Deployed infrastructure
The instrument
Not a survey, not a simulation — a deployed system generating structured behavioral data continuously, with complete version-control history and telemetry available to partners.
Discord progression bot
Live · ~100 registered accounts · a thirteen-person active coreThe five-tier ASCEND progression system: dimensional scoring, review workflow, daily tier scheduler. Every credit-earning action, review decision, and tier transition is event-logged. Repository since July 2025; progression phases deployed February–June 2026.
AI-assisted evaluation layer
Deployed June 2026 · content-groundedClaude-based rubric evaluation of discussions and reflections, graded against the actual source text, with a human audit queue; capstone parts stay human-approved — AI operating inside the coordination instrument, not merely advising its operator.
The room: Think Tank + Sunday service
Tank nightly since May 2026 · Sunday service since August 2026The Think Tank convenes at 6 PM Central (auto-scheduled nightly, attendance tracked; 117 sessions and 13 distinct attendees by September 2026). The Sunday live show runs as a service whose lesson block teaches one manifesto chapter, discussed in the Tank the same evening and shown as preached on the Atlas.
Email engagement pipeline
Live since May 2026Subscribe → double-opt-in verify → tier-triggered drip → weekly broadcast on a 33-week calendar, with send-log and open/click telemetry. The Sunday Discord question is mechanically pinned to the email week.
F.R.E.S.H. Stream Center
Live in production for every show OBS plugin powering the Wednesday, Friday and Sunday shows: multi-stream broadcasting, chat aggregation, and audio entrainment with per-session logging via EntrainmentLogger.
Synaptiq
In developmentAudio/cognitive-science toolkit with three layers: VST commercial, cognitive-state research, and quantum-cognition speculative — the research surface for audio-mediated cognitive-state intervention.
Falsifiable by design
The hypotheses
Eight hypotheses across seven families, each with named measurement infrastructure and an honest status. Compressed here — the canonical registry with full constraints, design notes, and corrections is the research statement (v0.6, 2026-09-11).
Tier-gated progression with reviewed written reflections improves argument quality and engagement durability, assessed against within-community baselines. Its activation premise measured null (317 asynchronous pushes → 0 completions); the live test now runs through H2.
Activation premise null · runs through H2A weekly calendar — one manifesto chapter taught at the Sunday service, discussed in the Think Tank the same evening, shown as preached on the Atlas — with advancement earned by witnessed participation across a season converts room presence into tier progression where self-directed curriculum did not. Revised in v0.6; pre-registered as Arc 1 (six Sundays; three members keeping the season is the bar).
Pre-registered · Arc 1 starts when Sermon 001 is datedDefined audio interventions (40 Hz gamma stimulation, HRV-paced breathing, binaural protocols, pink/brown noise) produce measurable changes in attention, working memory, and self-reported cognitive state versus silent control.
In developmentTargeted cognitive-state interventions can partially counteract attention and cognitive-function decrements associated with environmental neurotoxin exposure — designs restricted to exposures with live biomarkers.
Partnership-gatedStructured iterative collaboration between human operators and large language models compounds decision quality and execution velocity across iteration count — the production AI-evaluation layer is the current exhibit.
Accumulating dataMembers reached across coordinated channels — the weekly email broadcast and the Discord question-of-the-day — show longer activity tails and higher subsequent response rates than single-channel members.
Accumulating dataCoordinated same-chapter exposure across both channels improves concept retention, scored on question responses against matched control chapter pairs.
Partnership-gatedSustained exposure to explicit calibration rituals — verified reflections, post-game review huddles, institutionalized debiasing checks — shifts member discourse from moral-essentialist attribution toward mechanism-structural attribution, measured within-member against pre-exposure baselines.
Registered — instrument in developmentWhat we say plainly
We would rather under-claim and be checked than over-claim and be caught.
- The numbers, generated from the live database (as of 2026-09-11). 98 registered accounts, 2 above Observer. 12 members active in the last 30 days, 8 in the last 7 (member activity only — system-grant events excluded). 35 approved curriculum completions ever, from 7 members; none in the last week. 117 Think Tank sessions since May 9, 2026, 13 distinct attendees ever, 9 in the last 30 days. This is a thirteen-person room with a hundred registered accounts, and we say so. (The block is regenerated from the instrument on a stamp; a stale stamp fails our own doc gate.)
- A documented null result, run to completion. Through May 2026 the instrument recorded essentially zero organic tier advancement — passive onboarding into an instrumented progression system produced near-zero activation. One advancement followed the June–July countermeasures (a Theorist on June 29, 2026, under a beta-tester exception to the full content gate). Then the countermeasures ran to completion: 317 asynchronous pushes delivered, no curriculum completion attributable to any of them, and a pre-registered three-week Think Tank lesson ritual also null. The one conversion ever recorded was preceded by 191 voice-minutes in the room, so the program now runs through the room (H2). A correction, stated: our July figure of eighty-six active members counted launch-day system grants as member activity; the member figure was 13.
- We audit our own instrument and say so. The evaluation layer's first version scored reflections without the source text and discarded rejected attempts; fixed June 30, 2026, and disclosed — pre-fix scores are flagged in any analysis. A June 2026 recalibration of the dimension economy (passive presence no longer mints advancement value) came with a one-time reset; pre- and post-recalibration telemetry are treated as separate measurement regimes.
- Unpowered, and we say so. At current scale no between-group hypothesis is adequately powered. Claims are gated on declared power thresholds — within-subject analyses at roughly 300 weekly-active members, between-cohort designs at roughly 1,000+ — and scale with the data, not ahead of it.
- No IRB relationship yet. Telemetry is platform-operations data; no member has been enrolled as a human research subject. H4 and H6b are explicitly gated on partner-institution IRB supervision. Members opt in to research use of their data with
/research consent(revocable at any time) and can have their data erased on request — both built September 2026; nothing is analysed for publication without consent on file. - Pre-registered before inspection — and honest about the registry. Analyses are pre-registered in a git-timestamped registry before any data inspection. The registry was stood up in June 2026 and held no entries until September 10, 2026, when the July Think Tank experiment was filed retroactively from its July 17 prediction text (the entry says so) and the next test, Arc 1, was pre-registered before it runs.
Where we partner
Partnership categories
Coordination science & complex systems — the progression instrument as a small-scale testbed where coordination interventions are deployed and measured against live telemetry.
Cognitive neuroscience & audio intervention — Synaptiq as a research surface for audio-mediated cognitive-state intervention, with hypotheses tied to the existing literature.
Quantum biology & cognition (speculative frontier) — presented as bookmark-worthy futurism, not established science, and carrying no near-term claims.
AI safety & human-AI coordination — a live small-scale instance of the human-AI coordination question, including an AI evaluation layer running inside the production instrument.
Civilizational-scale planning & Type I infrastructure — coordination instruments scaled toward planetary problems on the long-arc Kardashev Type I target.
The four asks
What we’re asking for
Research collaboration — co-designed measurement protocols on the live instrument, joint telemetry analysis, co-authored peer-reviewed publication.
Grant co-application — OD9 LLC as operator of the live instrument; partner institutions provide academic credentialing, IRB processes, and methodological rigor.
Visiting researcher access — participant-observer engagement with the live community and anonymized telemetry under appropriate ethical and data-protection arrangements.
Methodological consultation — research-methodology, IRB-process, and publication-norm expertise to accelerate the transition from operational instrument to peer-reviewed output.
Start the conversation
Academic and partnership inquiries, methodological critiques, and corrections are all welcome — critique is a contribution here, not a threat.
research@offda9.comFull research statement (v0.6, September 2026) and complete telemetry/schema inventory available on request.