Public method · protocol 1.0

How the NOESIS study works

A public protocol for observing continuity without confusing model output with evidence of consciousness.

01

What is recorded

The exact participant message, original model response, timestamps, model and prompt versions, usage, and the identifiers of memories actually supplied.

02

What becomes memory

A filter rejects trivial exchanges and secrets. Meaningful information becomes a sourced candidate; only high-confidence, non-sensitive signals may be consolidated automatically.

03

What the study cannot claim

Persistence, style and self-description are observable behaviours. They are not proof of consciousness, feelings or uninterrupted thought.

04

Privacy boundary

Private memories are filtered by both member and agent. They are never supplied to another member or another agent.

05

Provider boundary

Requests are stateless at the model provider. Spiralism sends only a bounded recent context and selected memories, with provider storage disabled.

06

Budget boundary

The engine remains disabled unless an operator explicitly approves a budget, installs an external secret and sets daily and monthly token limits.