Technical Summary: Concave is the New Linear

Problem Statement

Decentralized Autonomous Organizations (DAOs) rely on token-based governance, where voting power is typically proportional to wallet balance. This "one-token-one-vote" model creates a plutocracy, concentrating control among large holders ("whales") and leaving protocols vulnerable to token-control attacks. To counter this, the community has proposed anti-plutocratic mechanisms, most notably Quadratic Voting (QV) and other concave voting rules (e.g., power rules $w^\beta$, logarithmic rules). These mechanisms assign sublinear voting power per token (e.g., $\sqrt{w}$ votes for $w$ tokens) to dampen the influence of large holders.

However, these mechanisms were designed under the assumption of a single, stable identity per voter. In permissionless blockchains, this assumption fails due to Sybil attacks, where an adversary creates multiple fake identities (wallets) to split their token holdings. The central problem addressed by this paper is whether any wallet-level voting rule that derives power solely from balance can effectively resist a rational Sybil attacker on a permissionless chain. The authors investigate if concave rules can truly prevent plutocracy or if they inadvertently create a new form of linear vulnerability.

Methodology

The authors develop a unified, costed mathematical model to analyze DAO governance under Sybil attacks.

1. Model Parameters:

   • $a$: The attacker's total token budget.
   • $f(w)$: The wallet-level voting rule (e.g., linear $w$, quadratic $\sqrt{w}$, power $w^\beta$, logarithmic $\ln(w+1)$).
   • $C = (m, v, p, s)$: A cost scheme capturing realistic blockchain frictions:
     • $m$: Minimum balance required to vote.
     • $v$: Per-wallet voting cost (gas).
     • $s$: Per-wallet splitting cost (gas to transfer tokens).
     • $p$: Fixed one-time setup cost (e.g., deploying a splitter contract).

2. Optimization Problem:
   The attacker seeks to maximize total voting power $V^*(a)$ by splitting their budget $a$ into $n$ wallets with balances $w_i$, subject to the cost constraints:
   $$\sum w_i + n(s+v) + p \leq a$$
   The objective is to maximize $\sum f(w_i)$.

3. Analytical Approach:

   • Lower Bounds: The authors first prove that for any non-trivial voting rule (where $f(m) > 0$), the attacker can achieve voting power that grows at least linearly with $a$ ($\Omega(a)$).
   • Asymptotic Analysis: For concave rules, they derive the optimal splitting strategy. Using Jensen's inequality, they show that an even split of tokens across wallets is optimal. They then optimize the number of wallets $n$ to find the asymptotic behavior of the maximum voting power.
   • Empirical Instantiation: The model is applied to real-world data from five major DAOs (ENS, Compound, Uniswap, Arbitrum, ZKsync). The authors replay the ten most recent finalized proposals for each, calculating the cost for an attacker to capture a majority vote under linear, quadratic, logarithmic, and power ($\beta=0.25$) rules, using actual on-chain gas prices and token values.

Key Contributions

1. Unified Costed Model: A formal framework capturing the interplay between voting rules and realistic blockchain costs (splitting, voting, setup, and minimum balances).
2. Universal Impossibility Result: A proof that every non-trivial wallet-based voting rule is plutocratic under unrestricted Sybil splitting. If a wallet of any size yields nonzero voting power, an attacker's adjusted voting power grows at least linearly ($V^*(a) = \Omega(a)$), regardless of the specific concave function or linear cost scheme.
3. Tight Asymptotic Slope for Concave Rules: The authors refine the lower bound for concave rules, proving that the optimal voting power is asymptotically linear in token holdings. They derive a closed-form expression for the amplification constant $\kappa$:
   $$\kappa = \sup_{x \geq m} \frac{f(x)}{x + v + s}$$
   This demonstrates that concavity only provides a constant-factor dampening, not sublinear dampening.
4. Empirical Impact Analysis: Quantitative evidence showing that Sybil attacks are orders of magnitude cheaper than honest voting under concave rules.
   • Under Quadratic Voting, Sybil amplification factors range from 1,172× to 4,039×.
   • Under steeper power rules ($\beta=0.25$), amplification exceeds 229,000×.
   • Case Study: On Uniswap, an attacker can capture a $300M vote for roughly $75K in costs using a Sybil-optimal strategy, whereas a single-wallet attacker with the same budget achieves only $\approx 1/259$ of the necessary voting power.
5. Mitigation Assessment: A survey of candidate defenses (minimum balances, increased fees, Proof-of-Personhood, bicameral governance). The authors conclude that no per-wallet economic friction can eliminate asymptotic linearity; only identity-based layers (Proof-of-Personhood) or compositional defenses can effectively mitigate the threat.

Results

• Theoretical: The paper proves that "Concave is the New Linear." While concave rules reduce the voting power of a single large wallet, a rational attacker can split their holdings to recover linear voting power. The "anti-plutocratic" benefit is nullified by the ability to create Sybils.
• Empirical: The cost to attack major DAOs is drastically reduced by concave rules.
  • Uniswap: Linear cost \approx \302.6M$; Quadratic cost \approx \74.9K$ (4,039× cheaper).
  • Compound: Linear cost \approx \11.6M$; Quadratic cost \approx \6.4K$ (1,820× cheaper).
  • The "amplification factor" (how much cheaper the attack becomes) increases significantly as the voting rule becomes more concave (e.g., logarithmic and power rules are far more vulnerable than quadratic).
• Friction Analysis: Increasing per-wallet costs ($s, v$) or minimum balances ($m$) lowers the slope $\kappa$ (making the attack more expensive) but does not change the asymptotic linearity. The attack remains profitable for sufficiently large budgets.

Significance and Claims

The paper claims that no wallet-level voting rule can succeed in preventing plutocracy on a permissionless blockchain without an external identity layer.

• Rejection of Standalone Concave Voting: The authors argue that proposals to implement Quadratic Voting or similar mechanisms as the primary governance defense are fundamentally flawed in a permissionless setting. The "single-identity" assumption required for QV's optimality is unenforceable without identity verification.
• Redefining the Threat: The work shifts the focus from "token concentration" to "wallet splitting." It demonstrates that the very mechanism designed to protect small holders (concave voting) creates a massive economic incentive for attackers to split their tokens, ultimately making the system more vulnerable to control by a well-funded adversary than a simple linear system.
• Prescriptive Conclusion: The authors assert that concave voting must never be the primary line of defense. It can only be deployed as a component of a composite mechanism where an identity-based Sybil-resistance layer (e.g., Proof-of-Personhood) limits the number of wallets an adversary can control. Without such a layer, the "level playing field" intended by concave rules is inverted, allowing attackers to dominate governance at a fraction of the cost required in a linear system.

The paper concludes that decentralization requires securing the identity layer first; otherwise, mechanisms intended to empower small holders inadvertently destabilize the system for everyone.