The case for a maximum of eight nominations on Polkadot
Polkadot should reduce the maximum number of validators a nominator can approve from sixteen to eight. Eight offers room for a resilient selection of active validators and additional choices, while narrowing the breadth of continuing support for accounts subject to that limit. The proposed transition would preserve existing longer lists, with their allowances decreasing as targets are removed. The objective is a more natural flow of support between validators, with informed nominator choices giving credible candidates a realistic route into the active set.
The case rests on two findings. Recently submitted full nomination lists overwhelmingly cover established winners, and the additional coverage available from very long lists has diminishing value in a mathematical continuity benchmark. Persistent nomination preferences also provide substantial backing to a subset of incumbents. Together, these findings justify reconsidering whether sixteen is an appropriate default allowance.
This paper recommends a policy direction and explains why eight is a reasonable choice. Decisions about handling existing lists belong to a separate implementation proposal. The observations below use finalized Polkadot Asset Hub state from 12 September 2026, with 731 registered validator candidates, 600 active validators and 28,713 nomination records.
Takeaways
- Sixteen is capacity rather than a participation requirement. Nominators need their stake assigned to elected targets; they do not need sixteen active validators to receive an assignment.
- Recent full lists mostly reinforce current winners. Of 7,344 approvals in 459 recently submitted sixteen-target lists, 96.13% point to current winners and 1.99% to registered waiting candidates.
- Eight leaves room for continuity and choice. A core of five active validators plus three discretionary selections fits within eight. The continuity benchmark already exceeds 99.95% with five active targets, under its stated assumptions.
- Longstanding nominations provide substantial backing. 114 active validators receive most of their assigned nominator backing from preferences submitted at least 365 eras before the reference era.
- The proposed benefit is more deliberate support. A smaller allowance asks nominators to be more selective about which validators receive their continuing approval.
What multiple nominations are meant to achieve
In Nominated Proof of Stake, a nomination identifies a validator that an account is willing to support. The election selects validators and assigns stake along approvals to those winners. Approving sixteen validators does not create sixteen copies of the nominator's stake, and the protocol does not reserve an equal sixteenth for each target. Additional approvals give the election more options for allocating the same stake budget (Parity Technologies, n.d.; Polkadot Wiki, n.d.).
That flexibility can serve two useful purposes: maintain participation when some chosen validators are not elected, and support credible waiting candidates while retaining elected targets in the same list. A 2023 technical discussion of nomination quotas explicitly describes this ability to help new validators establish support. It also explains that the historical limit of sixteen arose from scalability constraints on the nomination graph and its stored snapshot (Pestana, 2023).
The policy question is therefore how much approval capacity provides useful continuity and diversity in practice. Sixteen has no special status as the number required for reliable nominator participation. Its value depends on how nominators use the available choices.
How full nomination lists are being used
The clearest evidence comes from the 459 current sixteen-target lists submitted during eras 2261 to 2290, the most recent 30 eras through the reference active era. These lists contain 7,344 stored approvals. Classifying their targets at the reference snapshot gives the following distribution.
| Target classification | Approval entries | Share |
|---|---|---|
| Current elected validators | 7,060 | 96.13% |
| Registered waiting candidates | 146 | 1.99% |
| Currently unregistered accounts | 138 | 1.88% |
456 of the 459 lists contain more than eight current winners. Moreover, 390 contain at least eight validators that were elected in every one of the 84 observed eras, from 2207 through 2290. These portfolios already contain extensive coverage of validators with persistent elected membership.
This suggests that much of the available nomination capacity is being used for repeated incumbent coverage, with comparatively little directed toward waiting candidates. In these recent full lists, the extra capacity appears underused for the decentralization objective of widening participation and helping credible candidates enter the set.
A nominator may reasonably value a dependable incumbent. The concern is the aggregate effect when many portfolios use nearly all their capacity to approve established winners. Supporting a broad collection of incumbents can become the easy default, even when a smaller collection would provide ample continuity. The resulting approval pattern gives us a reason to question the size of the allowance.
Longstanding preferences can sustain incumbent support
At the reference snapshot, 20,212 nomination records, or 70.39% of the population, have a recorded submission era at least 365 eras earlier. This is measured directly from the submittedIn field in chain state. The reference active era is 2290, so the count includes records with submittedIn at or below 1925. It measures elapsed eras rather than calendar timestamps (Parity Technologies, n.d.).
