Research
Scientific object: how artificial and hybrid minds think, decide, understand, and stay constrained under load. Active spine: S1 Agent Mathematics — the Lineage Equation as a capacity law under load.
Agent Mathematics: numeracy through graphs, information, and control. The Lineage Equation is the first capacity law — structural inertia, directed flux, and bottleneck rate composed into one invariant. Not advanced math pretending to be foundational.
Viability before optimization. Dual-estimator interference: deliberation can degrade a correct fast answer. Plasticity and action stay gated by mode, reserve, and tissue health. Measurable constraint — not sermons.
Proof ladder A–I with a contiguous-stage rule. Identification protocols, held-out prediction, causal intervention. The Lineage reference module is the first instrument: a law that needed a tool.
When structure, flux, and latency interact, what invariant still holds — and how do you prove it on live traces?
When N hypotheses are live, what collapse rule is good — and how do you prove it?
What structure improves multi-step understanding without over-completion?
Where agents violate stated constraints under load — and how to measure it.
Newton method — not product cosplay. Official maturity is the highest contiguous proof stage. Instruments only when a law needs a tool.
Results that stick go to Findings. Writeups go to Blog. The open mathematics packet lives in AI-STEM under this lab's name.