Technical Summary: Proof of Useful Attestation (PoUA)

1. Problem Statement

The paper identifies a fundamental misalignment in consensus mechanisms for "attestation-native chains"—blockchains whose primary economic activity is the production and verification of cryptographic attestations (e.g., content provenance, AI-output attribution, supply-chain traceability).

Current Proof of Stake (PoS) systems are designed for general-purpose state transitions. In these systems, validators earn fees regardless of whether they process the specific application workload correctly or selectively censor it. Consequently, an adversary with sufficient capital can acquire stake, censor specific attestation schemas, or extract MEV from attestation reordering without triggering standard slashing conditions. The chain's economic security remains indifferent to the correctness of its dominant workload.

The paper argues that a chain hosting attestation contracts on a general-purpose L1 cannot defend the workload effectively because the consensus incentives are decoupled from the attestation work itself.

2. Methodology: Proof of Useful Attestation (PoUA)

PoUA is a consensus weighting primitive designed to tie a validator's influence directly to their history of processing valid attestation work. It is not a replacement for the underlying BFT primitive (e.g., Tendermint-style optimistic finality) but a layer that modifies how validator weight is calculated.

Core Mechanism

Validator weight ($w_v$) is computed as the product of bonded stake ($s_v$) and a non-transferable reputation score ($r_v$):
$$w_v(t) = s_v(t) \cdot r_v(t)$$

• Reputation Accumulation: Reputation grows based on the validator's contribution to processing valid attestations. The update function is additive and bounded:
  $$r_v(t+E) = \text{clip}_{[r_{min}, r_{max}]}(r_v(t) + \eta \cdot g_v(t) - \lambda \cdot b_v(t))$$
  Where $g_v$ is a fee-weighted measure of valid attestation processing (split between proposer and voter roles) and $b_v$ is a bad-behavior score.
• Non-Transferability: Reputation is intrinsic to the validator's identity on the chain and cannot be transferred or unbundled.
• Bounded Interval: $r_{min} > 0$ ensures new validators have non-zero weight (preventing cold-start lockout), while $r_{max} < \infty$ prevents runaway concentration of power.

Defense Against Compound Adversaries

The paper addresses the "compound adversary" (an attacker with both capital and the ability to control attestor sets and submitter addresses to farm reputation). PoUA employs a six-layer defense:

1. Layer 1 (Formal): Excludes attestations where the submitter address equals the proposer's address.
2. Layer 2 (Formal): Excludes attestations where the submitter address is within a transaction-graph distance $d$ of the proposer (preventing simple fund routing).
3. Layer 3 (Economic): A fraction $\tau_{burn}$ of every attestation fee is sent to a non-recoverable destination (e.g., a burn address). This creates a formal lower bound on the cost to grind reputation.
4. Layer 4 (Heuristic): Statistical detection of grinding patterns (schema concentration, graph density).
5. Layer 5 (Governance): An appeal process for false positives.
6. Layer 6 (Future): Zero-knowledge proofs of stake-graph independence.

3. Key Contributions

Formal Security and Cost-to-Attack Premium

The paper derives a formal cost-to-attack premium ($\kappa$) against a capital adversary. In a steady-state healthy chain, the cost to acquire a fraction $\rho$ of consensus weight is multiplied by the ratio of the average honest reputation to the minimum reputation:
$$\kappa = \frac{\bar{r}_H}{r_{min}}$$
With recommended parameters ($r_{max}/r_{min} \in [4, 10]$), PoUA offers a 4× to 10× multiplicative premium on the cost-to-attack compared to equivalent pure-stake PoS chains. This "moat" is formalized in Section 5.3.

Cost-to-Grind Bound (Lemma 1)

A central contribution is Lemma 1, which establishes a formal lower bound on the non-recoverable fees a compound adversary must pay to grind reputation. Even if an attacker evades Layers 1 and 2, the Layer 3 burn ensures:
$$F_{net, \text{per member}} \geq \frac{\tau_{burn} \cdot \Delta r}{\eta \cdot \alpha_{eff}(m, k)}$$
This converts the defense against reputation farming from a purely heuristic argument into an economic one.

Incentive Alignment

The paper models validators as profit-maximizers. It argues that because reputation is non-transferable and represents future revenue (via block rewards and fees), slashing a validator destroys not just their current bond but their future revenue stream. This "time-locked" incentive alignment is absent in pure-stake PoS.

Implementation and Integration

The paper provides a concrete integration specification for the Sovereign SDK rollup framework, detailing:

• Validator selection and vote weighting modifications.
• Reputation state storage (32 bytes per validator).
• Adaptive parameter rebasing mechanisms for $\tau_{burn}$, $\eta$, and $\lambda$ to handle fee market drift without governance intervention.

4. Results and Validation

The paper presents results derived from a reference simulator (prototypes/poua-sim/):

• Cost-to-Attack: Monte Carlo simulations confirm the analytical $\kappa$ curves, showing the 4×–10× premium holds across different validator set sizes (50 to 1000 validators).
• Layered Defense: Strategy-search heatmaps demonstrate that while individual defense layers (e.g., Layer 1 only) can be bypassed, the combination of all layers collapses the reward for grinding strategies to the baseline ($r_{min}$).
• Transition States: The paper quantifies the "warmup" and "recovery" periods where the premium is lower than the steady-state ceiling, providing operational guidance for chain launch and post-slash recovery.

5. Significance and Claims

The paper positions PoUA not as a new cryptographic primitive, but as a synthesis of existing concepts (reputation-weighted consensus, proof-of-useful-work, and restaking) tailored specifically for attestation-native chains.

• Modest Claims: The authors explicitly state that PoUA does not provide a complete cryptographic security proof for all rational deviations. The defense against compound adversaries relies on a formal economic floor (Lemma 1) supported by heuristic detection (Layer 4) and governance (Layer 5).
• Scope: The mechanism is designed for chains where the workload is attestation production. The paper explicitly excludes general-purpose smart contracts, arguing that adding them would dilute the specific security moat PoUA provides.
• Differentiation: Unlike restaking (EigenLayer), which reuses stake from a primary chain, PoUA constructs a native consensus mechanism where the security budget is sized to and aligned with the attestation workload. Unlike Proof of Useful Work (Filecoin/Helium), it does not require external hardware attestation; the "useful work" is verifiable at the protocol level via signature verification.

The paper concludes that PoUA offers a defensible security profile for attestation-native chains that generic L1s cannot replicate without rebuilding their consensus layer, provided the mechanism is stress-tested via devnet calibration and external review.