jordan/stemedb
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
3cfaa1e1d3
stemedb/ai-lookup/features/simulation.md
jordan 3cfaa1e1d3 feat: Complete Phase 1 (The Spine) - storage foundation 
Phase 1 delivers the complete durability and storage layer:

- WAL with crash recovery: Append-only journal with BLAKE3 checksums,
  fsync guarantees, and proper seek-to-EOF on reopen
- Storage engine: sled-backed KVStore with scan_prefix for range queries
- Content-addressed storage: H:{hash}, V:{hash}, E:{hash} key patterns
- Ingestor: Background worker tailing WAL, writing to KV with 8-byte
  aligned record headers for rkyv zero-copy deserialization
- Comprehensive tests: 31 tests covering crash recovery, round-trips,
  and multi-cycle durability

New crates: stemedb-wal, stemedb-storage, stemedb-ingest

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-31 14:15:34 -07:00

2.0 KiB
Raw Blame History

Simulation System ("The Infinite Game")

Last Updated: 2026-01-31 Confidence: High

Summary

The Simulation is an Agent-Based Modeling (ABM) environment that validates StemeDB under emergent, adversarial, and evolutionary pressure. It simulates a society of AI agents with conflicting goals living within the knowledge graph.

Key Facts:

  • Codename: "The Arena"
  • Purpose: Integration testing via societal stress tests, not unit tests
  • Architecture: stemedb-sim binary orchestrating agent swarms via tokio
  • Agents communicate only through StemeDB reads/writes

File Pointer: /simulation-vision.md (full vision document)

Agent Personas

Persona Goal Behavior
Scientist Converge on truth High-confidence assertions, cites sources
Troll Sow chaos Low-confidence contradictions, frequent forks
Believer Amplify consensus Trusts high-reputation agents
Skeptic Find variance Reduces confidence of unverified claims
Historian Preserve context Resurrects dormant truths with new evidence

The Gameplay Loop

  1. Assertion: Agent reads ground truth, creates assertion
  2. Fork: Adversarial agent forks reality with contradiction
  3. Lens Resolution: Query agent applies Lens (e.g., Consensus)
  4. Reputation Update: TrustRank adjusts agent reputations
  5. Decay: Unverified assertions fade via Dormancy Protocol

Success Criteria

  • Truth survives: High-reputation assertions outlive spam
  • Lenses work: Consensus lens filters Troll noise
  • Performance: 1000 concurrent agents without locking
  • Emergence: Trust clusters form without hardcoded rules

Metrics

Tracked via Prometheus/Grafana:

  • global_truth_convergence - Entropy of the graph
  • agent_reputation_distribution - Reputation spread
  • fork_depth_max - Deepest branch depth

Related Topics