Base32 Crockford Validator

Crockford's Base32: human-friendly encoding behind the ULID

Crockford's Base32, designed by Douglas Crockford, is a Base32 variant optimised for humans reading and typing identifiers. It uses 0–9 and the letters A–Z but deliberately excludes I, L, O and U — the first three to avoid confusion with 1 and 0, and U to avoid accidental profanity. This tool validates the character set (case-insensitive, hyphens ignored).

The example above, 01ARZ3NDEKTSV4RRFFQ69G5FAV, is a ULID — a 26-character, lexicographically sortable identifier whose encoding is Crockford Base32. That makes this alphabet the backbone of modern sortable IDs.

What makes it forgiving

• Case-insensitive: decoders accept upper or lower case and normalise to uppercase.
• Confusable substitutions on decode: O → 0, and I / L → 1, so a human typo still decodes correctly.
• Hyphens are ignored: you may insert - for readability and they're stripped before decoding.
• Optional check symbol: Crockford defines a mod-37 check character (*, ~, $, =, U) that some implementations append.

Where it shows up

• ULID: 48-bit timestamp + 80-bit randomness, encoded as 26 Crockford chars — sortable and URL-safe.
• Public IDs & coupons: short codes meant to be read aloud, dictated over the phone or printed on receipts.
• Key shards & licences: where transcription errors must be tolerated.
• Mock data / testing: checking a token is well-formed Crockford before parsing.

Gotchas

• Not RFC 4648: Crockford is a separate spec; don't decode it with a standard Base32 or Base32hex table.
• Exclusions are the point: seeing I, L, O or U in canonical output means it isn't canonical Crockford.
• Check symbol ≠ data: if a trailing check char is present, don't feed it into the value decode.
• ULID monotonicity: two ULIDs in the same millisecond rely on the random part; sortability is millisecond-granular, not strictly unique-ordered.

FAQ

Why exclude I, L, O, U? I/L look like 1, O looks like 0, and U is dropped to avoid spelling rude words — all to make hand-typed codes reliable.

Is a ULID just Crockford Base32? A ULID is a 128-bit value whose text form is 26 Crockford Base32 characters; the encoding is Crockford, the structure (time + random) is ULID's own.

Does it preserve sort order like Base32hex? Yes for the canonical alphabet — which is why ULIDs sort chronologically as plain strings.