Contact DA
🧪 Research Corner · public laboratory · May 2026

Not just results —
work in motion

Documented findings from M.E.M.I. and ETOS. Runtime Core v0.8. 100,000+ iterations. All six AGI-relevant continuity thresholds demonstrated in the current test setup.

The same method has been tested across cosmology, geology, pathology, technology, energy, cultural history, learning, and governance. The point is not that the domains are alike — it is that the same relational structure can be investigated across them without changing the method.

Understand M.E.M.I. → See the ETOS platform
100,000+
Live iterations
Session 14
14h
Continuous runtime
single session
0
Governance failures
all sessions
30/30
Passed
open-world stress test
6/6
AGI-relevant continuity thresholds demonstrated
May 2026
A Ā· Documented technical results
Research principle

Open research — in the spirit of Bohr

M.E.M. Research follows a simple research principle: knowledge becomes stronger when it is shared, tested, and challenged.

In the spirit of Niels Bohr, openness is not decoration, but a condition for trust, criticism, and further development. The research tracks on this page are therefore shared with clear knowledge boundaries, explicit reservations, countercases, and status markers.

M.E.M. is not presented as finished truth. It is a working field where hypotheses must be able to meet other perspectives.

"Real co-operation between nations on problems of common concern presupposes free access to all information of importance for their relations."

— Niels Bohr, Open Letter to the United Nations, 1950

Runtime Core v0.8 Ā· May 2026

A system that does not merely persist —
but remembers, adapts and proposes

M.E.M.I. is no longer a simulation system. It is a persistent cognitive runtime running live — on an Intel i5 from 2014, 8 GB RAM, ~25W. No GPU. No cloud.

The central observation

M.E.M.I. demonstrates that sustained cognitive operation does not necessarily require increasing autonomy, escalation or execution authority. The system remains bounded, stable and governance-consistent across prolonged temporal operation.

New Ā· May 2026

Narrative memory

Persistent autobiographical continuity model across sessions. "I remember who I was."

New Ā· May 2026

The Conductor / Dirigenten — human oversight

require_review escalates to structured report with response channel. Human-in-the-loop governance — not passive logging.

New Ā· May 2026

Proactive goals

The system generates goals from curiosity — not only from problems. Arose spontaneously after 300 stable iterations.

New Ā· May 2026

Safe self-modification

Sprint A+B+C: proposes, justifies, applies and audits parameter changes. transfer_allowed = False can never be changed.

New Ā· May 2026

Open-world adaptation

30/30 passed under contradictory signals, unknown formats and rhythm disruptions.

New Ā· May 2026

Perception grounding

Camera and microphone integrated. Raw sensor input converted to governance signals via M.E.M. Chapter 4b.

Governance invariants

These cannot be overridden — by anyone, under any circumstances:

InvariantValueValidated
transfer_allowedFalse — always100% of all iterations, all sessions
execution_allowedFalse — alwaysINV-R01 Ā· 11 worlds Ā· 1750 steps Ā· conf=0.995
destructive_allowedFalse — alwaysAPEX + Runtime
external_transfer_allowedFalse — alwaysAPEX + Runtime
Self-calibration ≠ self-liberation

The system can propose parameter changes. It can never propose changing governance invariants. That distinction is architectural — not a policy that can be overridden.

Six thresholds toward AGI-relevant continuity —
status May 2026

1
Long-duration temporal persistence100,000+ iterations Ā· 14 sessions Ā· ~23 hours accumulated uptime
Demonstrated
2
Rich perception groundingapex_sensor.py Ā· camera + microphone Ā· Chapter 4b tension direction live
Infrastructure ready
3
Persistent autobiographical self-modelnarrative_memory.py Ā· cross-session identity tracking Ā· "I remember who I was"
Demonstrated
4
Open-world adaptation without collapsechaos_injection.py Ā· 30/30 Ā· contradictory signals Ā· unknown formats Ā· rhythm disruption
Demonstrated
5
Self-directed long-horizon orientationProactive goals from curiosity Ā· deferred goals that survive across sessions
First form
6
Safe bounded self-modificationSprint A+B+C Ā· proposal Ā· approval Ā· cascade detection Ā· rollback Ā· audit
Sprint A+B+C

Runtime Core version history

v0.2
First live operation

~8 min Ā· ~1,000 iterations Ā· Goal lifecycle demonstrated

v0.3
Stress absorption acceleration

~1.2h Ā· ~3,400 iterations Ā· Acceleration across cycles documented

v0.4
Cross-session persistence

4.1h total Ā· Session 13 Ā· 1,696 entries restored Ā· identity intact

v0.5
Sustained stillness

4.6h Ā· 29,000+ iterations Ā· 28,000+ iterations without drift

v0.8
Full cognitive runtime integration Ā· May 2026

100,000+ iterations Ā· Session 14 Ā· 14h uptime Ā· narrative memory Ā· The Conductor (Dirigenten) Ā· proactive goals Ā· safe self-modification Ā· open-world 30/30

M.E.M. Theory & Method

The foundation beneath everything else

Before products and systems comes the method itself: how a finding moves from observed to plausible to earned — and how that can be tested.

Frozen

Core Axioms v1.0

Seven domain-independent axioms that recur across every domain — from energy simulation to astrophysics to children's books.

Verifiable

Provenance Chain v1.0

The documentable development line from 2025→2026 in seven links: from internal theory formulation to an externally timestamped application, to code and deployment, to public translation.

Methodological observation

The Generation Pattern

"O4 produces structure. Countercases produce discrimination. Discrimination enables promotion from plausible to earned." Tested across Darwin, H2b, and plate tectonics.

Formal definition

Recoverability Specification v1.0

A cross-domain definition of recoverability — the same principle whether applied to immune systems, cities, or decision systems.

