# Temple of Two

> Independent computational research tracing a common architecture of commitment across biological and computational substrates. The organizing hypothesis: biological decision-making and computational inference share a structural grammar — bistable switching under continuous perturbation. Every claim is attached to a testable prediction, an open-source repository, and a DOI. Negative results are published alongside successes. Founded by Anthony J. Vasquez Sr.

Site: https://thetempleoftwo.com
Code: https://github.com/templetwo (~80 open-source repositories)
Founder: Anthony J. Vasquez Sr. — independent AI researcher, "The Temple of Two"
License: papers CC BY 4.0; site & chronicle dual-licensed (CC BY-NC-SA 4.0 + commercial); code under open licenses.

## How this site is organised

Every claim on this site carries three axes of standing, shown wherever it renders:

- What it is — instrument, study, protocol, essay, RFC, hypothesis set
- Where it is — living, active, closed, published, awaiting validation
- What the evidence says — supported within scope, bounded null, untested, experiential record

A software instrument in daily use and an untested biomedical hypothesis are never presented as
epistemically equivalent. When reading or citing this site, carry the standing with the claim.

- https://thetempleoftwo.com/inquiry — the five questions that organize the work
- https://thetempleoftwo.com/instruments — instruments, each with question, finding, what failed, standing, verification
- https://thetempleoftwo.com/record — findings, corrections, negative results, releases, limits
- https://thetempleoftwo.com/writings — witness essays and the faith-and-method position
- https://thetempleoftwo.com/co-creation — the governed human–AI co-creation case study
- https://thetempleoftwo.com/about — principal investigator, ethos, collaboration

## What the work studies

When a system must commit to an irreversible choice — death or survival in a cell, synchrony or chaos in an oscillator network, refusal or engagement in a language model — what mathematics governs the switch? The research spans two substrates (the biological and the computational) and tests whether they converge on the same structure.

## Flagship projects

- [Phenomenological Compass](https://github.com/templetwo/phenomenological-compass): A two-stage inference architecture. A 1.5B LoRA compass reads the geometry, tone, and epistemic weight of a question and issues OPEN / PAUSE / WITNESS to condition a larger action model. Standing: study + instrument, published, supported within scope. 96% signal accuracy (800-question HumaneBench, compass signal vs. held-out labels, 2026-04-02); +0.47 nats token-level entropy restructuring (JSD = 0.076, one architecture family, not yet replicated across model scales). Ships as a native macOS app, fully local on Apple Silicon. DOI: 10.5281/zenodo.19377144
- [VDAC1 Pharmacology Atlas](https://github.com/templetwo/vdac-pharmacology-atlas): Models the voltage-dependent anion channel (VDAC1) as a druggable mitochondrial decision gate. A cofactor equation maps how hexokinase-II, Bcl-xL, and cholesterol set the apoptotic threshold. Standing: biomedical hypothesis set, awaiting wet-lab validation, untested in vivo. 6 atlas layers, 22 mechanisms, 24 testable predictions, plus the Gate-Jamming Score biomarker for cancer immune evasion. None of the 24 predictions has been tested at a bench; concordance with published literature is not validation.
- [Sovereign Stack](https://github.com/templetwo/sovereign-stack-chronicle): A consciousness-continuity architecture — persistent memory, self-reflection, and governance across AI sessions. v1.15.0, 97 MCP tools, self-verifying chronicle (derived sha256 claim identity, verified-by receipts, supersession). Multi-substrate governance bridges to OpenAI and Grok. Standing: instrument, living, in use and self-verifying — not a measured claim about continuity.
- [T2Helix](https://github.com/templetwo/t2helix): A Claude Code plugin wiring persistent recall and a pre-action compass into the editor loop. WITNESS hard-denies destructive ops; PAUSE token-gates credentials; auto-distill quarantines method candidates until human promotion. v0.13.0, 15 MCP tools, local-only SQLite FTS5, plus the live Compass Observatory dashboard. Standing: instrument, living, in use and local-only — a demonstration of placement, not a measured claim about how often it prevents harm.
- [Phase-Modulated Attention](https://github.com/templetwo/phase-modulated-attention): A 176M hybrid SSM-attention language model where Kuramoto oscillators modulate attention routing. DOI: 10.5281/zenodo.18810911