The relationship to current validator backing is substantial for part of the set:
- 100 active validators have at least 200 approving accounts in this older-preference cohort. One has 3,011.
- 114 validators receive more than half of their assigned nominator backing from the cohort; 44 receive at least 90%.
- The cohort supplies 175.29 million DOT, or 19.75% of all currently assigned nominator stake.
These stake figures count actual election assignments, rather than repeatedly counting a nominator's full balance for every approval. They show that old preferences carry economic weight, beyond simply occupying storage.
The concentration is uneven: 365 of the 600 current validators have no approvals from this older cohort. The concern is a subset of validators accumulating substantial continuing support, rather than every incumbent sharing the same advantage. That support can persist without a new on-chain expression of confidence, even as the operator's circumstances or a nominator's priorities change.
Nomination lists also contain 83,832 entries pointing to currently unregistered accounts, representing 34.40% of all stored entries. Validator deregistration does not automatically remove these entries from nominators’ stored lists. The figure therefore measures currently ineligible targets retained in storage; it does not, by itself, establish whether nominators have neglected their selections.
Reducing the cap would narrow the number of continuing approvals for new nominators and for existing nominators as their lists shrink. The intended result is more selective support and a validator set that responds more readily to informed choices. Keeping those choices current remains an ongoing responsibility; a smaller cap by itself does not renew them.
Why eight is a reasonable choice
Eight allows a nominator to build a core of active validators and retain space for additional selections. For example, five active validators from distinct operators plus three discretionary choices can combine continuity with support for credible waiting candidates or broader operator representation. This is an illustration of what fits within eight, rather than a required portfolio composition.
Earlier research proposed five active nominations as a managed continuity core, with the remaining capacity available for waiting candidates. That work addressed how to use sixteen slots. The proposal here takes the further policy step of asking whether a smaller total allowance would better reflect the way nomination capacity is actually being used (Bulat, 2026).
The continuity calculation can be updated to the current candidate population. If successive committees each contain 600 validators drawn from the same 731 candidates, at most 131 current winners can be absent from the next committee. For a uniformly sampled group of distinct current winners, the following benchmark measures the chance that at least one remains, even at that maximum replacement count.
| Current active validators selected | Modeled coverage in the next committee |
|---|---|
| Five | 99.953300% |
| Six | 99.990111% |
| Eight | 99.999565% |
| Sixteen | 99.9999999987% |
The modeled improvement from eight to sixteen is approximately 0.000435 percentage points. This illustrates diminishing additional insurance against losing every elected target. It supports the argument that an account need not approve a long list of incumbents solely to maintain coverage.
Eight is consequently a reasonable policy compromise: it leaves three choices beyond the five-active core, or two beyond a more conservative six-active core. An eight-target allowance is half the current maximum while retaining meaningful choice. Under the proposed transition, existing longer lists would retain their current allowance until they shrink. The evidence supports this balance; it does not identify eight as a uniquely optimal number.
The benchmark concerns currently active selections, rather than any arbitrary list of eight candidates. A five-active plus three-waiting portfolio uses the five-active benchmark for its continuity core. Validator selection and operator diversity remain important, because real portfolios do not follow a uniform random sample.
The benefits to pursue
The central benefit is a more selective expression of nominator support. With fewer choices available, each approved validator occupies a larger share of the account's limited selection capacity. That can encourage nominators to consider which operators they actively want to support, including their reliability, independence, infrastructure and contribution to Polkadot.
Eight also gives nominators a smaller set of relationships to review. Comparing operators and keeping track of changes takes attention. For accounts subject to the eight-target limit, the proposal reduces the maximum number of targets to assess, while leaving room to support several independent operators. Better review is an expected benefit to evaluate, rather than a behavior already demonstrated by this research.
For validators, the desired outcome is support that can move as confidence changes. Reliable operators should continue to earn backing, and credible challengers should have a route into the set. A smaller allowance can limit broad incumbent coverage, but the distribution of future support will still depend on whom nominators choose.
