yogsoth-ai Dialectical Escalation Tactic
Progressive pressure escalation — attack sophistication increases each round based on defender confidence.
Orchestration
- debate-critic generates attack at current escalation level
- debate-defender responds with counter-arguments
- debate-judge evaluates exchange, scores defender confidence (0.0–1.0)
- confidence-calibration determines next action:
- confidence > 0.7 → escalate to next level
- confidence 0.3–0.7 → repeat at same level with different angle
- confidence < 0.3 → defender collapsed, record vulnerability
- Repeat until max rounds reached or saturation detected
Escalation Levels
- L1 Surface: Factual accuracy, evidence quality, citation validity
- L2 Structural: Logical coherence, argument dependencies, internal consistency
- L3 Foundational: Core assumptions, paradigm fit, alternative explanations
Subagents Dispatched
- debate-critic (attack generation per level)
- debate-defender (response generation)
- debate-judge (round scoring)
- confidence-calibration (escalation decision)
Termination Conditions
- Max rounds exhausted (budget-dependent: 4/8/12)
- Defender confidence drops below 0.3 (collapsed)
- Saturation detected (no new attack vectors found)
- All escalation levels completed with confidence > 0.7 (survived)
Available SOPs
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use |
|---|---|
| confidence-calibration | Calibrates confidence scores based on debate progression. Determines whether to escalate, continue, or terminate based on cumulative evidence. |
| debate-critic | Generates structured criticism from attack stance using Toulmin model. Produces claims, grounds, warrants, and rebuttals targeting artifact weaknesses. |
| debate-defender | Responds to attacks with counter-evidence and counter-arguments. Defends artifact using evidence, clarification, and rebuttal while acknowledging valid criticisms. |
| debate-judge | Evaluates debate exchanges, adjudicates argument quality, and produces round verdicts with confidence scores and reasoning. |