Social Entropy Layered DIDs (SEL-DIDs) is a novel framework that enhances Decentralized Identifiers (DIDs) with cross-platform behavioral and social signals to solve three critical problems in decentralized identity: Sybil Resistance, Account Recovery, and Trust Bootstrapping.

Key Features

• Sybil-Resistant Identity - Increases the cost of forging identities by requiring cross-platform behavioral consistency
• Privacy-Preserving Verification - Zero-knowledge proofs verify policies without exposing raw behavioral data
• Self-Sovereign Storage - All credentials remain in user-controlled data vaults (Decentralized Web Nodes)

Read the full paper: GitHub Repository

I've been thinking about identity on the internet for a while now. Not the "log in with Google" kind—the deeper question of how we prove we're real humans in systems designed to be trustless.

Decentralized Identifiers (DIDs) are supposed to solve this. Generate a keypair, control your own identity, no middlemen. Beautiful in theory. But in practice, there's a problem: anyone can spin up a thousand DIDs in seconds. They're cryptographically valid but meaningless for distinguishing real humans from Sybil attacks.

The Problem Space

I kept running into three problems that existing DID systems don't really solve:

1. Sybil resistance — How do you do "one person, one vote" without centralized KYC or iris scans?
2. Account recovery — Lose your private key and your identity is gone. Social recovery exists but reintroduces trust assumptions.
3. Trust bootstrapping — A bare keypair tells a relying party nothing about whether there's a real human behind it.

The Insight

Here's what I realized: most of us already generate massive amounts of behavioral evidence across platforms. Years of messaging patterns, social graphs, content consumption habits, posting rhythms. This data has entropy—it's hard to fake at scale and expensive to replicate.

The naive approach would be to derive keys from this social data. That's insecure and defeats the purpose. Instead, what if we treat behavioral evidence as a layered, revocable credential attached to a cryptographic root?

Introducing SEL-DIDs

Social Entropy Layered DIDs keep the cryptographic root intact—you still control a high-entropy keypair. But on top of that, you attach a Social Entropy Vector (SEV): an aggregated representation of your cross-platform behavioral patterns stored as verifiable credentials in your own data vault.

The key properties:

• No single platform is mandatory — the system degrades gracefully
• Privacy-preserving — relying parties see zero-knowledge proofs, not raw data
• User-controlled — all credentials live in your DWN, not a central database
• Revocable — you can rotate, split, or delete credentials at will

How It Works

Imagine logging into a DAO governance platform. Instead of just proving you control a keypair, you also prove—via zero-knowledge proof—that your SEV indicates at least 3 years of consistent activity across multiple independent platforms.

The DAO never learns which platforms, who your contacts are, or what content you consume. They just learn: this identity has high social entropy and is expensive to forge at scale.

Comparison with Existing Systems

BrightID relies on social graphs and verification parties—useful but narrow. Gitcoin Passport aggregates stamps from various sources into a score—better coverage but still treats proofs as external badges. Worldcoin uses iris biometrics—effective but centralized and controversial.

SEL-DIDs take a different approach: integrate behavioral evidence directly into the DID structure, keep everything user-controlled, and use ZK proofs for verification. No mandatory operator, no biometrics, no single point of failure.

The Cold Start Problem

I'm not pretending this solves everything. New users with thin online histories will have low-entropy SEVs. That's fine—SEL layers are optional. Bare DIDs work in contexts that don't need Sybil resistance, and alternative evidence (community attestations, offline ceremonies) can supplement thin histories.

What's Next

This paper is architectural—it defines the model, threat assumptions, and protocols. The next steps are: designing concrete feature spaces, empirically evaluating cost-of-forgery, and building prototypes for real governance and funding platforms.

If you're working on decentralized identity, Sybil resistance, or proof-of-personhood systems, I'd love to hear your thoughts.

Read the full paper: GitHub