Where is the signal?
The 62.5% result is real as a description of the saved lineups. It is not equally strong across the proposed pairs.
Those lineups reuse 11 pairings, each tested in both directions. Several figures recur across pairs, so these are not 397 independent traditions—or even 11 independent histories.
Giving each pooled pair equal weight produces 61.8%. The aggregate signal is not just an artefact of unequal completed counts. Its interpretation still depends on the sources, descriptions and competing candidates.
Open the average.
First-place selections among completed lineups. Direction rows use different candidate lineups; this is not a controlled order experiment.
| Proposed pair | First / completed | Rate | Both directions |
|---|---|---|---|
| Quetzalcoatl / Viracocha | 36/37 | 97.3% | Quetzalcoatl → Viracocha: 19/20 Viracocha → Quetzalcoatl: 17/17 |
| Bochica / Viracocha | 35/37 | 94.6% | Bochica → Viracocha: 19/19 Viracocha → Bochica: 16/18 |
| Osiris / Quetzalcoatl | 31/36 | 86.1% | Osiris → Quetzalcoatl: 16/16 Quetzalcoatl → Osiris: 15/20 |
| Quetzalcoatl / Votan | 29/34 | 85.3% | Quetzalcoatl → Votan: 14/17 Votan → Quetzalcoatl: 15/17 |
| Nommo / Oannes | 31/37 | 83.8% | Nommo → Oannes: 14/17 Oannes → Nommo: 17/20 |
| Quetzalcoatl / Thoth | 28/37 | 75.7% | Quetzalcoatl → Thoth: 11/18 Thoth → Quetzalcoatl: 17/19 |
| Bochica / Quetzalcoatl | 27/39 | 69.2% | Bochica → Quetzalcoatl: 17/20 Quetzalcoatl → Bochica: 10/19 |
| Oannes / Viracocha | 24/35 | 68.6% | Oannes → Viracocha: 16/20 Viracocha → Oannes: 8/15 |
| Oannes / Quetzalcoatl | 5/39 | 12.8% | Oannes → Quetzalcoatl: 5/20 Quetzalcoatl → Oannes: 0/19 |
| Osiris / Viracocha | 1/31 | 3.2% | Osiris → Viracocha: 1/17 Viracocha → Osiris: 0/14 |
| Oannes / Osiris | 1/35 | 2.9% | Oannes → Osiris: 1/19 Osiris → Oannes: 0/16 |
Recalculated from the saved V1 by-pair results. Failed or missing responses are not scored as last place.
The retest worth doing.
Start with the existing source passages: compare complete episodes from teachers and founders, cite each proposed match on both sides, and score specific relationships separately from generic “teaching” or “travel”. Preserve uncertain readings.
Test a small reading rubric first, including changed actors and reversed arrival/departure checks. Then freeze a larger source collection, use close role-matched alternatives, and report every pair and its strongest competitors.
A high score would establish a resemblance within that collection. A shared origin would still require a historical argument.