Why it matters
Proof-of-stake replaces the cost of electricity with the cost of capital. A network is secured because attacking it requires controlling a large quantity of the asset, and misbehaving destroys part of it. The security budget is therefore an interest rate: the return a network must pay to keep enough capital committed, and the penalty it must impose to make dishonesty unprofitable.
For a holder of a tokenized fund or a payment claim on such a network, that rate and that penalty price settlement assurance. If the return is too low, capital leaves; if too high, everything stakes and unencumbered collateral becomes scarce. If the penalty is too mild, correlated failure is cheap; if too severe, only large operators can bear the tail risk and the validator set consolidates.
How it works
On Ethereum a validator earns for attesting to blocks, occasionally for proposing one and occasionally for sync committee duty. The reward scales with a base reward that falls as total stake rises: effective balance divided by the square root of total active balance, with a base reward factor of 64. The published duty weights sum to 64: 14 each for timely source and head, 26 for timely target, 8 for proposing, 2 for sync committee.
Penalties come in three grades. Missing a duty forfeits the reward and incurs a small charge, the ordinary cost of downtime. Slashing applies to provable equivocation (double voting, or proposing two blocks for one slot) and burns part of the stake immediately: 0.0078125 ether for a 32 ether validator, scaled linearly with balance. Then, at day 18 of the 36-day forced exit, a correlation penalty applies "whose magnitude scales with the total staked ether of all slashed validators". If the chain fails to finalise for more than four epochs, an inactivity leak begins and the stake of non-participating validators "gradually bleed away" until finality returns.
Solana does not have protocol-enforced slashing. Anza's post of 17 January 2025 introducing SIMD-0212 notes the protocol has been secure without it and proposes a schedule weighting vote violations at 1 and duplicate block violations at 10 under a quadratic function whose slow climb avoids "harsh penalties for the sake of giving the network the freedom to make occasional mistakes". It was in phased rollout, subject to a governance vote.
Economic mechanism
Two mechanisms operate, and conflating them is the common analytical error.
The first is the reward curve, which sets the equilibrium quantity of stake. Because the base reward falls with the square root of total stake, yield declines as capital enters and rises as it leaves: the curve is a supply schedule for security. A validator stakes while the return exceeds its cost of capital, operating cost and the option value of holding the asset unencumbered. Lido reported 2.2% on stETH on 25 July 2026, against 9,337,523 ether staked and $17.6bn of value. Ordering revenue and priority fees sit on top of issuance and vary with market activity, not protocol parameters.
The second is the penalty schedule, and it does not price honesty. It prices independence. An isolated double-signature (a misconfigured failover, a duplicate key on two machines) costs a fraction of a per cent of stake, deliberately, because such events are accidents. The correlation penalty escalates with the total stake slashed in the same window because simultaneous equivocation is the signature of an attack or a shared dependency. The schedule charges an operator for the probability that its failure coincides with everyone else's, which is the externality a single client, cloud region or middleware layer creates.
Who pays and who is paid follows. The network pays stakers with newly issued supply, diluting non-stakers; that transfer funds security. Stakers pay operators a commission retained from rewards. Operators bear the tail risk of slashing and, where they are institutional custodians, price or indemnify it. The party bearing correlated risk without pricing it is the liquid staking token holder, whose exposure is to the largest operator's dependency stack rather than their own.
Participants
Solo stakers run their own hardware and keys; node operators and staking-as-a-service firms run validators for others for a commission. Liquid staking protocols issue a receipt token against pooled stake and spread it across operators; Lido lists more than 900 across its curated, distributed and community modules. Custodians and prime brokers (Anchorage Digital, BitGo, Coinbase) provide institutional staking with key management and, sometimes, slashing cover. Protocol developers set the parameters: the Ethereum Foundation funds the research behind issuance and penalty design, while Anza and the Solana Foundation determine client behaviour and stake distribution.
Examples
The mass slashing of 10 September 2025 is the best-documented correlated event. Thirty-nine Ethereum validators tied to the SSV Network were slashed, not through any protocol compromise but because routine maintenance on Ankr's systems triggered slashings in one cluster while a secondary validator setup caused duplicate signing in another that had migrated from Allnodes two months earlier. Losses were small: one validator with a 2,020 ether stake lost about 0.3 ether, roughly $1,300. Fewer than 500 validators out of more than 1.2 million active had been slashed since the Beacon Chain launched in 2020. The cause is the instructive part: two offchain dependencies, invisible in the validator set's apparent diversity.
The SEC statements of 2025 changed the economics for US institutions. The Division of Corporation Finance addressed protocol staking in May 2025 and, on 5 August 2025, said that "depending on the facts and circumstances, the liquid staking activities covered in the statement do not involve the offer and sale of securities" under the 1933 or 1934 Acts. Pintail attributes the subsequent entry queue, nearly a year without clearing and the longest in Ethereum's history, to that clarity.
Solana is the counterfactual: a network carrying a large share of tokenized fund and stablecoin distribution has operated without protocol-enforced slashing.
Risks and limitations
The open dispute on Ethereum is whether the issuance curve should be capped. Pintail's April 2026 analysis reported the staking ratio passing 33% including the entry queue for the first time, and argues that the current curve, designed before liquid staking existed, tends toward a very high share; earlier research suggests 80% to 100% absent intervention. The stated harms are specific: solo stakers paying income tax on nominal yield face negative dilution-adjusted returns before receipt-token holders do and, once pushed out, do not return; and receipt tokens displace unencumbered ether as collateral. Staking operators and receipt-token holders answer that a cap cuts the security budget and expropriates existing stakers to solve a problem that has not yet materialised.
The sharpest formulation of concentration risk is "too big to slash": if one operator's stake is large enough that penalising it would destabilise the network or the asset's value, the penalty schedule stops being credible and economic security is nominal. There is no agreed threshold at which this binds, and no mechanism to enforce diversity in the offchain dependencies (clients, cloud regions, middleware, key management) that determine whether failures correlate.
Two further limitations. Published yields are gross of commission, tax and dilution, and the three together can turn a positive nominal return into a negative real one. And the absence of slashing is no evidence that slashing is unnecessary, any more than its presence is evidence that it works: Ethereum's schedule has never been tested by a deliberate attack, and neither has Solana's absence of one.
Key metrics
The figures that matter are the staking ratio and the entry and exit queues, which show whether the reward curve is clearing; nominal yield decomposed into issuance, priority fees and ordering revenue, since only the first is a protocol parameter; that yield net of commission and dilution; the largest operator's and largest client's share of stake; and the count of validators slashed, with the largest correlated event. That last is the only observable test of the dependency structure.
Related research
The ethereum.org rewards and penalties documentation is the authoritative statement of the reward weights, base reward formula, initial slashing charge and correlation penalty. Anza's SIMD-0212 post is the clearest reasoning on why a network might design a mild penalty curve deliberately. Pintail's April 2026 analysis is the most developed argument that a high staking ratio is itself a security problem, and the SEC's staff statements of May and August 2025 set the perimeter for US institutions.