MIT OpenCourseWareLecture 9: Modern Encryption: Key Concepts
MIT OpenCourseWare
Encryption as a Third Leg 0:00
You are told that encryption is a separate tool alongside distributed ledgers and smart contracts, and it can be used with either. For ledgers, it lets financial counts be altered securely without requiring trust, producing immutable, indexed records. For contracts, it helps parties overcome private information, limited communication, limited commitment, and confidential data problems, letting two or more agents solve a mechanism design problem while keeping their messages secret and their commitments binding, even if both sides might otherwise want to renege.
Public and Private Keys 4:00
You learn that a public key is a mathematical object used to turn a message into ciphertext, which can be shared openly without revealing anything, while a private key is needed only to decrypt it, and the algorithms make decryption without that private key virtually impossible. The comparison used throughout is prime numbers: two large primes multiplied together are easy to check but extremely hard to factor back apart, and the private key behaves like one of those hidden factors. Some current encryption methods, you are warned, are vulnerable to quantum computing, a point returned to later.
Hash Functions Explained 6:34
A cryptographic hash function takes any input and produces a fixed-size output called a hash, and the same input always gives the same hash. Hashes are one-way, meaning you cannot reconstruct the original input from the output, and they show an avalanche effect, so changing even one letter, as in turning 'jumped over' into a typo, produces a completely different, unrecognizable hash. This lets hashes act like signatures for documents or accounting records: you can hash a file, publish the hash, and later compare a candidate copy's hash to confirm nothing was altered, similar to sealing a document with the product of two primes and later verifying authorship by supplying one factor.
Puzzles, Primes, and Difficulty 14:05
Cryptographic puzzles are designed to be hard but controllably so, solvable only by trial and error within a set amount of time and effort, and factoring large primes is the classic example. You are shown two roughly seven-digit primes multiplied together, and told that factoring a number with n digits is believed to take exponential time, while verifying a proposed factorization takes only polynomial time, which is much faster. RSA encryption rests on this unproven assumption, so if anyone found a polynomial-time factoring method, RSA-encrypted messages would become breakable, a vulnerability that also connects to the threat of quantum computing.
Diffie's Public Key Breakthrough 18:00
Before Whit Diffie's design, the same machine that scrambled a message also deciphered it, so capturing an enemy's cipher machine, as in World War II, exposed everything. Diffie's innovation makes the deciphering key public and widely known while keeping only the private decryption key secret, flipping the old secrecy model on its head.
Rings, Groups, and Cyclic Structures 20:03
You are walked through the algebra behind this: a group or ring is a space where elements can be added or multiplied under rules like associativity, commutativity, and identity elements, and a generator is an element that, multiplied by itself repeatedly, produces every element in the group. A worked example uses a group of order 11 with generator 6, showing how repeated multiplication and taking remainders (the modulus) cycles through all the elements, which is why it is called a cyclic ring. The order of the group must be large, because in small groups an attacker can simply brute-force the private key through a method called baby step, giant step.
Alice, Bob, and Message Encryption 29:03
Using this cyclic ring structure, Alice generates a key pair by picking a random number x as her private key and publishing h, computed as the generator raised to x, along with the group itself. When Bob wants to send Alice a secret message, he picks his own random number y, computes a shared secret using Alice's public h, and sends Alice two pieces of ciphertext, one being the generator raised to y and the other the message combined with the secret. Alice, using her private key, can recover the shared secret and then invert it to recover the original message, a process that works only because of special properties of these encrypted spaces where powers combine multiplicatively.
Signatures for Outgoing Messages 38:39
Signing an outgoing message is harder than receiving one, but it lets a receiver verify that Alice truly sent the message, that she cannot deny having sent it, and that the message was not altered afterward. The Schnorr signature scheme is introduced as an example, involving parameter generation with a prime p, a hash function, and a generator, followed by key generation where a random integer a becomes the secret key and the generator raised to a becomes the public key, before moving into the actual signing steps using another random number k.
Verifying a Digital Signature 42:32
The lecture walks through a signature exchange between Alice and Bob using a shared generator g and a private key a. Bob sends Alice a random constant c, Alice combines it with her private key and a value k to produce a signed response, and Bob checks the signature by comparing two expressions built from g, the public key, and the values exchanged earlier. Working through the algebra shows the two sides are always equal when the steps are followed correctly, which means Bob can be confident the message could only have come from Alice, since only she could have produced a response that makes the equation hold.
Polynomials and Quantum Resistance 46:33
Because standard integer-based encryption could eventually be broken by quantum computers, newer schemes use polynomials instead of plain sets of integers. Coefficients are drawn from a finite field, such as F2 or F11, and a modulus polynomial keeps the highest power under control. A related method called ring learning with error adds small random polynomials into the mix, and because this error can compound with repeated operations, modern encryption relies on techniques like bootstrapping to periodically strip out accumulated error. The intuition for why this resists quantum attacks is that different coefficients can be thought of as controlled by different agents, so cracking one does not expose the rest.
Homomorphic Encryption and Secret Sharing 51:02
Homomorphic encryption relies on a mapping between two structures that preserves relationships, so a function can be applied to an encrypted message and still produce the correctly encrypted result once decrypted, without ever exposing the original value. Multi-party computation, also called secret sharing, lets several people combine private numbers, such as personal wealth, by passing along sums clouded with random noise that each person later removes, so that only a combined result, like an average, is revealed. These techniques can be layered together, so a value can be encrypted, operated on, and decrypted without any party ever seeing the raw underlying data.
Zero-Knowledge Proofs and Pedersen Commitments 58:31
A zero-knowledge proof lets a prover convince a verifier that they know something without revealing what it is, illustrated by a special-pen guessing game and a graph-coloring puzzle repeated many times until the verifier is statistically convinced. This idea extends to ledgers between banks like Goldman Sachs, JPMorgan, and Barclays, where a Pedersen commitment obscures each transaction value using two generators, g and h, while still letting a regulator verify that sums balance without seeing individual amounts.
Bitcoin's Merkle Tree and Mining 1:05:34
Returning to Bitcoin, transactions are organized into a Merkle tree where each node is the hash of its children, so changing any transaction changes the final root. Each block header includes a nonce, and miners repeatedly change this nonce to try to produce a double-hashed block ID smaller than a target number, such as two to the 187th power, which happens rarely and requires trial and error, making it the basis of the mining difficulty.
Choosing the Right Encryption Tool 1:09:30
The lecture closes with a decision guide for picking an encryption method. If sensitive data can be revealed after use, a commit-reveal method like Pedersen commitments works. If not, but the situation can be reduced to a true-or-false statement about the data, zero-knowledge proofs apply. If the data still needs full concealment but can be managed by one entity, homomorphic encryption is the tool, and if it must be split across multiple entities, multi-party computation is needed.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
