Overcommit Protection

The Overcommit Protection mechanism is a core reliability safeguard that balances capital efficiency with network security. It defines the maximum concurrent workload a prover can handle relative to their total stake, ensuring the network remains stable and responsive.

The Concept of Overcommitment

In a strict 1:1 collateral model, a prover would be limited to taking on tasks where the total minStake requirements exactly match their locked assets. However, requiring 100% collateral for every concurrent task can be capital-inefficient for reliable, professional operators.

Overcommitment allows the network to scale its throughput by permitting provers to manage a larger volume of “Active” tasks than their raw stake would normally allow. This ensures that the network’s computational capacity is not artificially capped by the speed of token liquidity.

Capacity Management

The marketplace tracks the Assigned Stake of every prover. A prover’s ability to bid on a new request is determined by the following eligibility rule:

Required Stake = Request.minStake + (AssignedStake * overcommitBps / 10,000)

• Assigned Stake: The sum of the minStake requirements for all requests currently assigned to a prover that have not yet reached a final state (Fulfilled or Refunded).
• The Overcommit Parameter (overcommitBps): A protocol-wide variable that determines how much “weight” is given to existing assignments when calculating a prover’s required collateral for a new job.
• Capacity Release: As soon as a request is completed, the associated minStake is removed from the prover’s Assigned Stake tally, immediately restoring their capacity to bid.

Throughput vs. Safety

The overcommitBps value is a vital lever for tuning the network’s economic health. Its value represents a direct trade-off between the total volume of proofs the network can handle and the level of collateral backing those proofs.

• Higher overcommitBps (Safety-Focused): By increasing this value, the protocol makes existing assignments “consume” more of the prover’s stake. This leads to higher collateralization per task and reduces systemic risk if a single prover fails, but it requires more total capital to maintain high network throughput.
• Lower overcommitBps (Throughput-Focused): A lower value reduces the “weight” of existing assignments, allowing a prover to take on a significantly higher volume of proofs with the same amount of total stake. This maximizes capital efficiency for provers but means a larger portion of the network’s active work is shared against the same pool of collateral.