How does the incentive mechanism verify each endpoint's real usage without exposing its raw meter data?

The smart meter emits a cryptographic proof (e.g. a zero-knowledge proof) of its readings; the raw data goes only to the DSO, while the chain verifies the proof rather than the numbers.

Rewarding forecast accuracy requires checking everyone's actual usage — but publishing raw consumption on a shared ledger would destroy privacy. The resolution:

• At cycle end the smart meter collects the real consumption/production, generates a verifiable data proof, and sends the raw data only to the DSO.
• The proof is uploaded to the endpoint's on-chain contract; the DSO computes each participant's prediction deviation and passes the results to the consensus contract, which verifies them and announces the next proposer.

Using zero-knowledge proofs — proofs that a statement is true while revealing nothing beyond its truth — the network can confirm "this endpoint's reported deviation is correct" without ever seeing the household's actual numbers. Verifiability and privacy are satisfied at once.

Go deeper:

• Zero-knowledge proof (Wikipedia) — Proving a statement is true while revealing nothing beyond its truth.