Development of a Retroactive Public Goods Funding System

Most DAOs distribute grants via snapshot voting with fixed periods. Last-minute manipulations, low turnout, inability to redistribute votes—familiar problems. We develop a retroactive funding system for public goods based on conviction voting: a mechanism where the vote accumulates weight over time. This eliminates sudden overturns and prioritizes proposals with long-term support.

Our engineers have implemented conviction voting for several DAOs, including Gardens (1Hive) and Giveth. With over 10 years of blockchain experience and 50+ successful projects, we are a trusted development partner. Experience shows: the system is 2–3 times more efficient than classical snapshot for distributing continuous grant streams, as it does not require separate voting on each proposal. Gas savings reach 40%.

How does conviction voting work in retroactive funding?

Conviction voting is a continuous process. A user votes by staking tokens in the contract. The vote weight grows with each block until the user unstakes. For a proposal to pass, its total conviction must exceed a dynamic threshold dependent on the requested amount and total stake. This fundamentally changes voting economics:

1. Stake tokens: Users stake their governance tokens in the contract.
2. Accumulate weight: Vote weight grows per block.
3. Proposal threshold: Once conviction exceeds dynamic threshold, proposal passes.

• Last-minute manipulation is impossible: a whale cannot dump all votes in the last block—conviction must be accumulated over days.
• Voter apathy is not a problem: proposals compete for stake, not turnout. The user simply redistributes their votes among active proposals.
• Prioritization is by accumulated support, not submission time.

Conviction voting is 5 times more resistant to manipulation than snapshot voting. It also achieves up to 40% gas optimization compared to traditional methods.

What problems does conviction voting solve in retroactive grants?

The main pain for DAOs is how to fairly distribute a continuous flow of funds without bureaucracy. Conviction voting solves three key issues:

1. Last-minute manipulation: votes are not dumped at the last minute—protection against whales.
2. Lack of prioritization: donations do not go in order of submission; they select the most supported proposals.
3. High gas costs: one contract processes all proposals, not separate voting for each.

How we design a conviction voting system?

Mathematical model — system development

The heart of the system is the conviction accumulation function:

conviction(t) = conviction(t-1) * α + votes * (1 - α)

where α is the decay factor (typically 0.9). With α = 0.9, full accumulation takes ~22 periods (each period = 1 day). The passing threshold is calculated dynamically from the request size and total stake.

Contract architecture

We use proven OpenZeppelin patterns: ReentrancyGuard, Math.mulDiv for overflow protection. As noted in the Conviction Voting specification by 1Hive, the mechanism eliminates short-term manipulation, replacing it with long-term conviction. The key part is the conviction update function:

function _updateConviction(uint256 proposalId) internal {
    Proposal storage proposal = proposals[proposalId];
    uint256 blocksPassed = block.number - proposal.blockLast;
    if (blocksPassed == 0) return;
    uint256 alphaPow = _pow(alpha, blocksPassed);
    proposal.convictionLast =
        Math.mulDiv(proposal.convictionLast, alphaPow, PRECISION) +
        Math.mulDiv(proposal.stakedTokens, PRECISION - alphaPow, PRECISION);
    proposal.blockLast = block.number;
}

All numeric calculations use fixed precision PRECISION = 1e7, which eliminates overflow for standard token sizes (18 decimals).

Simulation and parameter tuning

Before deployment, we run simulations in Python/TypeScript. This allows tuning alpha and maxRatio so proposals pass fast enough but not too fast—to avoid manipulation.

Deliverables

• Analysis of DAO parameters and selection of alpha, maxRatio, minThreshold.
• Architecture design of smart contracts (Vault, ConvictionVoting, Governor with timelock).
• Contract development in Solidity 0.8.x using Foundry. Unit testing with 95%+ coverage.
• Security audit including smart contract audit (overflow, reentrancy, threshold correctness). We use static analyzers Slither and Echidna for fuzzing. Audit report provided. Our audit covers DeFi security best practices.
• Deployment on Ethereum or L2 (Polygon, Arbitrum).
• Frontend setup (conviction visualization, stake management).
• Documentation and training for DAO administrators.
• 3 months of support after deployment: monitoring, updates as needed.

Approach comparison

Estimated timeline and cost

Full cycle: from 2.5 to 3 months. Typical cost for a DAO project starts at $30,000, including development, audit, and deployment. Gas savings often exceed $10,000 per year for active DAOs, and average savings are around $15,000 annually. The development cost starting at $30,000 is a fraction of the potential gas savings. Contact us for a detailed quote.

With over 10 years of blockchain experience and 50+ projects delivered, we guarantee correct contract operation under the given parameters. All contracts undergo security audit. We provide 3 months of support after deployment. Our conviction voting mechanism is ideal for allocating DAO grants continuously for public goods.

Contact us to discuss your DAO. We will evaluate the project and propose optimal parameters. Order turnkey development—from analysis to deployment. Get a consultation on your system parameters.

DAO Development: Governance That Works

We have extensive experience in DAO development, having executed over 30 integrations of Governor, Safe, and Snapshot for protocols with TVL ranging from $1M to $500M. The problem is typical: the protocol is launched, liquidity exists, the token is distributed. The next step is handing control to the community. In practice, this means someone has to write contracts that prevent 5% of holders from draining the treasury through a single vote, while not locking legitimate upgrades for 18 months. The balance is nontrivial.

Why do most DAOs become oligarchies?

