Modern Cryptography
Different cryptographic tasks, constructions and security assumptions
The earlier halls distinguish cryptanalytic breaks, operational misuse, obsolescence, and unresolved systems; a correctly used one-time pad remains unbreakable. The constructions here serve different purposes. DES and AES are symmetric block ciphers; DES is now unsafe because its 56-bit key is too small. RSA supports distinct public-key encryption and signature schemes, with appropriate encoding and padding such as OAEP and PSS. Diffie–Hellman is key agreement, and SHA-256 is a hash function. These building blocks explain encryption, shared-secret establishment, signatures, and integrity; none alone provides a complete secure system.
For the broader landscape — ChaCha20, ECDH, post-quantum standardisation, and the cipher-vs-protocol taxonomy — see the Modern Cryptography wing. For the design rule that underpins every card in this hall, see Kerckhoffs's Principle.
DES was standardized in 1977. Its 56-bit key became vulnerable to practical exhaustive search, demonstrated by EFF’s Deep Crack in 1998.
Diffie–Hellman derives a shared group element over a public channel. Authentication, appropriate groups and key derivation are needed for secure protocol use.
RSA uses a public modulus and exponents to support public-key constructions. Secure encryption and signatures need specified padding, adequate parameters and correct implementation.
AES standardises a 128-bit block cipher with 128-, 192- or 256-bit keys. Secure use also requires an appropriate mode, key management and implementation.
SHA-256 produces a 256-bit digest for messages shorter than 2^64 bits. Hashing has distinct roles in signatures, integrity checks and other protocols; a bare hash is not authentication.
ECDSA signs with elliptic-curve arithmetic. The toy model shows how nonce reuse can expose a signing key under the stated conditions; it is not a production implementation.
Shamir’s finite-field construction lets k valid shares reconstruct a secret. Fewer shares preserve its prior distribution when coefficients are chosen uniformly and independently with distinct nonzero share coordinates.
Zero-knowledge formalizes what a proof reveals. This small Schnorr model illustrates honest-verifier zero knowledge and extraction from repeated commitments, not production security.
The museum’s closing lesson: Modern cryptography specifies security goals, attack models and assumptions. Public designs, suitable parameters and careful implementations matter together. Historical weaknesses do not imply that every classical system is broken, or that modern primitives are secure under every use.
Research case studies: what was actually attacked?
UA-8295: altered firmware and forensic evidence · SoDark: eight-round radio linking protection · NFC: authentication, configuration and counterfeit hardware
Compare GPU search, GPRS faults and conditional-linear DES. Each result identifies the target, attacker access and limits of the evidence.