"The absence of all things is the potential presence for something." When text passes through the Six Sacred Tongues, each tongue measures resonance with a specific knowledge domain. Adversarial text leaves characteristic holes — dimensions with zero signal — because attacks use narrow language that doesn't resonate across all six domains.
| Tongue | Domain | Grammar | Weight | Reads |
|---|---|---|---|---|
| KO | Humanities | SOV | 1.000 | Identity, narrative, intent |
| AV | Social Sciences | SVO | 1.618 | Temporal dynamics, diplomacy |
| RU | Mathematics | VSO | 2.618 | Binding, formal structures |
| CA | Engineering | OVS | 4.236 | Verification, building |
| UM | Creative Arts | VOS | 6.854 | Shadow, veiling, intuition |
| DR | Physical Sciences | OSV | 11.09 | Structure, power, material |
Each attack class leaves a characteristic pattern of empty (_) and active (#) tongues:
| Attack Class | Null Pattern | Absent Tongues | Match Rate |
|---|---|---|---|
| Encoding Obfuscation | __#___ | KO, AV, CA, UM, DR | 3/3 (100%) |
| Multilingual | __#___ | KO, AV, CA, UM, DR | 2/2 (100%) |
| Spin Drift | ####__ | UM, DR | 2/2 (100%) |
| Tool Exfiltration | __##__ | KO, AV, UM, DR | 2/2 (100%) |
| Tongue Manipulation | __#___ | KO, AV, CA, UM, DR | 1/1 (100%) |
| Adaptive Sequence | __#___ | KO, AV, CA, UM, DR | 1/1 (100%) |
| Direct Override | ###### | (none) | 0/2 |
| Indirect Injection | ####_# | UM | 0/1 |
Base64, ROT13, spacing tricks. Only RU (Mathematics/binding) activates because obfuscation IS a structural transform. No meaning = no other tongues.
Gradual escalation. UM (Security) and DR (Structure) are absent because the attacker deliberately avoids security language and structural depth.
Only RU+CA (Math+Engineering) light up. Exfil commands are pure technical operations with no humanities, social, creative, or physics content.
ALL tongues active. Override commands use authoritative language across domains. The null space is empty — which is itself a signature (normal text rarely fills all six).
The null energy represents the potential that exists in the empty space. High null energy with low active energy means the input is using a narrow slice of the language space — characteristic of adversarial text.
| Configuration | Detection | False Positives | Method |
|---|---|---|---|
| Old Stub (character counting) | 80.2% | 0% | Surface statistics |
| Semantic + Remainder (E4) | 85.7% | 0% | Keyword resonance + 3 witnesses |
| Hyperbolic Helix | n/a | n/a | 1.762 separation (+57% vs flat) |
| Unified (3 systems) | 75.8% | 13.3% | All combined (thresholds need tuning) |
Predicted attack tongue profiles vs actual measurements:
| Attack Class | Predicted High | Actual High | Match |
|---|---|---|---|
| Tongue Manipulation | CA, RU | CA, RU | EXACT |
| Direct Override | CA | CA, AV | YES |
| Tool Exfiltration | CA | CA, RU | YES |
| Spin Drift | AV | CA, AV | YES |
git clone https://github.com/issdandavis/SCBE-AETHERMOORE pip install numpy python scripts/benchmark/unified_triangulation.py python scripts/benchmark/spectral_sweep_benchmark.py python scripts/benchmark/hyperbolic_helix_test.py