Where does key collide?

The parameter that controls how a cipher encrypts and decrypts, and the only part of the system that is meant to stay secret. Kerckhoffs argued in 1883 that a cipher's security must rest on the key alone, since the algorithm will eventually become known; the modern reading is that everything but the key is public. Diffie and Hellman's 1976 paper New directions in cryptography split the idea in two: a symmetric key that both sides share, and an asymmetric pair in which a public key encrypts and a private key decrypts. The referent is a string of bits chosen from a large space, and its job is control over a transformation. It has nothing to do with a tonal center, a database row, or a map's symbols; the only thing shared with those meanings is the English word.[1][2]

Unit

A secret bit-string that parameterizes a cipher

Recovery means

Key recovery is what an attacker attempts; a strong cipher makes it infeasible

Measured by

Key length in bits and the work required to recover the key by search

The pitch and scale a passage is organized around, so that one note feels like home and the others lean toward or away from it. A key is named by its tonic and mode, as in C major or A minor, and it is written at the start of a staff as a key signature of sharps or flats. Rameau's 1722 Traité de l'harmonie grounded the sense of a governing center in the fundamental bass, and modern texts such as Aldwell, Schachter, and Cadwallader's Harmony and voice leading treat the key as the frame within which chords take on function. The referent is a relationship among pitches heard over time, not a secret and not an identifier. Sharing the word with the cryptographic meaning is a coincidence of English; a key signature controls nothing that a cipher key controls.[3][4]

Unit

A tonic and mode organizing a set of pitches

Recovery means

Not applicable

Measured by

The key signature written on the staff and the tonic that the music resolves toward

An attribute, or set of attributes, whose values identify a row in a relation and never repeat across rows. Codd's 1970 paper on the relational model made the key the way a row is addressed: a primary key names each tuple, and a foreign key holds a value that must match some primary key elsewhere, which is how tables refer to one another. The referent is a uniqueness and reference constraint over data, checked by the database. It is not a secret to be hidden, and unlike the cryptographic key it is usually visible; its value is not control over a transformation but the guarantee that one row can be told apart from every other. It shares only the word with a musical key or a map legend.[5]

Unit

An attribute set that uniquely identifies a tuple

Recovery means

Not applicable

Measured by

Whether the attribute values are unique and non-null across all rows