What Is a Hash? (And Why MD5 Had Its Day)

September 29, 2026 · 6 min read

Download a Linux ISO and the site lists a long hex string next to it — sha256: 9f86d081884c7d659... — with instructions to "verify the checksum." Git commit IDs are hashes. When you log in somewhere, the server compares hashes, not your actual password. Hashing is one of the most-used and least-understood ideas in computing. The core concept, though, is simple: a hash function takes any input and produces a fixed-size fingerprint of it.

The three promises of a hash function

A cryptographic hash like SHA-256 makes three guarantees:

Deterministic. Same input, same output, every time. hello always hashes to 2cf24dba5fb0a30e26e83b2ac5b9e29e1b161e5c1fa7425e73043362938b9824. No randomness, no exceptions.
One-way. Given the hash, you can't reconstruct the input. There's no "unhash" button. The only way to find an input matching a hash is to guess inputs until one matches — and with 2²⁵⁶ possibilities, you'll be guessing for longer than the universe has existed.
Avalanche. Change one character of the input and the output changes completely and unpredictably. hello and hellp produce hashes with nothing in common. There's no "close" — either every bit matches or it doesn't.

That combination — deterministic, irreversible, chaotic — is what makes hashes useful as fingerprints. Two files with the same SHA-256 hash are, for all practical purposes, the same file.

What people actually use hashes for

Verifying downloads. That ISO checksum: hash the file you downloaded, compare with the published hash. Match means the download wasn't corrupted (or tampered with). One flipped bit anywhere in a 4 GB file changes the hash entirely.
Storing passwords. Good systems never store your password — they store hash(password + salt). When you log in, they hash what you typed and compare. A database breach then leaks hashes, not passwords. (Salting — adding a unique random value per user before hashing — defeats precomputed "rainbow table" attacks. And for passwords specifically, slow hashes like bcrypt or Argon2 are preferred over SHA-256, precisely because SHA-256 is fast and attackers love fast.)
Git. Every commit ID is a hash of the commit's contents. Change history and the IDs change — that's how Git detects tampering and how content-addressing works.
Deduplication. Cloud storage services hash file chunks; identical hashes mean identical data, stored once.
Data structures. Hash tables (the thing behind Python dicts and JS objects) use non-cryptographic hashes for speed — same idea, weaker guarantees.

So what happened to MD5?

MD5 was the hash function of the 90s and 2000s. Then researchers found collisions — two different inputs producing the same MD5 hash. At first it took serious computing power; by the late 2000s, researchers were generating colliding files on a laptop, and someone famously created a rogue CA certificate using an MD5 collision. A hash function where attackers can craft collisions is broken for security purposes. Full stop.

Does that mean MD5 is useless? Not quite. For non-security uses — checksums where nobody is attacking you, hash tables, quick file identification — MD5 is fast and fine. The rule is simple: if an adversary could benefit from faking a hash, don't use MD5 (or SHA-1, which fell the same way). For anything security-related, SHA-256 is the current sane default, with SHA-3 and BLAKE3 as modern alternatives.

One more misconception to kill

Hashing is not encryption. Encrypted data can be decrypted with the key; hashed data can't be "decrypted" at all. When someone says "the passwords are encrypted with MD5," they're confused twice over — it's hashing, not encryption, and MD5 shouldn't be near passwords anyway. Words matter here because they imply very different security properties.

Try it: hash any text instantly with our free SHA-256 Hash Generator — change one character and watch the avalanche effect yourself. All in your browser, nothing uploaded.