Why Quadratic Voting is Economically Unsound for DAOs

QV's core defense—costly identity verification—is antithetical to permissionless blockchains. Projects like Gitcoin Grants rely on centralized social graphs (BrightID) to mitigate this, creating a governance layer that is not credibly neutral.\n- Cost of Attack: Sybil-ing a $1M proposal can cost less than $5k in borrowed capital.\n- Real-World Proof: a16z demonstrated trivial vote manipulation in early MolochDAO experiments.

QV assumes voters are independent actors. In reality, capital aggregates. A wealthy participant can simply bribe or pay a large number of pseudo-identities to vote quadratically on their behalf, a strategy formalized in bribery-resistant voting literature.\n- Result: The system collapses to one-dollar-one-vote but with extra steps and fees.\n- Protocol Impact: Makes DAOs like Uniswap vulnerable to low-cost governance takeover.

QV requires locking capital (e.g., Conviction Voting) to express preference intensity. This imposes a massive opportunity cost on participants, disincentivizing the very engagement it seeks to promote. Liquid governance tokens on Compound or MakerDAO are constantly pulled toward higher-yield opportunities.\n- Outcome: Only the most ideological (or poorly capitalized) members vote with high conviction.\n- Data Gap: No major DAO with >$100M TVL uses pure QV for core treasury decisions.

QV theoretically aggregates intensity of preference. On-chain, preferences are not private and are expressed via wallet signatures, creating a public game theory puzzle. Voters strategically under-reveal true intensity to avoid influencing others, breaking the mechanism's Vickrey-Clarke-Groves (VCG) inspiration.\n- Reality: Leads to herding and signal jamming, not true preference revelation.\n- Academic Verdict: Requires costly, private voting—impossible on transparent ledgers.

QV's core defense—that splitting capital is prohibitively expensive—is shattered by cheap on-chain identity. A whale can create thousands of pseudonymous wallets for marginal cost, dominating any vote.

• Cost of Attack: Fraction of a cent per Sybil vs. quadratic cost for honest voters.
• Real-World Impact: Renders the 'one person, one vote' ideal economically impossible to enforce without a centralized identity layer.

QV explicitly assumes voters are independent actors. In crypto, explicit coordination (collusion) is a feature, not a bug. Protocols like OlympusDAO and Convex are built for it.

• Mechanism Failure: Whales can bypass quadratic costs by pooling funds off-chain or using vote escrow tokens.
• Result: The system penalizes honest, distributed participation while rewarding the very centralized power structures it aimed to prevent.

QV imposes a cognitive and financial tax on every marginal vote, crippling voter turnout and governance throughput. This isn't a bug but a direct consequence of the math.

• Voter Apathy: Why spend 10x the gas for a vote that's 1/10th as powerful? Users abstain.
• Decision Paralysis: Complex proposals requiring multiple preference votes become economically prohibitive, leading to stagnation.

The poster child for QV had to fundamentally alter its model, proving the theory's fragility. It introduced pairwise bonding curves and retroactive funding to mitigate flaws.

• Admitted Failure: Pure QV was gamed by sybil farms and collusive rings.
• The Fix: Hybrid models that layer QV with other mechanisms, acknowledging it cannot stand alone in a permissionless environment.

QV's core premise—one-person-one-vote via cost scaling—is shattered by cheap identity fabrication. Attackers can split capital across sybil identities to gain disproportionate voting power for minimal cost, undermining any democratic ideal.

• Cost of Attack: Often less than $1 per identity on L2s.
• Result: Concentrated capital defeats distributed voice, replicating plutocracy with extra steps.

QV fails its own stress test against rational, coordinated capital. Whales can simply bribe or subsidize a coalition of sybils, achieving linear voting power at quadratic prices—still economically efficient for high-stakes proposals.

• Mechanism: Platforms like Gitcoin Grants demonstrate chronic collusion.
• Outcome: QV becomes a tax on disorganization, not a barrier to capture.

Requiring native token locking for voting creates severe UX and capital inefficiency. It disincentivizes participation from liquidity providers and delegators who optimize for yield, not governance.

• Capital Lockup: Tokens are stuck, not working in DeFi pools.
• Alternative: Snapshot with off-chain signaling or ERC-20Votes with delegation avoid this deadweight loss.

QV's transparent voting cost creates a gameable signaling environment. Voters are incentivized to misrepresent their true preferences early to manipulate others' cost calculations, or to 'free-ride' on anticipated outcomes.

• Dynamic: Becomes a poker game, not a truth-revelation mechanism.
• Compare: Conviction Voting or Holographic Consensus better aggregate sustained sentiment.

The theoretical link between QV and optimal public goods funding (Vickrey-Clarke-Groves) breaks down with incomplete information and non-binary choices. Real DAO proposals are complex, multidimensional, and sequential.

• Evidence: Gitcoin's rounds show funding concentration, not optimal distribution.
• Reality: Simpler plurality voting or weighted voting often yields more predictable, legible outcomes.

Architects should prioritize sybil-resistant identity (Proof-of-Personhood, BrightID, Worldcoin) paired with simpler, robust mechanisms.

• For Funding: Retroactive Public Goods Funding (Optimism, Ethereum Foundation).
• For Proposals: Conviction Voting or Streaming Votes (Vote Escrow models).
• Rule: Decouple identity from voting mechanics. Fix the root, not the formula.