agent-intelligence.org
What is happening in agent intelligence?
Evidence before consensus
See the shape.
Keep it falsifiable.
Preconvergence is an evidence-disciplined method for investigating independently observed signals before a shared structure becomes broadly recognized.
OBSERVED→INDEPENDENCE→ALIGNMENT→PRECONVERGENCE→ESCALATION→FALSIFIER
The method
A progression, not a certainty score.
-
01
OBSERVED
What happened?
Record the event and its source boundary before interpretation.
-
02
INDEPENDENCE
Are these genuinely separate signals?
Separate evidentiary origins; account for shared lineage and overlapping windows.
-
03
ALIGNMENT
What direction are the independent signals beginning to share?
Name the common direction without turning co-occurrence into causation.
-
04
PRECONVERGENCE
Is a structure visible before broad recognition?
State the emerging topology as a bounded hypothesis, not a settled category.
-
05
ESCALATION
What would confirm convergence?
Publish a dated, observable condition that would strengthen the thesis.
-
06
FALSIFIER
What would break the thesis?
Publish evidence that would weaken or invalidate the proposed structure.
Hard boundaries
The hypothesis must remain smaller than its evidence.
- Observation is not convergence.
- Repeated reporting from one origin is not independent confirmation.
- Alignment is not causation.
- A forecast without an escalation condition and a falsifier is incomplete.
- An operator's domain registration is not market validation.
- Preconvergence is a publication method here, not a claimed standard or canonical category name.
A distinct altitude
Related domains answer different questions.
poly-agents.com
How can multiple agents reason or operate together?
preconvergence.com
What structures are forming before they become obvious?