Update from Parano1d

V2 upgrades the proof system without restarting its history. A node joining after the v2 upgrade must still be able to verify how the present came from genesis. That requirement makes the upgrade itself part of the proof design.

The first v2 proof binds to the selected last pre-v2 block. Its authenticated origin carries the existing validity obligation into the new relation. A participant verifies this connection rather than trusting an operator's declaration that the old chain was valid.

The activation release contains both historical and new proof matrices. I also implemented and qualified the next step: authenticated certificates that let a later node verify the historical obligation while omitting the old matrix bytes. The new matrices and authenticated origin evidence remain.

The isolated qualification tested this with actual node processes.

A fresh node built without the legacy matrices synchronized from an empty database, reached the source's exact State and verified a contract receipt. It then restarted with no provider running and verified the retained evidence again.

Missing, old-format and corrupted certificates caused receipt verification to fail. Reconnecting a provider recovered valid evidence at the same chain tip.

A separate run covered both historical proof classes. After a contract call's body was pruned, a fresh node without the old matrices verified its receipt, then produced new blocks and a new call accepted by the ordinary transition peer.

All of this is documented as isolated-network qualification. A mainnet retirement certificate must bind to the actual selected activation predecessor.

The security accounting keeps the historical failure terms when old files are retired. The end-to-end post-quantum assessment includes the upgrade and retirement path under its published premises.

For Parano1d, independent verification from genesis has to survive software and proof-system changes as well as transaction pruning.

Certificates, cold synchronization and production records https://github.com/proof-native/parano1d/blob/main/research/v2_feasibility/results/2026-09-25-common-input-budget/REPORT.md