# Observation-to-Decision Stack **Entity class:** Systems concept ## Definition An **observation-to-decision stack** transforms conditions in the world into evidence, features, identities, classifications, and actions. A disciplined version keeps four layers distinct: ground truth, observed evidence, inferred state, and predicted state. It also records provenance and uncertainty at every transformation. ## ICE examples - [[wiki/ICEYE|ICEYE]]: Earth state → SAR acquisition → imagery/analytics → disaster or ISR decision. - [[wiki/Space Intelligence|Space Intelligence]] and [[wiki/ICE Commodity Traceability|ICE CoT]]: satellite observations → land-cover map → farm-plot test → supply-chain compliance decision. - [[wiki/ICEPIC|ICEPIC]] and [[wiki/FALCON Search and Analysis|FALCON-SA]]: source records → entity resolution → relationship/lead → investigative decision. - [[wiki/IBCS Integrated Collaborative Environment|IBCS ICE]]: sensor tracks → common air picture → mission-command planning and engagement decision. - [[wiki/Institute for Cross-Disciplinary Engagement|Dartmouth ICE]]: disciplinary observations → cross-domain dialogue → public reasoning, not operational command. - [[wiki/Intercontinental Exchange|Intercontinental Exchange]]: farm and supply-chain facts → normalization → geospatial/environmental validation → risk and due-diligence state → commodity-market workflow. ## Recurring ICE archetype Across otherwise unrelated entities, ICE repeatedly names the place where heterogeneous reality becomes a common representation: - [[wiki/ICEPIC|ICEPIC]] resolves people, records, and relationships; - defense and intelligence ICE environments fuse sensors, systems, and mission users; - [[wiki/Institute for Cross-Disciplinary Engagement|Dartmouth ICE]] joins disciplines and epistemic traditions; - [[wiki/ICEYE|ICEYE]] turns planetary surfaces into persistent radar observations; and - [[wiki/Intercontinental Exchange|Intercontinental Exchange]] normalizes economic and physical-world facts for markets, benchmarks, compliance, and institutional action. The shared archetype is functional rather than organizational: **integrate heterogeneous reality into a common representational environment so that something can know, decide, or act.** ## Relationships - **methods:** [[wiki/Data Fusion|Data Fusion]], [[wiki/Multi-Sensor Fusion|Multi-Sensor Fusion]], [[wiki/Entity Resolution|Entity Resolution]], [[wiki/Machine-Readable Policy|Machine-Readable Policy]], and [[wiki/Human-in-the-Loop|Human-in-the-Loop]]. - **risks:** provenance loss, false joins, model error, policy overreach, and confusing inference with fact. - **disambiguation route:** [[wiki/ICE|ICE disambiguation]]. ## Sources / Provenance - Editorial synthesis grounded in the primary system sources linked from [[wiki/ICE|ICE]].