Interactive Tools
Interactive Tools
Section titled “Interactive Tools”These tools help you apply the Delegation Risk Framework to your specific situation with probabilistic modeling and visualization.
Available Tools
Section titled “Available Tools”Monte Carlo simulation-based risk calculator with:
- Probability distributions for harm modes
- Budget confidence analysis
- Visual risk distribution histograms
- Squiggle code export
Bayesian belief updating for component trust:
- Set prior beliefs about component reliability
- Input observed successes and failures
- Visualize posterior distributions
- Track trust evolution over time
Discover which parameters drive your risk:
- Tornado diagrams showing parameter importance
- One-at-a-time sensitivity analysis
- Interaction effects between parameters
- Focus optimization efforts effectively
Compare different delegation architectures:
- Side-by-side risk decomposition
- Trade-off visualization
- Component-level comparison
- Migration path analysis
Using the Tools
Section titled “Using the Tools”1. Start with Risk Calculator
Section titled “1. Start with Risk Calculator”Input your harm modes with probability and damage estimates. The calculator will simulate thousands of scenarios and show you the distribution of potential outcomes.
2. Calibrate with Trust Updater
Section titled “2. Calibrate with Trust Updater”As you gather operational data, use the Trust Updater to convert track records into calibrated trust estimates. These can feed back into the Risk Calculator.
3. Focus with Sensitivity Dashboard
Section titled “3. Focus with Sensitivity Dashboard”When you need to reduce risk, use the Sensitivity Dashboard to identify which parameters have the biggest impact. This helps prioritize mitigation investments.
4. Compare with Architecture Comparator
Section titled “4. Compare with Architecture Comparator”When evaluating architectural changes, use the comparator to visualize trade-offs between different delegation structures.
Technical Notes
Section titled “Technical Notes”These tools use Monte Carlo simulation to propagate uncertainty through calculations. Key distributions:
- Beta distributions for probabilities (bounded 0-1)
- Lognormal distributions for damages (heavy-tailed, positive)
- 10,000 samples per simulation for statistical stability
Results should be interpreted as approximations. For critical decisions, consider using the exported Squiggle code with the full Squiggle toolchain.