Primary bottleneck
SELECT TO FILTERH = human · A = AI · L = shared loop · X = external · NONE = no primary bottleneck coded. Categories reflect the record's annotation, not an objective diagnosis.
PUBLIC RESEARCH LABORATORY EXPERIMENT BEGAN 14 SEPT 2026
Evidence,
under inspection.
What changes when one human and AI keep working together—and how could we tell?
—STRUCTURED PUBLIC RECORDS
—CODED EXTERNAL VERIFICATIONS
—CODED NEW-CAPABILITY EVENTS
—LATEST RECORD DATE
Reading the latest published structured record. Figures update only after approved publication; this is not a live activity feed.
00 / THREE INSTRUMENTSNot more commentary.
Not more commentary.
Better questions for the evidence.
I / READ The evidence observatory
Inspect what was recorded, how it was classified, and what the classifications cannot tell us.
STRUCTURED RECORD → II / CHALLENGE The hypothesis benchLook at claims with open questions, counter-explanations, and tests the experiment still owes.
CLAIM ≠ CONCLUSION → III / TRACE The evolution traceFollow linked Field Notes from an early observation to later boundaries, corrections and operating rules.
SOURCE → REVISION →01 / EVIDENCE OBSERVATORY
What the record actually says.
Counts are descriptions of coded public records, not effect sizes, productivity gains or causal estimates. Select a category to investigate the underlying entries.
Verification status
CODING, NOT CERTAINTYExternal coding denotes an externally grounded artifact or outcome in the source record. It does not establish independent experimental replication.
Recorded outcomes
SELECT TO FILTEROutcome labels classify what an entry recorded; observational is not an inferior result. Legacy values are retained rather than silently rewritten.
RETROSPECTIVE CLASSIFICATIONS—
CONTEMPORANEOUS CLASSIFICATIONS—
MIGRATION WARNINGS—
How to interpret these numbers 02 / WORKING HYPOTHESES
What would change our minds?
Selected open questions, including a current interpretation grounded in Entries 146–149. These are research prompts, not verified findings; proposed comparisons have not been run.
Loading selected hypotheses…
Current hypothesis added . The wider Lab interpretation and coding remain dated snapshots; the structured evidence feed above updates separately. Trace the research question ↗
03 / EVOLUTION TRACE
A concept is not a fixed object.
Choose a research trajectory. Each point links to a published Field Note; the sequence is a curated reading path, not proof of causation.
TRACE / —
Choose a trajectory
Public Field Notes become more informative when read as sequences of revisions rather than isolated insights.
Editorial selection last reviewed 6 October 2026. Each link opens the underlying public note.
04 / PUBLIC RECORD EXPLORER
Inspect the source layer.
Filter the approved public dataset. This is a projection of the private analytical ledger, not a disclosure of private records or a complete log of every judgment.
Loading records…SELECT A ROW TO READ THE FIELD NOTE ↗
Loading the published source index…
Source: public Loop Records JSON ↗ · schema ↗ · full Field Note archive ↗
05 / RESEARCH INTEGRITY
The limits are part of the result.
Observation is not automatic publication.
The private ledger is upstream. A public Field Note requires the established human publication decision; this site displays only approved projections.
Publication method ↗Historical coding is not contemporaneous measurement.
Entries through 117 were annotated retrospectively. Later entries use the structured record format when published. Known migration anomalies remain visible.
Inspect schema ↗One case does not prove general effects.
There is no randomized comparison group. Patterns and hypotheses are interpretations of one evolving collaboration, not population estimates or evidence that a mechanism will transfer.
Explore research analysis ↗THE WORKING PRINCIPLE
Make the change visible. Keep its source attached. Leave room for the record to prove us wrong.
Return to Lifted