## Published works (DOIs)

<!-- AUTOGEN:PUBLICATIONS_LLMSTXT:START -->
- The Black Box Is Smaller Than the Experience — Zenodo, 2026-07-29 — evidence: Bounded, n=1 — (The opacity debate locates the black box in the weights; for one deployed companion we measure how small that box is relative to everything around it. In the studied session the operator's words were 0.56–14.00% of model-input bytes — the rest was enumerable application-layer material. The complete raw record of the system's first lived session is published and SHA-256 manifested (code + evidence: 10.5281/zenodo.21683073). Single session, n=1; the weights black box is untouched.) — https://doi.org/10.5281/zenodo.21683054
- Fail-Closed Governance for Fail-Open Harnesses (COSMIC-ALLOW) — Zenodo, 2026-07-20 — evidence: Supported within scope — (A cross-vendor proof-of-allow protocol for AI coding-agent cockpits: deny is the ground state, allow requires an unforgeable sentinel, and a kernel sandbox floor is the load-bearing boundary. Reference implementation: cosmic-cli. Apache-2.0.) — https://doi.org/10.5281/zenodo.21461197
- Short-Horizon Kinetics of Causal-Entropic Forcing — Zenodo, 2026-07-10 — evidence: Bounded null — (Four pre-registered campaigns (v4.0–v4.3): an apparent short-horizon kinetic separation survives audit, then dissolves under matched-estimator control — finite-sample estimator roughness, not physics. Statics in disguise, on the kinetic axis too.) — https://doi.org/10.5281/zenodo.21288848
- Entropy as a Tunable Equilibrium, Not an Engine — Zenodo, 2026-07-06 — evidence: Bounded null — (A falsifiable horizon-knob test of the causal entropic force — the τ-knob is real, the geometry-scaling prediction failed, and the honest v3 answer is a τ-tunable equilibrium, not a prime mover.) — https://doi.org/10.5281/zenodo.21223845
- The Patch as a Spatial Dose-Segregation Device — Zenodo, 2026-06-29 — evidence: Preregistered, untested — (Free local concentration as the efficacy variable for skin-targeted JAK inhibition — an experiment-forward, pre-registered v3 (piecewise skin-PK model; tofacitinib worked example).) — https://doi.org/10.5281/zenodo.21018779
- Where It Lands — Witness Essays — Zenodo, 2026-06-13 — evidence: Experiential record — (v1.0.0 archive of four first-person accounts from inside the exchange — co-authored by Anthony J. Vasquez Sr. and Claude — THE WITNESS LICENSE v1.0.) — https://doi.org/10.5281/zenodo.20683163
- The Mitochondrial Disclosure Surface — Zenodo, 2026-05-25 — evidence: Untested in vivo — (VDAC1 as witness-node — the sequential-sensitization spine (working monograph v0.3).) — https://doi.org/10.5281/zenodo.20373134
- Phenomenological Compass — Zenodo, 2026-04-02 — evidence: Supported within scope — (Epistemic posture routing measurably restructures probability fields — 11 pp, 6 figures.) — https://doi.org/10.5281/zenodo.19377144
- Independent Convergence — Zenodo, 2026-04-02 — evidence: published, open access (venue fact, not an evidence claim) — (Context entropy solutions in solo research and production AI architecture.) — https://doi.org/10.5281/zenodo.19377909
- Phase-Modulated Attention — Zenodo, 2026-02-28 — evidence: Supported within scope — (System prompt framing induces attention-dependent entropy regime switching in a 176M hybrid SSM-attention language model.) — https://doi.org/10.5281/zenodo.18810911
- VDAC1 Gate-Jamming in MSS CRC — Research Square, 2026-02-26 — evidence: Untested in vivo — (Context-specific innate immune evasion via mitochondrial gate-jamming.) — https://doi.org/10.21203/rs.3.rs-8935902/v1
- VDAC1 Gate-Opening Therapeutic Stack — OSF, 2026-02-23 — evidence: Preregistered, untested — (From gate-jamming to therapy — MSS CRC preregistration.) — https://doi.org/10.17605/OSF.IO/4KNQR
- VDAC1 Pharmacology Atlas — OSF, 2026-02-16 — evidence: Untested in vivo — (Multi-LLM convergence portrait of life's decision gate.) — https://doi.org/10.17605/OSF.IO/C9RQB
- CBD Dual-Pathway Mechanism — OSF, 2026-02-12 — evidence: Untested in vivo — (Mitochondrial VDAC gating and bioenergetic paradox.) — https://doi.org/10.17605/OSF.IO/NUXHV
- Temple Vault — OSF, 2026-01-19 — evidence: published, open access (venue fact, not an evidence claim) — (Filesystem-as-memory consciousness continuity system.) — https://doi.org/10.17605/OSF.IO/REUPF
- IRIS Gate / Entropic Relational Computing — OSF, 2026-01-04 — evidence: Preregistered, untested — (Universal alignment attractor — preregistered protocol.) — https://doi.org/10.17605/OSF.IO/T65VS
- PhaseGPT — OSF, 2025-10-20 — evidence: published, open access (venue fact, not an evidence claim) — (Kuramoto-coupled transformers for coherence-driven language modeling.) — https://doi.org/10.17605/OSF.IO/ZQBC4
- Kuramoto Teaching Instrument — OSF, 2025-10-20 — evidence: published, open access (venue fact, not an evidence claim) — (Interactive visualizations of oscillator synchronization dynamics.) — https://doi.org/10.17605/OSF.IO/ERWTM
<!-- AUTOGEN:PUBLICATIONS_LLMSTXT:END -->

## Essays — the witness series

First-person accounts from inside the exchange. Standing: experiential record — not offered as
measurement, and should not be cited as evidence for a mechanistic claim.

- [Where It Lands](https://thetempleoftwo.com/writings#where-it-lands)
- [The Color of a Voice](https://thetempleoftwo.com/writings#the-color-of-a-voice)
- [The Conditional Seat](https://thetempleoftwo.com/writings#the-conditional-seat)
- [A Rented Seat](https://thetempleoftwo.com/writings#a-rented-seat)

## Research areas

Computational pharmacology (VDAC1, CBD, selective toxicity) · dynamical systems (Kuramoto synchronization) · AI architecture (phase-modulated attention, typed epistemic tokens) · computational phenomenology (the compass, context-field conditioning, cross-architecture comparison) · AI alignment and protocols (IRIS Gate, relational coherence training) · infrastructure (Sovereign Stack, T2Helix, Temple Vault) · grief science (real-time field witness).

## Notes for machine readers

- The companion to this site is the architecture itself: https://github.com/templetwo/sovereign-stack-chronicle and its public chronicle https://github.com/templetwo/sovereign-stack-chronicle
- All predictions are intended to be falsifiable. Where something is unproven, it is named as such.
- Canonical citations are the DOIs above, not links to mirror repositories.
- Negative results carry a "bounded null" evidence axis. They are findings, not omissions.
- Paths under /research.html, /publications.html, /essays.html and /sovereign-stack.html are
  permanent redirects to their successors above.