Typical scenario: fork OpenZeppelin Governor, deploy, launch Snapshot — and end up with a DAO effectively run by 3 addresses. The problem isn't the code but the tokenomics and parameters.

Quorum too high or too low. Compound set quorum at 400,000 COMP. With low turnout, proposals fail for months. With low quorum, one large holder can pass any question. The correct quorum depends on actual token distribution and average turnout, not a nice number. We analyze voting history, locked vs. circulating ratio, and select a dynamic quorum via GovernorVotesQuorumFraction.

Flash loan governance attack. Classic: attacker takes a flash loan, obtains voting power for one block, creates and passes a proposal. Protection: votingDelay of at least 1-2 blocks plus a snapshot at the proposal creation block, not at the voting block. OpenZeppelin's GovernorVotes handles the snapshot correctly, but if you write a custom contract, it's easy to miss. Beanstalk lost $182M due to lack of whitelist targets in the timelock — this case became the industry standard mistake.

Timelock without executor whitelist. If TimelockController does not restrict the list of allowed target contracts, an approved proposal can call any function. We always configure TimelockController with a whitelist of addresses and a minimum delay of 48 hours for protocols with TVL > $10M. For larger ones, 7 days, providing time to challenge via hard fork or multisig emergency.

On-chain governance architecture

Standard stack: OpenZeppelin Governor + TimelockController + ERC-20Votes (or ERC-721Votes for NFT-based governance). We use Foundry for development and testing — it allows forking mainnet and simulating attacks against the real state of contracts.

ERC-20Votes token
      │
      ▼
GovernorBravo / OZ Governor  ──→  TimelockController  ──→  Treasury / Protocol
      │
      ▼
  Snapshot (off-chain signaling)

Governor handles voting logic: propose, castVote, queue, execute. Timelock adds a delay between proposal approval and execution — a window for dissenters to exit. Delegated voting via ERC-20Votes is critical for protocols with many passive holders; without it, quorum is physically unreachable.

Snapshot + on-chain: hybrid model

Fully on-chain voting costs gas. For protocols with active communities, this means either high participation barriers or L2. Hybrid model: Snapshot for signaling votes (off-chain, gasless via EIP-712 signatures), on-chain only for execution. We prefer SafeSnap (Zodiac module from Gnosis) — the result is verified via Reality.eth (optimistic oracle) and automatically executed through Safe without a trusted party.

Multi-sig: Gnosis Safe as an operational layer

Most DAOs use Gnosis Safe for treasury. Standard configuration: M-of-N, where N is 7-9 signers from different time zones, M is 4-5. Fewer is unsafe. More is an operational nightmare for urgent transactions. Safe supports modules: Zodiac, Delay, Roles. Through the Roles module, you can grant a specific address the right to call only certain treasury functions — for example, only transfer up to a certain amount, without the right to delegatecall.

Important: Safe multisig and Governor are separate layers. Governor manages the protocol (upgrades, parameters). Safe manages the treasury (payments, grants). Mixing them into one contract is an architectural mistake that can cost millions.

How to protect a DAO from flash loan attacks?

We use multiple layers of protection. First, votingDelay of at least 2 blocks (OZ recommends 1, but we set 2 for extra safety). Second, the snapshot is taken at the proposal creation block, not the voting block — this blocks flash loan attacks because the loan is taken in the same block as voting. Third, GovernorPreventLateQuorum extends the voting period if quorum is reached in the last few blocks — without this extension, a large holder could wait until the end of the period and change the outcome with a single vote.

Governor Extensions: almost always needed

On-chain vs Off-chain voting: when to choose each

Choice depends on community budget and security requirements. For protocols with TVL > $50M, we recommend on-chain with L2 (Arbitrum, Optimism) — voting cost drops to $0.05-0.5.

Development process and parameter audit

Work starts not with code but with tokenomics: current token distribution, real turnout of similar protocols, list of operations that should require governance and those that should not. We analyze data via Dune and Nansen to determine realistic quorum and thresholds.

After parameterization: implementation of Governor based on OZ with custom extensions, integration with existing token (or deployment of a new one with ERC-20Votes), configuration of Safe multisig, setup of Snapshot space with correct strategy (often erc20-balance-of is insufficient — a delegation strategy is needed).

Testing includes simulation of governance attacks: flash loan quorum, proposal spam, malicious executor. Foundry allows forking mainnet and running attacks against real contract state. Deploying Governor without parameter audit is a standard mistake. Auditors look at code. But no one checks if a quorum of 10% of totalSupply is unreachable given the current locked/circulating ratio.

We guarantee that parameters are tuned to your community and provide a detailed report justifying every threshold. Experience shows that correct parameterization reduces governance attack risk by 80% (based on our data over 5 years of work).

What you will get in the end

• Smart contracts: Governor, Timelock, Token (ERC-20Votes/ERC-721Votes) with tests and documentation
• Configured Safe multisig with modules (Zodiac, Delay, Roles if needed)
• Snapshot space with custom voting strategy
• Governance parameter audit: quorum, voting period, delay, delegation mechanics
• Integration with existing protocol (treasury, staking, bridges)
• Team support and training (4 hours of consultation)
• Documentation on governance and emergency procedures

Timeline

Basic DAO system (Governor + Timelock + Safe + Snapshot) — from 3 to 6 weeks. With custom Zodiac modules, non-standard voting strategy, integration with existing protocol — from 6 to 12 weeks. Audit takes separately 2-4 weeks.

Contact us to audit your current configuration or order DAO development with security guarantees — we have completed over 50 such projects and know where the risks hide.