[PROPOSAL] Standardizing DID Key Recovery: Three Viable Architectures (Social, Temporal, and Distributed)

Dear CCG Community,

Within the UDNA CG w3c-cg/udna: This repository holds the work of the W3C
Universal DID-Native Addressing (UDNA) Community Group. We are developing a
standard for addressing Decentralized Identifiers (DIDs) on the web.
<https://github.com/w3c-cg/udna>, we have been exploring a critical gap in
the DID ecosystem: standardized key recovery. We believe the path toward
global adoption requires a robust approach to recovery that avoids both
permanent identity loss and vendor lock-in.

After extensive research and prototyping, we have identified three distinct
approaches that appear viable across different operational and trust
models. We are seeking community feedback on these mechanisms to advance
this work toward a formal specification.
The Three Proposed Models1. Social ZKP Recovery (Threshold-Social)

Analogy: Like giving pieces of a treasure map to five trusted friends,
where any three can come together to find the treasure—but they never
actually see the map.

The Technical Model: Threshold trust with zero-knowledge proofs (ZKP).

How it works: Uses Feldman Verifiable Secret Sharing (VSS). Recovery keys
are split into t-of-n shares. Guardians hold shares but never reveal them
to the user or each other. Recovery uses Schnorr ZKPs to verify share
possession.

The Magic: No single guardian can reconstruct the key, and even t-1
colluding guardians learn nothing. Your friends prove they could help
without ever seeing your key.

Use Case: Individual sovereignty, family protection, and communities with
existing trust networks.
2. Deterministic Temporal Recovery (Time-Lock)

Analogy: Like a safety deposit box that automatically opens after one year,
but only for your chosen beneficiary—and the lock itself is mathematical,
not dependent on a bank employee.

The Technical Model: Time-locked recovery via Verifiable Delay Functions
(VDFs).

How it works: A master seed is encrypted with a VDF-derived key. A
beneficiary must compute the VDF for a specified duration (acting as a dead
man's switch).

The Magic: No one can cheat the clock. The computer must perform sequential
calculations that cannot be parallelized. It is a mathematical guarantee of
time, like a digital hourglass.

Use Case: Inheritance planning, long-term asset custody, and
cross-generational identity transfer.
3. MPC-Mediated Recovery (Distributed-Provider)

Analogy: Like a bank vault requiring three officers to open simultaneously.
Every month, the locks are changed, so a stolen key from last month is
useless.

The Technical Model: Threshold signatures with proactive refresh.

How it works: Uses fROST (Flexible Round-Optimized Schnorr Threshold
signatures). Key shares are distributed across independent providers.
Proactive Secret Sharing (PSS) refreshes shares periodically (epoch-based
synchronization).

The Magic: The key never exists in one place. Even if an attacker
compromises a provider, the stolen share becomes useless after the next
automatic refresh cycle.

Use Case: Enterprise identity management, institutional custody, and
high-assurance environments.
Why These Three

These three models emerged from analyzing real-world recovery scenarios.
For users who trust their friends and family but not completely, social ZKP
recovery fits. For those who want their children to inherit their digital
identity, temporal recovery applies. For organizations requiring bank-grade
security with multiple independent authorities, MPC-mediated recovery is
appropriate.

Together, they form a comprehensive toolkit that can be mixed and matched
based on user needs and risk profiles.

Repository: github.com/sirraya-labs/did-kr
Questions for the Community

Do these three models cover the majority of use cases the community is
currently seeing?

Should these be specified as separate extensions or unified under a single
recovery verification relationship?

Are there existing CCG work items we should align with or build upon?

Are there DID Method implementers interested in testing these recovery
hooks or reviewing a draft?

We are seeking initial feedback over the next two weeks to determine
interest in a unified specification. Pending community interest, we intend
to submit this as a formal CCG Work Item and move toward registering these
as a DID Extension.

We are looking for implementers interested in building wallet support, use
case contributors with additional scenarios, and co-editors , co-sponsors
to help shape the specification.

Looking forward to your thoughts and feedback.

Best regards,

Amir Hameed Mir
Founder, Sirraya Labs
On behalf of the UDNA CG
amir@sirraya.org

Universal DID-Native Addressing (UDNA) | Community Groups | Discover W3C
groups | W3C <https://www.w3.org/groups/cg/did-native-addr/>