Currently, when an E3 computation fails, there is no mechanism to:
- Detect and mark the failure on-chain
- Refund the requester for work not completed
- Compensate honest nodes for work they already performed
This creates poor UX for requesters (they lose their entire payment even if no work was done) and economic unfairness for nodes (they don't get paid for partial work completed).
Currently, when an E3 computation fails, there is no mechanism to:
This creates poor UX for requesters (they lose their entire payment even if no work was done) and economic unfairness for nodes (they don't get paid for partial work completed).