What is JSON? A Practical Introduction

JSON in one paragraph

JSON is a text format for representing structured data: objects of key–value pairs, ordered arrays, and a handful of value types. It was derived from JavaScript's object syntax (hence the name) but is language-independent — every mainstream language parses and produces it natively. Its entire grammar fits on a business card, which is precisely why it displaced heavier formats: there is almost nothing to learn and almost nothing to get wrong mechanically.

{
  "name": "Ada Lovelace",
  "born": 1815,
  "fields": ["mathematics", "computing"],
  "active": false,
  "advisor": null,
  "address": { "city": "London", "country": "UK" }
}

The six value types

Everything in JSON is one of six types. Strings: double-quoted, with backslash escapes (\n, \", \\, \uXXXX). Numbers: integer or decimal, optional exponent — no leading zeros, no NaN or Infinity. Booleans: true and false, lowercase. Null: the explicit absence of a value. Objects: unordered key–value pairs in braces, keys always double-quoted strings. Arrays: ordered values in brackets.

Composition is unrestricted — arrays of objects containing arrays of objects, as deep as your data demands. That recursive simplicity is the whole specification (RFC 8259).

The rules that break people

Almost every invalid JSON document violates one of five rules, all inherited from JSON being stricter than JavaScript. No trailing commas — [1, 2, 3,] is invalid. Double quotes only — 'single' quotes fail, on keys and values alike. Keys must be quoted — {name: "Ada"} is a JavaScript object literal, not JSON. No comments — neither // nor /* */ exist in the grammar. No undefined, NaN, or Infinity — use null, or strings, or restructure.

If you hand-edit JSON, a validator with exact error positions turns these from twenty-minute hunts into five-second fixes — which is exactly what our JSON Validator does, including a one-click repair for all five mistakes above.

Where JSON is used

REST and GraphQL APIs speak JSON almost exclusively. Configuration lives in it (package.json, tsconfig.json, composer.json). Databases store it natively — PostgreSQL's jsonb, MongoDB's documents. Logs ship as JSON lines for machine parsing. Browsers expose JSON.parse and JSON.stringify as built-ins, and every HTTP client defaults to it. Learning to read JSON fluently is learning to read the modern web's wire format.

JSON vs. the alternatives, briefly

Against XML: JSON is lighter (no closing tags), maps directly to data structures, but lacks attributes, namespaces and schema maturity — XML still owns document-centric and enterprise domains. Against YAML: YAML is more human-friendly for hand-written config (comments! no quote ceremony) but its whitespace sensitivity makes it error-prone; JSON wins for machine interchange. The formats interconvert losslessly in most practical cases — our JSON to YAML converter does it in one click.