CSP as Spatial Reasoning

Structured binary encodes spatial state. Deterministic projections produce consumer-specific outputs. No training. No inference. Just bit manipulation — achieving what GNNs do with learned weights.

Q Where exactly is the shooter? 14 bytes
0101011001001010001011110001001001010110000000000000011100000100101000000000001111110000110000000000000100000010
version [0:1] offsetType [2:3] precisionTier [4:5] semanticPresent [6:6] anchorOverride [7:7] mgrsCell [8:39] structuralOffset [40:79] confidence [80:87] semanticFlags [88:95] eventType [96:103] priority [104:111]
SOURCE FIELDSDERIVEDPROJECTIONSPrecisionOperatio…Offset TypeStructur…Semantic?yesMGRS CellMGRS GZD…StructureID 7FloorFloor 4FacadeSWDepth0/15Unit3Confidence0.94EventHOSTILEPriorityIMMEDIATEPositionMGRS + offsetClassificationevent contextReliabilityconf + sourceAUDIOSpeech synthesisNAVWaypoint JSONTACMap markerFIRESFire mission
1 / 6

Header Decode

The first byte encodes 5 control fields. Structural offset mode tells us this is a building reference. Operational precision gives 0.1m resolution. Semantic flag indicates event classification follows.

BYTES 56
A GNN needs:
  • Training data: 10K+ building layouts
  • GPU inference: spatial transformer
  • Feature extraction pipeline
  • 3D point cloud processing
~50–200ms (GPU)
CSP needs:
  • 14 bytes of structured binary
  • Byte-level deterministic decode
  • Structure overlay lookup table
  • Zero training, zero weights
<1ms (deterministic)

Consumer Projections

VOX Audio Tactical

HOSTILE. Building 7. Floor 4. SW. window 4. IMMEDIATE.

CONF
94%
FALLBACK: [4A2F] 12 56
NAV Autonomous Nav
CONF
94%
FALLBACK: [4A2F] 12 56
TAC Tactical Display
CONF
94%
FALLBACK: [4A2F] 12 56
FDC Fires Digital
{
  "status": "ENGAGEMENT_DENIED",
  "reason": "Insufficient confidence: missing semantic pair (eventType/entityClass), sourceType missing or inferred-only",
  "mgrs": "[4A2F] 12 56",
  "confidence": "0.94"
}
CONF
94%
FALLBACK: [4A2F] 12 56