Hash functions and SHA-256

Wonder

1. Why does changing one character in a file alter half its hash?
2. If endless possible inputs share a finite set of outputs, why is it still infeasible to find two that collide?
3. What made SHA-256 safe when its predecessor, SHA-1, was broken?

Explore

1. A cryptographic hash function maps arbitrary-length data to a fixed-size digest. It is deterministic: the same input always yields the identical output. It is fast to compute but infeasible to reverse. This one-way property defines preimage resistance. If an attacker holds only the hash, they cannot derive the original message. This irreversibility secures integrity checks and identity proofs across distributed systems.
2. SHA-256 produces a 256-bit output regardless of input size. Whether the input is a single byte or a gigabyte file, the digest remains exactly 32 bytes. This fixed length allows consistent storage and comparison. The function processes data in 512-bit blocks through 64 rounds of logical operations. Its design follows the Merkle–Damgård construction, chaining a compression function across the blocks; the compression function’s mixing is what scrambles the output.
3. The avalanche effect is a critical feature: flipping one bit in the input changes approximately half the bits in the output hash. This sensitivity ensures that even minor data corruption or tampering results in a completely different digest. It makes hash functions ideal for verifying file integrity and detecting alterations. Any modification, no matter how subtle, breaks the match between the expected and computed hashes.
4. Collision resistance is the most demanding property. Finding two distinct inputs that produce the same hash requires roughly 2^128 operations because of the birthday bound. For SHA-256, this computational cost renders brute-force attacks impractical with current hardware. In contrast, SHA-1 suffered from a practical collision attack demonstrated by researchers at Google and CWI in 2017, known as SHAttered. This vulnerability forced migration away from SHA-1 for security-critical applications.
5. Bitcoin applies SHA-256 twice, a practice called double hashing. Block headers and transaction IDs are hashed with SHA-256, then the result is hashed again with SHA-256. Satoshi never explained the choice; a common explanation is protection against length-extension attacks, a known property of Merkle–Damgård hashes such as SHA-256. Every block hash and transaction ID on the chain is the output of this double pass.
6. Open any terminal or online tool and compute the SHA-256 hash of the string ‘MSTR Cat’. Then compute the hash of ‘MSTR cat’, changing only the capitalization of one letter. Compare the two 64-character hexadecimal outputs. They will be entirely different, sharing no recognizable pattern. This immediate divergence illustrates the avalanche effect in action: a single bit change scrambles the entire digest, confirming that the function treats inputs with extreme sensitivity.

Key facts

Exercise nine, six, three, one

Fact check

Dates, figures and claims checked against the published sources listed below.

• NIST, FIPS 180-4: Secure Hash Standard (2015)
• A. M. Antonopoulos & D. A. Harding, Mastering Bitcoin, 3rd ed. (2023)
• D. Stevens et al., 'The first collision for full SHA-1' (2017)