The internet was originally created to connect computers, not humans. The design choices of the past render today's internet inadequate in safeguarding ethical and democratic values.
User data is collected, owned, and managed by a handful of powerful organizations, leading to a reality where your digital identities are defined by what Google knows about you. Meanwhile, trust is broken on the internet—without a thorough background check, nothing can be verified or trusted. Social media platforms have become breeding grounds of misinformation, scams, hate speech, and bots.
Self-Sovereign Identity (SSI), or Decentralized Identity, represents a suite of human-centric technologies designed to empower individuals with full control over their digital identities and personal data. Users granularly decide what information they share, allowing them to, for instance, verify their age without exposing their entire ID card or prove to landlords that their income is sufficient without handing over their bank statements.
The W3C-standardized Verifiable Credentials Data Model defines a trust relationship involving three key roles: the Issuer, Holder, and Verifier. For illustration, a government (Issuer) issues a passport (Credential) to a citizen (Holder). Traditionally, when a citizen checks in at a hotel (Verifier), they must physically hand over their passport to the receptionist, who then verifies and copies the passport. This process is time-consuming and susceptible to security risks, such as fraud and identity theft. Verifiable Credentials offer a significantly more secure and seamless authentication process.
Several democracies are actively developing VC-based digital credentials to empower their citizens. In alignment with the eIDAS 2.0 regulation, the European Commission is realizing its vision of a cross-border European Digital Identity for all EU residents by 2027.
At present, various types of Verifiable Credentials are in development, each carrying its unique advantages and limitations. Balancing the delicate trade-off between security and privacy can be challenging. As of today, BBS+ Verifiable Credentials are the most private, because they are the only type of VCs that can achieve Issuer Unlinkability, on top of Selective Disclosure and Verifier Unlinkability:
- Selective Disclosure allows users to select the type and amount of data shared with the Verifiers. For instance, one can verify their age without revealing their name to access a service, aligning with the Data Minimization principle outlined in GDPR.
- Unlinkability separates users’ actions from their identities. There are two layers of unlinkability: Verifier Unlinkability and Issuer Unlinkability. Verifier Unlinkability prevents Verifier-Verifier collusion but not Verifier-Issuer collusion. A VC is Issuer-Unlinkable if no cryptographic identifier known to the Issuer is shared with the proofs to the Verifier (e.g., public key or correlatable signature).
Two Use-Cases:
- Security-focused: With reusable digital identity, preventing impersonations and identity theft. Users include KYC providers and their clients, such as governments, banks, insurances, and other actors that already KYC their own users.
- Privacy-focused: With ZKP and unlinkability, providing proof of humanity to use-cases that generally don’t “officially” KYC their users for now but would benefit from doing so in a way that still respects user privacy, such as social media platforms.
The distinction between security-based and privacy-based VCs reflects the direction that the community is currently taking. However, within the next 5-10 years, there is a possibility that the two VCs would merge, particularly with the development of performant general-purpose client-side zkVMs (see the ZKP section).
Data can optionally be addressed via its semantics (JSON-LD, RDF, etc.)
Advantages:
- The only reasonable and relatively mature solution to efficiently fight against sybil attacks & identity theft, and create an identity layer for the Web (one account = one human). Alternatives are based on either biometrics (e.g., Worldcoin) or social graphs (e.g., Proof of Humanity). The former is unethical, insecure, and useless, while the latter currently does not scale, is not private, requires users to pay to create an account, and lacks maturity & adoption (see their corresponding sections).
- Adoption is slow but steady, with many major state actors running large-scale pilots or creating new digital identity regulations (e.g., eIDAS in Europe).
- Respects user anonymity (if done right, i.e., using privacy-focused VC flavor for use-cases that require maximum privacy).
- While identity wallet integration brings challenges, we are seeing promising paths forward to make interoperability with apps smoother, through standardization efforts from web standards organizations as well as from mobile OS vendors Google and Apple.
- Censorship-Resistance: No call-home & self-sovereign (if done right): The issuer is not supposed to be able to know that a holder has used a verifier’s service, and a given credential may be used by an infinite number of services, with a portion of them being of direct benefit to the issuer, incentivizing the issuer to issue the credential in the first place, besides regulation.
- The use of a decentralized verifiable data registry ensures that the whole process is auditable and secure.
- It’s a standard that can be used for data other than pure identity information (such as “reputation” data based on the Web of Trust, see DWN & Web5).
Limitations:
- Requires accepting a trust model that may not be suited for every situation. In particular, the issuer must be trusted by both the verifier and the holder. The issuer could issue wrong attributes to certain individuals and arbitrarily censor certain individuals from receiving credentials. If the issuer is corrupted, then the whole model collapses, even though having a decentralized Verifiable Data Registry allows for third parties to audit the issuer, to a certain degree.
- Truly decentralized, immutable, and permissionless Verifiable Data Registries probably need to rely on a blockchain, which costs money and is hard to scale. (Note that recent encouraging developments may allow the community to get rid of the necessity to use blockchains, at least for certain use-cases.)
- Not there yet technology-wise: lots of differing standards, lack of maturity.
- Practically no real adoption yet, in most parts of the world.
- Even if wallet adoption were widespread, interacting with third-party wallets to integrate with online services is a huge added friction for users, which they are not used to (similar problem as with crypto wallets).
- Reasonable fear of a dystopian Orwellian usage of the technology, involving the adoption of security-focused VC flavors, which do not provide issuer unlinkability, for privacy-focused use-cases.
- A large portion of SSI actors focus on classic client-server applications, ignore or compromise with censorship-resistance & privacy requirements (especially issuer unlinkability), and mainly focus on reusable KYC & the identity theft aspect of Verifiable Credentials (banking & governments being the targeted clients). The direct interests in SSI of the issuing organizations (be they public or private) are largely related to security rather than privacy. As a result, lots of lobbying is done to push security-flavored high-assurance Verifiable Credentials, while the lobbying towards privacy-preserving credentials is pushed by a minority, often close to the DWeb & Web3 movements.
- Lack of trust by a portion of the general population towards their institutions.
- Because of its underlying technological disruption, the process of zkKYC (zero-knowledge KYC) is not intuitive and creates friction: “verify your identity now, so you can later post anonymously!”
- Not absolutely censorship-resistant: issuers can still arbitrarily revoke credentials from users—or impersonate them, even though they can’t do it privately if a decentralized verifiable data registry is used, and as a result, issuers could be held accountable for any misuse. Nevertheless, there is an assumption of trust between the holder and the issuer built into the model—which is not necessarily a bad thing depending on the political bias and context. Our opinion on this matter is that human identity does not exist in a vacuum, that society is built upon trust, that governance cannot be magically solved by a machine, that politics is part of our human nature, and that digital identity is only a tool that can be used to hold human organizations accountable, and to uphold the fundamental human rights of each and every member within these organizations. Among all the existing innovative identity systems that attempt to solve proof of personhood, apart from early-stage attempts to create decentralized governance which would get rid of trusting the issuer such as Proof of Humanity, we think this model is the most transparent and trust-minimized identity model one can wish for.
Could it be useful for our requirements?
SSI and the privacy-preserving flavors of Verifiable Credentials are the crucial credential infrastructure that we need to provide an authentication system that is both privacy-preserving and protects against bots and sybil attacks, with comprehensive support for hard security issues such as account recovery and revocation.
Verifiable Credentials are also highly interoperable across protocols, which is a wanted feature.
