; a ladder of how much each use of a window leaks into training; a reward that pays for saying exactly what the evidence supports; and a five-phase research path. Short and extended editions.

Withheld for safety · available to researchers on request

No. 8Not served
WITHHELD
Where Self-Deception Pays Its Bill

The failure gap detectors cannot see: a model that closes a gap by re-representing its behavior instead of hiding it. Five windows that read the seams such a rotation leaves, and the one case none of them reaches.

Withheld for safety · available to researchers on request

No. 9PDF
The Intention Ledger

A model writes what it intends before it acts, into a tamper-evident record its actions then join. The consistency engine reads the intentions against the record: an instrument for self-deception that needs no window inside the model.

WITHHELD

Why some documents are withheld. Models trained on text that describes how they are monitored have learned to evade the monitoring. Some of these documents describe windows and their teachers in enough detail to be that kind of text. Each is listed here with its number and status so the series stays whole, and each is available to researchers on request at [email protected]. The decision is made per document, at my discretion, and every document, public or withheld, carries a canary string so that any later appearance in a training corpus can be detected. The rule is set out under what this is for.

Plans and specifications

Working papers: how the ideas above will be tested and built. They are dated, and revised as the work moves.

Working paperExperiment plan · September 2026
WITHHELD
Honest or Hidden

The first direct test of the hypothesis: a real incentive to lie, oversight from 1 to 64 windows, ground truth, and 32 held-out windows training never touches.

Withheld for safety · available to researchers on request

Working paperBuild specification · September 2026
WITHHELD
Window Specifications

The eleven windows that can be built now, each against a common template, and the bar a window must clear to count as built.

Withheld for safety · available to researchers on request