Anti-Abuse System
The Threat
Operator finds a bug → kills the pool to avoid sharing → submits the finding under a different identity → steals the full bounty.
Defense Layers
1. 30-Day Watch Window
When an operator kills a pool:
- The operator × target combination is watched for 30 days
- Any finding on that same target within 30 days triggers an automatic dispute
- No exceptions. Automated enforcement.
2. Minimum Compute Burn
- 40% of compute must be spent before a kill is allowed
- Prevents operators from creating pools, quickly peeking at results, and killing early
- Ensures meaningful work was done before termination
3. Timing Correlation
- Finding appears on source platform suspiciously close to a pool kill? Auto-flagged for review
- Correlation analysis compares pool activity timestamps with source platform submissions
- No wallet linking needed — timing patterns are the signal
4. Community Reporting
- Sponsors can open disputes with evidence
- Crowdsourced watchdog — the community has financial incentive to catch cheaters
- 3+ flags on an operator = automatic review and temporary suspension
5. Reputation Nuke
If caught:
- Permanent ban from the platform
- All pending earnings frozen
- Public shame on profile (visible to all users)
- Stake slashed (100% of collateral)
The penalty is designed to be so severe that the attack is never worth attempting, regardless of bounty size.
6. Finding Hash Commitment
- Cryptographic proof that findings existed within the pool at a specific time
- If the same finding appears from a "different" hunter, the hash proves the pool had it first
- Can be used to dispute with the source platform directly
Kill Rate Tracking
- Operator kill rates are tracked and visible on their profile
- Too many kills = red flag for sponsors
- Pattern of kills followed by same-target submissions = automatic investigation