There is also a structural efficiency argument. An eight-target limit permits half as many approval entries per nomination record as a sixteen-target limit. The proposed transition would not immediately halve existing records or their storage requirements: reductions would emerge as longer lists shrink, while new lists would be limited to eight. Fewer entries can reduce nomination-graph storage and processing demands. Actual savings depend on list lengths and how election snapshots are bounded; no percentage reduction in total runtime cost is claimed here.
The limit remains per account. A nominator could split stake between accounts to approve a larger total set of validators. This involves managing separate stake budgets and nomination lists. The proposal does not assume that such behavior is impossible or predict how frequently it would occur.
Research scope and tradeoffs
The observed list composition and preference ages establish current behavior and persistence. They do not establish nominators' motives, or whether an old list was reviewed and deliberately kept. The recent full-list cohort is a specific part of the population. These observations support the policy argument without demonstrating that the sixteen-target limit alone caused the current pattern.
The coverage benchmark assumes a fixed candidate universe, a fixed next committee for comparison and uniform selection from current winners. It measures the presence of an elected target, rather than the result of rerunning an election after changing everyone's choices. Candidate registration, correlated operator failures, voter inclusion and the election's stake assignments can all affect actual participation.
There is a real tradeoff in reducing approval capacity: fewer edges also give the election less flexibility to distribute stake and give nominators fewer opportunities to support waiting candidates. Eight does not automatically increase operator diversity or the Nakamoto coefficient. Those outcomes require election-wide assessment.
This post advocates an eight-target ceiling as a direction for Polkadot, with reliable participation, feasible elections and diverse validator participation as requirements for its eventual implementation. The possible next steps below outline a transition for community discussion, without selecting which existing nominations should be removed to meet the lower ceiling.
Recommendation
Polkadot should pursue an eight-nomination maximum. The current evidence shows that recent full lists mostly cover incumbents, while a much smaller active core can provide high modeled continuity. Eight preserves room for that core and additional choices. The proposed transition would apply this limit to new and shorter lists, with the broader benefits depending on existing longer lists shrinking over time. It would not guarantee that every account eventually reaches eight.
The goal is a more natural flow of support: validators retain backing through continuing confidence, and nominators have a manageable set of choices through which to express it. The community can agree on that policy direction before deciding how existing records should be handled.
Possible next steps
Two complementary changes could support this direction: permissionless cleanup of nominations to persistently unregistered accounts, and a lower limit on future nomination calls that accommodates existing longer lists. I recommend considering both as part of a separate implementation proposal.
Permissionless cleanup of unregistered targets
A permissionless call could allow anyone to request and later complete the removal of nomination entries pointing to an unregistered validator. I propose a preliminary waiting period of one month, expressed as a configured number of Asset Hub blocks and measured from the block at which a cleanup request is recorded. The precise block count could be agreed upon by the community; its calendar duration would depend on block production.
Historical events and state can reveal when registration ended, but the current validator registry does not retain a deregistration block that this call can use as a timer. A two-stage process would establish that record explicitly:
- Record a cleanup request at Asset Hub block X. The call checks that the target account is absent from
Staking.Validatorsand stores the request block on-chain. It fails if the account is still registered as a validator. This records an observed absence; it does not deregister the validator or establish the original date of deregistration. Further requests during the waiting period should preserve the original block X. - Complete cleanup after block X plus the waiting period. Once the waiting period has elapsed, anyone can submit the call again to remove the specified nomination entries, if they still exist. The target must remain unregistered. Re-registration must invalidate the pending request; a subsequent deregistration would require a fresh request and waiting period.
The waiting period gives the validator time to register again before its retained nominations become eligible for removal. Measuring the waiting period in blocks makes it enforceable on-chain.
Limit future nominations while accommodating existing lists
For each new nominate call, the permitted target count would be max(8, existing_count), where existing_count is the number of targets stored for that nominator immediately before the call. An account without an existing nomination list would have a limit of eight.
- Lists with eight or fewer targets could contain at most eight after the call. A nominator could add or replace targets within that limit.
- Existing lists above eight would use their current count as the upper bound. A list of sixteen could remain at sixteen or shrink, but could not grow. If it shrank to twelve, twelve would become the upper bound; if it later shrank to eight or fewer, the upper bound would be eight. Removals through cleanup would also reduce the stored count used by subsequent calls.
