What is the starting point of a quality model for digital identities, and which two kinds of quality does it distinguish?
A risk assessment of the application (the relying party, RP) determines how good the identity it relies on has to be; the model then separates the quality of the authentication from the quality of the attributes.
A relying party is the service that consumes an identity: a tax portal, an online bank, a library login. Before choosing any technology, it asks what happens if this identity is wrong? A forged login to a library account is an annoyance; a forged login to a tax return or a health record is a serious harm. The answer to that question, the risk assessment (Risiko-Analyse), sets the required quality.
The quality model then looks at two separate things:
- Quality of authentication (Qualität der Authentifizierung): how sure can the RP be that the person logging in really is the person the identity belongs to? This covers registration, the authenticator, the operator and, in a federation, the transport of the login confirmation.
- Quality of attributes (Qualität der Attribute): how trustworthy are the data delivered about that person, such as name, date of birth, address or age?
They are distinct because a perfectly authenticated user can still carry a wrong, outdated or self-invented address, and a correct attribute is worthless if anyone can log in as that user.
Go deeper:
Risk assessment (Wikipedia) — the general method of weighing likelihood and impact that feeds the choice of a level.