# Error handling in DKG

- We are primarily interested in a green-path scenario
- In order to make error handling / fault tolerance / etc., easier, we're going to bucket all issues into two buckets
- "Deviation from the norm will be punished unless it is exploitable"
- Soft deviation
- Recoverable error state
- Results in degradation of the quality of the service
- Example: Increased latency
- Nodes are punished by a "slap on the wrist"
- Hard deviation
- Unrecoverable error state
- Results in a critical failure of the protocol
- Example: Failure to complete DKG ceremony within the bounds of the user-selected security parameters
- Nodes are punished by slashing
- Track-record of nodes is stored on-chain and can be indexed and queried
- This information may be used, for example, to prevent sampling certain nodes
- Product discussion needed