# Decentralized Identifiers
- Daniel Burnett
- Brent Zundel
- [Issue tracker](https://github.com/w3c/did-core/issues)
- [Mailing List](firstname.lastname@example.org)
## Table of Contents
- [Key scenarios](#Key-scenarios)
- [Detailed design discussion](#Detailed-design-discussion)
- [Considered alternatives](#Considered-alternatives)
- [Stakeholder Support or Opposition](#Stakeholder-Support-or-Opposition)
- [References and acknowledgements](#References-and-acknowledgements)
A decentralized identifier, or DID, is a new URI scheme. Unlike existing HTTP URLs, email addresses, commercial account identifiers, most government identifiers, etc., the DID was purpose-designed to support cryptographically-controlled identifiers backed by a decentralized oracle (such as a distributed ledger or blockchain).
Syntactically, a DID looks like this: `did:btcr:xyv2-xzpq-q9wa-p7t`. The three parts of a DID, separated by colons, are:
* The string “did”, representing the `did` URI scheme.
* A *DID method*, in this case `btcr`, that must be one of the values in the to-be-created DID Registry
* A string that is interpreted according to the rules of the specific DID method, in this case `xyv2-xzpq-q9wa-p7t`.
* Key management flexibility
* Management without dependence on external authority
* More flexibility for credential issuance and usage
* More flexibility for cryptographically anchored transactions
* Key discovery
* Permanent identifiers, that can be used for various use cases:
* Issue of long term documents
* Unique representation of artifacts
* Transient identifiers, that can be created and destroyed at ease
* Low cost
* Minimal or zero administrator
* Service discovery
* Provide a purpose-built solution
* Existing protocols may offer some of these features, but they are either incomplete or are not purpose-built with our features in mind.
- Creating decentralized identities
- Initiating, facilitating, or aiding a complete disruption of existing systems
- On the contrary, a pragmatic, incremental specification process can minimize disruptions in an incremental transition
- Replacing DNS
- Guaranteeing high-level or general-purpose identity assurance
- Content-addressed data stores
- For that, you might want to look into IPFS, DataShards, the Textile project, or IPLD
- Human-readable identifiers
- Defining or specificing DID-based authentication or authorization mechanisms
- We do everything possible to support those efforts and align on primitives, but those topics are out of scope here
- Identifiers for bearing or binding to Personally Identifiable Information (PII)
- We do everything possible to make DIDs useful identifiers for sophisticated PII-handling systems, but any binding of PII to discoverable identifiers carries some form of risk
- Other non-goals are these from the essay "7 Myths of SSI" by Timothy Ruff ([part 1](https://medium.com/evernym/7-myths-of-self-sovereign-identity-67aea7416b1) and [part 2](https://medium.com/evernym/7-myths-of-self-sovereign-identity-b16648c3090d))
- Self-sovereign means self-attested.
- SSI attempts to reduce government’s power over an identity owner.
- SSI creates a national or “universal ID” credential.
- SSI gives absolute control over identity.
- There’s a “main” issuer of credentials.
- There’s a built-in method of authenticating.
- User-centric identity is the same as SSI.
## Key scenarios
For key scenarios, please see [Use Cases and Requirements for Decentralized Identifiers](https://w3c.github.io/did-use-cases/)
We specifically encourage reviewing the [set of focal use cases](https://w3c.github.io/did-use-cases/#focalUseCases) from that document.
## Detailed design discussion
### Identifiers, not Identity
It is important to note that a DID is a *decentralized **identifier***, and not a *decentralized **identity***. While a DID could be used as an identifier in an identity system, the standard itself is expected to only define the identifier itself, making no claims about interpretability as an identity.
Although the standard will not expressly enforce privacy, it is a strong goal of the work that DIDs, and specific DID methods, can be used in ways that strongly reduce correlatability of personally-identifiable information.