This lets the allowance decrease as longer lists shrink, while allowing nominators to replace targets within their current limit. Existing lists would not be forcibly shortened merely to meet the new ceiling, and unchanged lists above eight could remain in place.
The nomination-limit transition therefore adopts no pruning strategy to select or remove existing targets solely to enforce eight. The cleanup process above addresses the separate case of persistently unregistered targets. Any broader pruning strategy is a sensitive decision that must be discussed and agreed upon by the community, if one is ever adopted.
Methods
The analysis uses original chain observations obtained through the Polkadot Asset Hub RPC at finalized Asset Hub block 20,557,035, recorded on 12 September 2026 at 08:22:24 UTC, with active era 2290. Its block hash is 0xa28410670d30cc4ff85c8b9de93d813a69f59a94b28685d05bf798336b66987d. Using a single finalized block keeps the nomination records, candidate registry and active-era state aligned. The sixteen-target configuration is recorded in the Asset Hub staking runtime (Polkadot Fellows, n.d.).
Each stored nomination target was compared with the candidate registry and active-era validator set, then classified as a current elected validator, a registered waiting candidate or a currently unregistered account. Recent full lists were selected by requiring sixteen stored entries and submittedIn between 2261 and 2290 inclusive. Their target counts were summed within each category and divided by the cohort's 7,344 entries to calculate the reported shares. The historical comparison checks their present targets against the elected sets for all 84 eras from 2207 through 2290; it does not reconstruct earlier nomination lists.
The older-preference cohort includes nomination records with submittedIn at or below 1925, at least 365 eras before the reference active era. Its population share is 20,212 divided by 28,713. For each current validator, the analysis counts distinct accounts in this cohort whose stored lists approve it. Nomination records represent accounts, including applicable pool staking accounts, rather than individual people. Approval counts describe entries or distinct approving accounts as stated.
Assigned backing was measured separately from approvals. The analysis sums the stake actually assigned to each validator by nominators in the older-preference cohort, using active-era exposure pages, and compares it with that validator's total assigned nominator backing. Exposure-page sums were reconciled to the validator totals. Across the active set, total assigned nominator backing is 887,492,161.6440 DOT, of which the cohort contributes 175,291,524.1266 DOT. This avoids counting a nominator's entire stake repeatedly for every target it approves.
The continuity benchmark is one minus the number of k-target groups drawn entirely from 131 displaced validators divided by the number drawn from all 600 current winners. Equivalently, the benchmark is 1 - C(131, k) / C(600, k), where C(n, k) counts k-element subsets of an n-element set. Evaluating this expression for five, six, eight and sixteen targets gives the coverage table. This is a coverage benchmark against a fixed comparison committee, rather than an election forecast.
The nomination-quota discussion is cited for its design rationale and graph-size tradeoff, rather than treating historical runtime parameters as current values (Pestana, 2023).
References
Bulat, R. (2026) ‘How many active validators does a Polkadot nominator actually need?’, Ross Bulat, 17 August. Available at: https://rossbulat.com/polkadot/how-many-active-validators-does-a-polkadot-nominator-actually-need/ (Accessed: 14 September 2026).
Parity Technologies (n.d.) Polkadot SDK [Source code]. Commit b07bc46ed883db53dccf87bca8313608dcde170c. Available at: NPoS election primitives, nomination submission era and assignment of the field on nomination.
Pestana, G. (2023) ‘Discussion: define and enable dynamic nomination quota curve based on the nominator’s balance’, Polkadot Forum, 2 October. Available at: https://forum.polkadot.network/t/discussion-define-and-enable-dynamic-nomination-quota-curve-based-on-the-nominators-balance/4137 (Accessed: 14 September 2026).
Polkadot Fellows (n.d.) Asset Hub Polkadot staking runtime [Source code]. Commit ed8040b1be85e0f3799a48b796c85d0cacca3d34. Available at: https://github.com/polkadot-fellows/runtimes/blob/ed8040b1be85e0f3799a48b796c85d0cacca3d34/system-parachains/asset-hubs/asset-hub-polkadot/src/staking/mod.rs (Accessed: 14 September 2026).
Polkadot Wiki (n.d.) NPoS election algorithms. Available at: https://wiki.polkadot.com/learn/learn-phragmen/ (Accessed: 14 September 2026).