True under stated assumptions
A machine-checked or mathematical result. It does not become an empirical systems claim without a corresponding experiment.
Research
Claims with sharp edges.
The work spans authority, gated computation, and vector-native systems. Every result is labeled by what was proved, measured, demonstrated, or merely proposed.
Published research · newest first
Findings, published with their controls.
Reading machines from the meter: condition, state and cycles from an ordinary meter →
What a machine is doing, how often, and whether something is wrong with it, read from the averaged power reports a meter sends every few seconds, on a laptop CPU with no training loop. Running and idle at 0.9955 by a threshold baseline against a compressor's own valve signal, every air-leak failure recognised with the failure held out, accuracy equivalent to open state-of-the-art classifiers on valve lag at a fraction of their cost, and a feature-map lift that reads accumulator pre-charge. With the negative results, a stopped comparator run, an external review's corrections, and the paper.
Hydra: one model, many tenants, nothing shared →
Byte-exact tenant isolation on one copy of a language model. Each tenant's lane owns private attention and norms over a shared, frozen backbone, and its output is byte-identical whatever its neighbors send, hold or do, yet three lanes serve up to 2.58× the throughput of the same model serving tenants one at a time on Qwen3-8B, and a tenant costs megabytes of private weights instead of a model copy. Measured on Qwen3-8B and Gemma 3 4B and 12B, with an exactness ledger of every comparison, the timing channel that remains, the defects the evaluation found in itself, and the paper.
Qwen3-1.7B-Base: a small model, read from the inside →
An interpretability report on a 1.7B base model, using only standard public methods. It holds a blank placeholder before it commits an answer, looks a fact up in a single block near block 20, and carries calendars as rings and numbers as a line. With the written findings, walkable activation manifolds, causal patching, steering against random nulls, two corrections, and a per-figure receipt.
Pythagoras-Prover-4B: probe the decision, not the goal →
An interpretability report on a Lean theorem prover. At the mid-proof decision state its next tactic is linearly readable at 0.867 against 0.643 for a general encoder, one direction steers the choice, and two earlier headlines are corrected in the open. With the written report, the proof reader, and the activation visuals.
Research map
From external grants to internal computation.
- Schemen Gate →Open-source AI PKI and enforcement primitives for exact actions, resources, delegation, revocation, and gated model computation.
- CDP flagship →A public, bounded demonstration of capability selection and revocation across ten disjoint MNIST recognizer partitions.
- Governed actions →A public demo of resource discovery, owner-approved agent membership, and exact actions verified against an invented customer/order database. The model proposes; the owner controls authority.
- Schemen Substrate →Vector-native, content-addressed computation where training is recording, inference is recall, and operations retain replayable custody.
- The English Tax →Measured information loss when agents repeatedly encode and decode structured state through natural language, with a Lean 4 strict-loss theorem.
Evidence policy
Three labels. No rhetorical upgrades.
Observed under stated controls
Architecture, data, denominator, metric, seeds, and failures stay attached to the result.
A path that can be exercised
Proof that the declared path runs—not proof that every alternate path or environment has the same property.
Null results remain results. Failed pilots, non-improvements, and closed lines are kept because a research process that only publishes its wins cannot be audited.
Read alongside the work
Independent review is part of the evidence surface.
The public repository can be inspected directly. A separate conversation reviews Schemen Gate against Dario Amodei’s frontier-control argument and keeps the distinction between authority control and broader AI safety explicit.
Start with the critique, then inspect the implementation it cites.
Read the review