# Evolution 123 Baseline Comparison Tables

## Part III — Open-World Capability Acquisition (32 Unfamiliar Families, 160 Sealed Instances)
| Arm | Solved | Accuracy | 95% Wilson CI | Mean Probes/Task |
|---|---:|---:|---|---:|
| `1_no_learning` | 5/160 | 0.0312 | [0.0134, 0.0711] | 0.0 |
| `2_frozen_e120` | 0/160 | 0.0000 | [0.0, 0.0234] | 3.0 |
| `3_retrieval_only` | 33/160 | 0.2062 | [0.1508, 0.2755] | 5.0 |
| `4_no_abstraction_synthesis` | 29/160 | 0.1812 | [0.1293, 0.2482] | 5.0 |
| `5_no_causal_model` | 75/160 | 0.4688 | [0.3931, 0.5459] | 4.719 |
| `6_random_policy` | 18/160 | 0.1125 | [0.0724, 0.1708] | 1.0 |
| `7_compute_matched_baseline` | 30/160 | 0.1875 | [0.1346, 0.255] | 3.844 |
| `8_full_cain_e123_organism` | 145/160 | 0.9062 | [0.8511, 0.9424] | 5.719 |

## Part IV — Causal World Model (48 Confounded / Shifting / Unpredictable Worlds)
| Arm | Correct | Accuracy | 95% Wilson CI |
|---|---:|---:|---|
| `1_passive_pattern_matching` | 1/48 | 0.0208 | [0.0037, 0.109] |
| `2_compute_matched_passive_2x` | 0/48 | 0.0000 | [0.0, 0.0741] |
| `3_frozen_e120_binary_model` | 15/48 | 0.3125 | [0.1995, 0.4533] |
| `4_random_interventions` | 24/48 | 0.5000 | [0.3639, 0.6361] |
| `5_e123_causal_world_model` | 44/48 | 0.9167 | [0.8045, 0.9671] |

## Part VII — Real Local Neural Generalization (`Qwen2.5-0.5B/1.5B`, 0 Weight Updates, n=24)
| Arm | Solved | Accuracy | 95% Wilson CI |
|---|---:|---:|---|
| `A_base_model_alone` | 0/24 | 0.0000 | [0.0, 0.138] |
| `B_base_plus_retrieval` | 0/24 | 0.0000 | [0.0, 0.138] |
| `C_base_plus_causal_memory` | 1/24 | 0.0417 | [0.0074, 0.2024] |
| `D_base_plus_learned_abstractions` | 7/24 | 0.2917 | [0.1491, 0.4917] |
| `E_base_plus_complete_cain_organism` | 7/24 | 0.2917 | [0.1491, 0.4917] |
| `F_frozen_previous_cain_e120` | 0/24 | 0.0000 | [0.0, 0.138] |

## Part X — Collective Intelligence (Matched Budget=12/Task, n=48 Heterogeneous Tasks)
| Architecture | Solved | Accuracy | Mean Quality | False Consensus | Coord Overhead |
|---|---:|---:|---:|---:|---:|
| `1_single_specialist` | 22/48 | 0.4583 | 0.5938 | 0 | 0 |
| `2_independent_agents` | 2/48 | 0.0417 | 0.1833 | 35 | 48 |
| `3_centralized_orchestration` | 41/48 | 0.8542 | 0.9236 | 0 | 96 |
| `4_hierarchical_specialist_team` | 39/48 | 0.8125 | 0.9028 | 0 | 144 |
| `5_adversarial_review` | 48/48 | 1.0000 | 1.0000 | 0 | 144 |
| `6_evidence_based_consensus` | 47/48 | 0.9792 | 0.9931 | 0 | 96 |
| `7_dynamic_team_composition` | 48/48 | 1.0000 | 1.0000 | 0 | 80 |