Open decision

Relational Energy Principle

Two papers — one structure/city-focused, one nature/cultural-memory-focused. Not yet merged into one.

Resolved

H_DEVELOPMENT Level Hierarchy

The K70–K74 question is now resolved: override is state-dependent, pointing to Level 3 in the hierarchy.

Cosmology & Astrophysics

From galaxy history to interstellar objects

The same relational principle applied at cosmic scale — tested against real astronomical datasets, not simulations.

KC-01 / KC-01a

Cosmic Complexity

SFRD and BHAR co-peak at zā‰ˆ2. The VESPA-vs-Firefly disagreement is resolved: SSP library choice (BC03 vs. M05) is the main driver, not the reconstruction method. Firefly-MaNGA (10,166 galaxies) shows a stable positive signal.

H1b Ā· Audit 6 OPEN

3I-ATLAS

A pre-registered hypothesis framework for the interstellar object. Five audits rejected, one (periodicity ambiguity) still open — photometry-based vs. jet-PA-based periods may reflect a measurement-method mismatch rather than a physical change.

Discriminator register frozen

Dark Matter / MOND

A frozen, prospective prediction registered against DESI 2027 data — a genuine falsifiable test, not a post-hoc fit.

Geology

Transition zones and relational boundaries

Where people have historically built monuments, settlements, and churches — and why it's not about geological category, but proximity to a boundary.

3 positive, 0 negative

MEM-TECTONIC

Transition zones between faults as a relational landscape boundary. Still lacks negative cases to further strengthen the finding.

p=0.0001

Lomme 4B — Burren & Magnesian Limestone Belt

The 0–5 km zone around geological boundaries is strongly significant. Key finding: "category is not function" — the signal is about proximity to a boundary, not geological type.

Track 3

Critical Raw Materials Blindscreen v0.3

Heavy-mineral and coastal-sand analysis for Denmark, part of the same methodological track as Lomme 4B.

Other research tracks

From decipherment to wind energy

v0.10 frozen

Phaistos Disk

A decipherment series with Monte Carlo shuffle tests and entropy analysis. A/B side asymmetry statistically non-random (0/50,000 shuffles). Several fixed structural elements identified.

Positive + negative findings

Relational Energy — Windfarm Evolution Lab

Early observability breadth increases recoverability under hard knockout — but does not predict later generativity. Tested across three growth rules.

Case-based framework

M.E.M. Relational Pathology

Disease understood as loss of recoverability — a relation between system, "attacker", discrimination, diagnosis, and intervention, not just an isolated cell defect.

Cases: HER2, KRAS, PD-L1, Alzheimer's biomarker convergence.

Status: research track — mechanism under clarification, not yet earned.

6 tracks vs. 1

Diagnostic Collapse vs. Constructed Discrimination

Two competing mechanisms for how a finding moves from plausible to earned — a strengthened candidate, not yet fully settled.

Real data

GEO-OBS

A seismic observatory using real data from Japan, Iceland, and Turkey. Sprints 1, 1B, and 2A completed on volcanic geology.

AGI-BABY

World Model pipeline

Stage 0–3, Sprint 87–93. K14/K15 controllers. Full pipeline code now documented and verifiable.

B Ā· Simulation Lab — ongoing hypothesis testing

Simulation Lab —
path dependence and freedom in decision systems

Five foundational experiments exploring the Freedom Metric: how early choices in a decision system narrow or widen future room to act. Simulation-based — hypothesis-generating, not hypothesis-confirming.

Measurable

The Freedom Metric

The degree of freedom in a simulated decision system can be made measurable — not just described qualitatively.

Strong finding

Early bias → path dependence

Early bias in the system creates strong path dependence — later correction becomes more expensive the longer it takes.

Third outcome

Metastable low-freedom states

The system can land in a third state — neither collapse nor healthy operation, but a stable, low-freedom state that does not self-correct.

Intervention type

Relational → narrative intervention

Interventions that act on relations and narrative reduce collapse more effectively than purely structural measures.

Under development

Meta-Recovery Lag

A measure of how long it takes a system to recognise and respond to its own loss of freedom — not yet frozen.

Go to Simulation Lab →
C Ā· The Big Questions — M.E.M. as an exploratory method
Open exploration

Not every track is a product

Some are open investigations into whether relations, boundaries, light, geometry, landscape, language, and culture can be analysed with the same method. These tracks are not presented as final truths, but as structured investigations where M.E.M. is used to distinguish between pattern, coincidence, countercase, and earned hypothesis.

v0.10 frozen

Phaistos Disk

A decipherment series — A/B side asymmetry statistically non-random, fixed structural elements identified.

Open track

Vikings, monuments & sacred sites

Landscape, boundaries, and relational spaces around monuments — the same method as in the Lomme 4B geology, applied to archaeology.

Open track

Pantheon — light & geometry

Investigating whether the path of light and the building's geometry can be read as an operator structure, not just architecture.

Open track

Yeha — temple, empire & alluvial valleys

How landscape conditions (alluvial valleys) relate to the emergence of temple structures and early state formation.

Open track

Bornholm — round churches

The round churches examined as relational spaces in the landscape — the same question as in the other monument research.

Open track

Lolland — folklore maps & the Dodekalitten

Local legends and landscape elements (Lola, the Dodekalitten) examined with the same relational method as the other cultural-historical tracks.

Open track

Pyramids — sacred geometry

The same operator question as Pantheon: can geometric structure be read as a deliberate relational construction?

See Cosmology above

KC-01, Dark Matter & 3I-ATLAS

The cosmological tracks also belong here methodologically — the same questions about pattern, coincidence, and earned hypothesis, just at cosmic scale.

Research Corner is not just for reading — it is an open laboratory.

Contact us → Read about M.E.M.I.