Validator concentration: count control, not validator keys
Validator concentration is a question about who can coordinate consensus weight. A network may have many validator keys while a smaller set of operators, custodians, or governing arrangements influences their behavior. Count keys first, then build a separate control view. Neither a token-holder ranking nor the number of staking depositors answers this question.
Ethereum’s attack-and-defense documentation explains why the fraction of stake controlled matters to particular attacks. Those protocol-specific conditions should not be exported to a different proof-of-stake system. This worksheet measures disclosed dependencies; it does not estimate an attack probability or label a network safe.
Choose the control question
Specify the network, observation epoch, active validator set, and weighting basis. Distinguish deposited balances from the weight actually used by consensus. State whether an operator can choose software, sign messages, select block construction services, or redirect withdrawals. These powers can belong to different parties.
A liquid staking pool is not necessarily one operator, and several brands do not necessarily represent independent operations. Keep a legal or governance grouping alongside an operational grouping. Record the reason for every aggregation. A shared withdrawal destination is a lead, not proof that signing decisions share a controller.
The principal output is a table with validator identifier, effective weight, operating entity, evidence for that entity, delegation relationship, and unresolved ownership. Retain the original key-level table so another reviewer can undo an uncertain grouping without reconstructing the whole dataset.
Hypothetical control regrouping
Assume a fictional network has 100 equally weighted validators. Brand Cedar advertises 30 keys, but deployment records identify three independent operators running ten each. Two other brands operate fifteen keys apiece through the same service company. The raw brand ranking and the operating-service ranking now answer different questions.
Under a disclosed-operator view, the service company controls 30 keys. Cedar contributes three groups of ten if the independence evidence supports that treatment. If contracts allow Cedar’s governing body to replace operators, record that longer-term authority in another column rather than erasing the operating distinction.
Suppose ten further validators have no reliable operator evidence. Do not divide them into ten independent entities by default. Report known group shares and a range under explicitly different unknown-group assumptions. The example is hypothetical and does not describe any real staking pool.
A concentration worksheet
| Field | Rule for this review |
|---|---|
| Weight denominator | Total active effective weight at the same epoch |
| Operator share | Group weight divided by that denominator |
| Shared service | Hosting, signing, client or block construction dependency |
| Grouping evidence | Deployment disclosure, contract rights, dated attribution |
| Unknown bucket | Unresolved weight, with sensitivity treatment |
| Change trigger | Operator rotation, delegation change or new control evidence |
Calculate a top-group share only after the grouping rule is fixed. If using a concentration index, write the formula, such as the sum of squared group shares, and state that it summarizes the selected grouping. It cannot reveal an undisclosed coordination agreement or replace protocol-specific threat analysis.
Interpret the bounds
Run at least two views: independently disclosed operators and broader organizational dependencies. A large difference between them is useful evidence about uncertainty. It is not permission to select whichever ranking supports a preferred narrative. Explain which rights create the difference and what would resolve it.
Check how the map changes when weights move while keys remain constant. A stable key count can conceal a changing distribution of voting influence. Likewise, an operator leaving a pool may alter the control map without changing aggregate token supply. Refresh these observations together before comparing periods.
Common errors include equating depositors with validators, treating every address as an organization, and summing weights from different epochs. Describe a dated exposure to coordinated behavior, not a claim about intent. A spot-asset memo can then discuss this dependency without turning concentration into a price target.
FAQ
Is a pool’s market share its operator concentration?
Only under an explicitly justified control definition. Depositor governance, operational signing and withdrawal authority may be separated.
Can unknown validators be called decentralized?
No. Unknown means attribution is unresolved. Show how plausible grouping assumptions affect the concentration result.