What should research institutions do to deploy resilient data architecture?

A subsidiary question under Q-0008 (concrete actions). Where Q-0008 spans all three audiences (institutions, funders, researchers), Q-0015 isolates the institutional half.

The answer is the institutional cluster of recommendations:

• R1 (C-0040): Conduct an architectural audit of existing data infrastructure within twelve months, classifying every dataset by its current architectural tier and recording its exposure to each §3 failure mechanism.
• R2 (C-0041): Deploy at least one Tier 3 preservation node — a BitTorrent seeder, a Tor relay, a Forgejo instance, an IPFS pinning node, or an AT Protocol personal data server — on existing institutional infrastructure within twelve months.
• R3 (C-0042): Modify the data deposit workflow so that every dataset produces, at the point of deposit, the verification artifacts (content-addressed identifiers, cryptographic hashes, signed attestations, machine-readable metadata) that funder mandates are converging on.

These three recommendations together specify the institutional response: audit what you have, deploy what you don't, instrument the workflow so compliance becomes a byproduct of operation rather than a manual procedural follow-through.