Decentralization is easy to claim and hard to measure. EmpoorioChain's approach is to write the measurement down first, publish where the network falls short, and make the gate the thing that decides when mainnet is even discussable. This post is that gate, as it stands in September 2026.
The current state, plainly
The public testnet has two validators. Both are operated by Empoorio. Both sit in the same provider (IONOS, AS8560) and the same data centre in Logroño. GRANDPA requires two thirds of the validator set to finalize; two thirds of two is two. Losing either validator halts finality. Fault tolerance is zero, and a provider-level incident takes both down at once.
The decentralization report says this in its first paragraph and labels the network not suitable for mainnet. That is the correct label.
The gate
| Requirement | Minimum | Current | Status |
|---|---|---|---|
| Active validators | 10 | 2 (read from Session::Validators) | Not met |
| Independent external operators | 4 | 0 demonstrated | Not met |
| Maximum concentration per operator | No majority control | Empoorio controls 100 % | Not met |
| Uptime per validator, 7-day window | ≥ 99 % | Pending Prometheus | Pending |
| Homogeneous, audited client version | 100 % reported | Pending | Pending |
The gate is considered passed only when a reproducible audit script produces JSON evidence with no critical RPC errors, EmpooScan or Grafana shows the active validators publicly, and every row above is green.
What counts as an independent operator
A validator is counted as external in the report only if its operator:
- publishes the identity of the responsible entity;
- exposes a private metrics endpoint to the authorised monitoring stack;
- keeps uptime at or above 99 % over a 7-day window;
- runs an approved client version;
- shares neither infrastructure nor keys with another operator counted as independent;
- accepts the incident-response playbook.
The last two rules are the ones that separate decentralization from its appearance: ten validators run by three friends on one cloud account are one operator.
Why publish a gate you fail
Because the alternative is to publish numbers you invent. The previous version of this same report — from May 2026 — listed three validators in three AWS regions. They did not exist. The validator inventory file contained template data using Substrate's well-known development accounts, and the report presented it as real infrastructure. The audit script that was supposed to verify it did not read the chain at all: pointed at a refused connection, it returned the same three fictional validators and exited with success.
That was found and corrected on 5 September 2026. The script now reads Session::Validators and Aura::Authorities from a live RPC, compares them to the inventory, and fails loudly if they disagree or if no RPC answers. The report you are reading is the first one built on that footing.
The path from two to ten
Recruiting external operators is the work. The operator guide (validator in 30 minutes), the hardware baseline, the key-management rules and the incident playbook exist so that an external team can join without a call to the core team. There is also a ceiling to plan around: Aura and GRANDPA are configured with MaxAuthorities = 32, while the staking pallet declares MaxValidators = 100. Beyond 32, the authority set is silently truncated with a warning — a validator could be elected, have its bond locked and never author a block. Raising or reconciling that limit is on the list before the set approaches it.
Based on DECENTRALIZATION_REPORT.md (2026-09-05) and VALIDATOR_IN_30_MINUTES.md.