Observation filtered through subjectivity, made structural.
A Gleam library for content-addressed, exhaustive decision trees. Every observer has bias -- weight-shifting, legacy patterns, filters. Bias makes it typed, exhaustive, and diffable.
observable
+-- observer-A
| +-- decision-1
| | +-- action-a
| | +-- action-b
| +-- decision-2
| +-- action-c
+-- observer-B
+-- decision-3
+-- action-d
Every level is content-addressable (SHA-256). The tree is diffable at every level. Decision variants are exhaustive -- every variant has actions, no catch-all.
gleam add biasFull documentation: docs/
- What Bias Is -- the concept, the types, content addressing.
- Modules -- how the seven modules compose.
- Agent Guide -- what future agents need to know.
gleam testLICENSE.md contains the Apache-2.0 licence required by Hex. The actual
governing terms are the systemic.engineering License v1.0.