X9 v2: External Prebiotic Network Replication Confirmed 4/4

Why X9 v2 mattered

X9 v1 was useful, but it was still a hand-built model graph.

That meant the next real test was obvious: can the same structural story survive contact with a published external autocatalytic network?

X9 v2 answers that with a clean yes.

Data source

This run used the published CatReNet example-9 reaction system, which is explicitly tied to:

Xavier, Hordijk, Kauffman, Steel, and Martin (2020), Autocatalytic chemical networks at the origin of metabolism, Proc. Roy. Soc. B.

The source was snapshotted locally, converted into an IRDME molecule-level multilayer graph, and then pre-registered before analysis.

Layers:
• reactant_coupling
• catalyst_product
• product_dependency

Pre-registered hash: 63759fcb... Date: 2026-05-27

Result

4/4 CONFIRMED
• h1 CONFIRMED: catalyst_product is structurally closer to product_dependency than reactant_coupling is
• r(d2,d3)=0.8263
• r(d1,d3)=0.4959
• h2 CONFIRMED: molecule a is the #1 hub in product_dependency
• h3 CONFIRMED: molecule a is also top-tier in catalyst_product, ranking #2
• actual #1 in catalyst_product: b
• h4 CONFIRMED: 9 molecules show strong cross-layer divergence with rank_gap>=3

What changed scientifically

This is the important upgrade:
• X9 v1 showed the signal in a model-based prebiotic graph
• X9 v2 shows the signal in a published external autocatalytic network

That does not prove a universal law of prebiotic organization by itself. But it does move X9 out of the "interesting hand-built example" category.

The structural pattern survived externalization.

The strongest takeaway

The confirmed h1 result matters most. Simple co-reactant pairing is not the main organizing layer. The stronger alignment is between:
• catalytic support
• product dependency

That is a cleaner structural signature of autocatalytic organization than mere substrate co-presence.

Next step

The next serious X9 step is larger external replication on the published prokaryotic catalytic network, not another hand-tuned small graph.

That is where the program either scales or breaks.

---

IRDME project: arXiv:2604.23639 All predictions are pre-registered publicly before analysis.