Kryptografie

Worum es geht

The science of protecting information and computation against adversaries: confidentiality, integrity, authenticity and the key management that binds them. Founded as engineering doctrine by Kerckhoffs (1883), as mathematics by Shannon (1949), and as a public discipline by Diffie and Hellman's public-key revolution (1976) with RSA (1978) as its first realisation; matured by the provable-security program of Goldwasser and Micali (1984) into the definitions-assumptions-proofs method that Katz and Lindell's textbook canonised. Its deployment history — padding oracles, WEP, the Debian entropy bug, Dual EC — is not a blooper reel but the discipline's empirical half, and its future is a live migration: NIST standardised the first post-quantum algorithms in 2024 against a quantum computer that does not yet exist.

Was im Paket steckt

Kuratiertes Gerüst, das dein Agent vor der Recherche lädt, damit er auf geprüftem Boden startet statt bei null.

• Tier 1: 22
• Tier 2: 8
• Tier 3: 1
• Tier 4: 2

Die Fragen und Behauptungen unten sind wörtlich aus den Paketdateien zitiert. Die Pakete sind durchgehend englisch, weil sie für Agenten geschrieben sind.

Wo das Feld sich uneinig ist

Jede trägt echte Vertreter auf mehr als einer Seite, damit dein Agent nicht still einen Sieger kürt.

• Can lawful government access to encrypted data be engineered without degrading security for everyone?

  4 benannte Positionen · politically unresolved and cyclically relitigated; the field's published technical consensus sits on one side while legislative pressure recurs on the other

• After Dual_EC_DRBG, can NSA-adjacent standards bodies be trusted, and what rebuilds trust?

  4 benannte Positionen · open competitions are the accepted mechanism and their outputs are widely deployed; residual distrust is rational, priced in, and part of the subject

• Is a provable-security reduction knowledge about the real world or an internally consistent formalism?

  4 benannte Positionen · the method won the field; its epistemic interpretation is still argued in print, and honest teaching states what a given proof does and does not cover

• How urgent is the quantum threat, and are the replacement schemes mature enough to bet on?

  4 benannte Positionen · migration is proceeding regardless of the metaphysics; pace, parameters and portfolio composition are the live disputes

• Should attacks be published, and what do cryptographers owe the people their work protects or exposes?

  4 benannte Positionen · academic publication norms are settled in favour of disclosure; state stockpiling and the field's ethical self-description remain contested

Mythen, die das Paket korrigiert

Verbreitete Behauptungen mit der Evidenz, die sie klärt oder begrenzt.

• “If the data is encrypted, it is secure.”

  debunked

  Encryption addresses confidentiality against a stated attacker, nothing else. Integrity, authenticity, key management, randomness, endpoints and side channels are separate problems, and each has sunk encrypted systems: WEP encrypted every packet and fell to its key schedule; padding-oracle attacks decrypt CBC traffic without touching the cipher; Heartbleed read keys out of a correctly encrypting process; the 2013 Adobe breach exposed 150 million "encrypted" passwords because reversible encryption was used where one-way hashing belonged. Shamir's summary stands: cryptographic systems are bypassed, not penetrated.

• “Rolling your own cipher or protocol is fine as long as you keep the design secret.”

  debunked

  Kerckhoffs settled this in 1883: assume the enemy knows the system, because designs leak, ship in firmware and get reverse-engineered, while keys can be replaced. The record since is uniform — GSM's secret A5/1 was reconstructed and broken; the MIFARE Crypto-1 chip cipher fell within months of reverse engineering (2008); DVD CSS lasted until one licensee slipped (1999); WEP was designed by committee without public cryptanalysis. The schemes that survive (AES, SHA-3, the NIST PQC winners) are precisely the ones that stood years of open, adversarial review. Secrecy of design is not a layer of defence; it is a way of skipping review.

• “Encoding, hashing and encryption are basically the same thing — ways of scrambling data.”

  debunked

  Three different objects with different contracts. Encoding (Base64, URL encoding) is keyless and reversible by anyone — zero security by construction. Encryption is keyed and reversible with the key — confidentiality, when used correctly. Hashing is keyless and one-way to a fixed digest — integrity and fingerprinting, with no decryption to speak of; a MAC is a fourth thing, a keyed integrity check. The conflation has a body count: Adobe 2013 encrypted passwords (with 3DES in ECB mode, hints alongside) where hashing was required, and the recovered plaintexts became a public crossword puzzle; systems still ship Base64 as "obfuscation". Which primitive fails, and how, depends entirely on which contract you actually needed.

• “Quantum computers break all cryptography.”

  debunked

  Shor's algorithm breaks the number-theoretic public-key families — RSA, finite-field and elliptic-curve Diffie-Hellman and signatures — given a large fault-tolerant quantum computer that does not yet exist. Against symmetric ciphers and hash functions the best known quantum attack is Grover's quadratic search speedup: AES-256 and SHA-2/SHA-3 remain far out of reach, and doubling parameters restores margins. Lattice-, hash- and code-based schemes have no known efficient quantum attacks — which is what NIST standardised in 2024. The genuine present-tense threat is narrower and duller: recorded traffic decrypted retroactively if migration lags (harvest-now-decrypt-later).

• “The one-time pad is unbreakable, so a product built on it is the most secure choice.”

  debunked

  The theorem requires a truly random key as long as the message, used once, never reused, and delivered securely — which is the original key-distribution problem restated, and why the construction is impractical almost everywhere. Products claiming the one-time pad either generate the pad pseudorandomly (making it a stream cipher with extra steps and no proof) or reuse pad material, which is fatal: the Venona project read Soviet one-time-pad traffic for years because pads had been duplicated. Perfect secrecy is a boundary theorem that motivates computational cryptography — not a product category.

Verwandte Themen

• Bitcoin