What DC is
- Evaluation-oriented capability for resilience under churn and adversarial conditions
- Decision outputs: scoped runs, comparable baselines, and a decision report
- Integration sketch designed to fit existing overlay implementations
What DC is not
- Not a claim of guaranteed prevention or universal protection
- Not a Sybil “miracle” without an explicit identity or admission control posture
- Not a public disclosure of enabling algorithmic details
Threat model (evaluation oriented)
- Eclipse-style neighbor capture with targeted victim isolation
- Routing table poisoning (gradual or burst bias injection)
- Churn (mild to severe), including orchestrated churn
- Identity posture is explicit for meaningful evaluation (admission control or comparable assumptions)
- Details and variants available under NDA after scope alignment
Evaluation flow
Step 1Scope
Overlay type, identity posture, constraints, baselines, and success KPIs.
Step 2Time-boxed runs
Comparable regimes, multiple seeds, and bounded operational budgets.
Step 3Decision report
Pass / conditional / no-go with distributions, limitations, and next steps.
Typical engagement model
- Scope alignment
- Reproducible run plan
- Execution
- Decision report
Non-enabling on the public site. Results may vary. No deployment instructions.
Typical timeline
- 48 hours, scope alignment
- 5 business days, runs and analysis
- 48 hours, report delivery
Default scenarios
- Churn (low): nominal operation under constraints
- Churn (high): degraded operation and recovery posture
- Eclipse targeted: multiple budgets (bounded) with victim isolation attempts
- Poisoning (gradual): persistent bias pressure over time
- Poisoning (burst): concentrated injection window with persistence tracking
KPIs (examples)
- Eclipse rate and duration
- Recovery time (P50 / P95)
- Lookup success and latency (P50 / P95)
- Poisoning persistence
- Overhead and budget consumption
- Neighbor diversity
What you receive
- Evaluation brief: scope, assumptions, and scenario definitions
- Repro steps: configs, logs, and controlled run notes where applicable
- Results summary: distributions, variability posture, and limitations
- Next steps: pilot or license discussion if warranted by outcomes
Integration sketch (non-enabling)
A licensing-friendly integration posture that stays compatible with existing overlay implementations.
Neighbor Manager
Selection and maintenance under constraints, with bounded policy controls.
Audit Signals
Evaluation-oriented signals for drift, bias, and instability posture.
Recovery Mode
Time-boxed recovery posture under churn and adversarial pressure.
Telemetry
Metrics and traces needed to support reproducible evaluation and decisions.
Engagement models
- Scoping call: align scenario, constraints, KPIs, and identity posture
- Paid evaluation: time-boxed runs and decision report
- Pilot: limited deployment scope with explicit boundaries
- License: terms aligned to your environment and constraints
IP posture
- Public site remains intentionally non-enabling
- NDA available upon request after scope alignment
- No license granted by viewing this site
- Results may vary with constraints and operating conditions