Slide contents

1. 1. Modeling Reliability from a Dependency Graph

   Alexander Polomodov × Anatoly Krasnovsky

2. 2. The graph narrows the chaos search

3. 3. The evidence lives in several versions

4. 4. A hand-built model ages before the system

5. 5. Telemetry becomes an executable model

6. 6. The graph sees only the mandatory path

7. 7. Monte Carlo plays out virtual incidents

8. 8. Model and system measure one outcome

9. 9. DeathStarBench is complex but controllable

10. 10. Replication matched to four decimals

11. 11. Five rates preserve the overall trend

12. 12. The error changes sign with stress

13. 13. Replicas help inside visible structure

14. 14. The model assumes binary fail-stop

15. 15. Shared fate lies outside the call graph

16. 16. Uncertainty forms the chaos backlog

17. 17. Research becomes a platform pipeline

18. 18. A pilot starts with one operation

19. 19. Before introducing chaos, you can build a model

    Paper · https://doi.org/10.1145/3786582.3786823

    Code · https://github.com/a-a-k/socialnet-resilience

    Author · https://t.me/mb3rlab

    The graph selects hypotheses. Chaos checks reality.