### DID Methods
A key component of the decentralized identifier specification is the notion of [DID methods](https://w3c.github.io/did-core/#methods). DID methods are a means by which DIDs are "decentralized", as any number of DID methods may be created, each with different properties and assurances. The DID specification defines what a conforming DID method specification must contain. A DID method specification describes a particular type of DID and defines how it may be created, updated, resolved, or deactivated. It also must provide security and privacy considerations for that DID method.
### Assertion of Control of a DID
A key property of a DID is that it has a controller. Commonly, this entity will use public-key cryptography to assert control of a DID. This assertion of control may be used as an authentication mechanism. The standard does not define new authentication mechanisms, but does reference existing ones.
### DID Documents
For some DID methods the URI string itself is sufficient to understand how to verify control over the DID, including key rotation and revocation. An example is the “key” method, which interprets the third part of the DID as a public key. To support other, more complex, DID interactions, the concept of a *DID document* is introduced. This document typically provides other details regarding how to assert control over the DID or connect with the DID subject. The DID Document for a DID may change over time, but at any moment, there is only one DID Document for a DID. A DID method defines how a DID resolves to a DID document for that method. [Here is an example DID document.](https://w3c.github.io/did-core/#example-2-minimal-self-managed-did-document)
### DID Resolution
Just as with HTTP URIs, DIDs are expected to be resolved and dereferenced. As per https://tools.ietf.org/html/rfc3986, *resolution* is the process of determining the operations that can be performed on/with the resource identified by the DID, along with any [access control or authorization](#Assertion-of-Control-of-a-DID). Resolution results in a [DID Document](#DID-Documents) and possibly some metadata about the resolution process. Though the resolution process itself is out of scope for this version of the DID specification, the specification does define abstract function signatures that a conforming DID resolver must implement against. For more detail, [please refer to the specification.](https://w3c.github.io/did-core/#did-resolution)
### DID Dereferencing
*Dereferencing* is the process of performing the operations identified during resolution. One directly supported capability today is the use of service endpoints. [Service endpoints](https://w3c.github.io/did-core/#service-endpoints)
## Considered alternatives
The alternatives that prompted the creation of DIDs were considered well before the initiation of the Working Group to develop this specification from the preliminary work done by the Credentials Community Group. Listed here are some of the other identifier approaches whose characteristics motivated the creation of DIDs.
### Proprietary account names
A common identifier nowadays is a username or email address. These identifiers tend to be issued by entities that can, administratively, change or remove them without the permission or even knowledge of the identifier holder. DIDs were explicitly created to allow for an identifier that could not be modified or taken away from a controller by any third party.
### Open ID / OIDC
One possible [use for DIDs is as an authentication mechanism](https://www.w3.org/TR/did-use-cases/#authenticate). While at first glance it may appear that DIDs compete with existing authentication mechanisms such as OIDC or WebAuthn, they are actually complementary to such technologies and with some effort [may even be incorporated with them.](https://medium.com/decentralized-identity/using-openid-connect-with-decentralized-identifiers-24733f6fa636)
### Other URI Schemes
Some of the uses of DIDs may be possible with existing URI schemes. What DIDs provide that the others do not is the unique combination of the following characteristics:
- proof of control,
- the ability to define specialized DID methods,
- the simplicity of DID creation,
- the ability to rotate keys and change endpoints associated with an identifier without needing to alter the identifier, and
- the potential for universal resolvability.
## Stakeholder Support or Opposition
There is broad interest in and support for DIDs across the identity community. Early in the Working Group's history there was a concern around DID Documents only having a JSON-LD-based format defined. Since then the specification has added both JSON-only and CBOR formats. The number of entries already in the [DID Method Registry](https://w3c.github.io/did-spec-registries/#did-methods) is evidence of the interest in this work.
## References and acknowledgements
In alphabetical order, many thanks for valuable feedback and advice from Juan Caballero and participants in:
- Decentralized Identify Foundation (DIF)
- Internet Identity Workshop
- Rebooting the Web of Trust
- W3C Credentials Community Group
- W3C Decentralized Identifier Working Group