What is a secure encryption system?
Surely, such a system should not allow to extract the original information (called the plaintext) from the encrypted form (called a ciphertext).
But is this all? Not really, since in many applications the extraction of some partial information (regarding the plaintext) may be harmful enough. Indeed, this harmful partial information may vary from application to application.
In 1982, a WIS scientist and her collaborator put forward a robust definition of security, which suffices for all possible applications. Furthermore, they (and subsequent work) showed that such a robust level of security can be achieved whenever a minimal level of security can be achieved. Their encryption scheme uses randomness as part of the encryption process, which is unavoidable for the robust definition of security.
The definitional approach underlying their proposal was adopted in almost all subsequent research in cryptography, which extends way beyond the domain of encryption (see Zero-Knowledge, Secure Multi-Party Protocols).
In 2012 this work served as the the basis for the bestowing of the Turing award on its co-authors.