| Gen 0 | Score: 76.08 | Best: 76.08 | Status: baseline | Change: Agent D&C approach | |
| Gen 1 | Score: 88.02 | Best: 88.02 | Status: improved | Change: Individual disambiguation instead of pairing | |
| Gen 6 | Score: 99.28 | Best: 99.28 | Status: improved | Change: Batched disambiguation with recursive pairing | |
| Gen 7 | Score: 99.28 | Best: 99.28 | Status: unchanged | Change: Cascading disambiguation (same as Gen 6) | |
| Gen 8 | Score: 99.52 | Best: 99.52 | Status: improved | Change: Group size constraint check after unite | |
| Gen 9 | Score: 99.82 | Best: 99.82 | Status: improved | Change: Group size checks in initial pairing and odd-value phases | |
| Gen 10 | Score: 99.61 | Best: 99.82 | Status: no improvement | Change: No Phase 1 approach (processes all N values directly) | |
| Gen 11 | Score: 99.85 | Best: 99.85 | Status: improved | Change: Subset-sum uniqueness check after cascade rounds | |
| Gen 12 | Score: 100.00 | Best: 100.00 | Status: PERFECT | Change: Pre-cascade subset-sum check resolves forced groups before disambiguation queries | |