QR Data Capacity

Error correction works by encoding extra redundant codewords alongside your actual data, and those redundant codewords take up space within the same fixed symbol size. A higher EC level (like H, roughly 30% redundancy) reserves more of that fixed space for error-recovery information, leaving less room for the payload than a lower EC level (like L, roughly 7% redundancy) at the identical QR version.

At the same error correction level, version 10 holds many times more data than version 1, since each higher version adds both more modules per side and, correspondingly, far more total encodable area (module count scales roughly with the square of version, not linearly) -- a 57x57 version-10 symbol has a much larger grid than a 21x21 version-1 symbol.

A higher version increases raw capacity by growing the physical symbol, while a higher error correction level spends part of that same capacity on damage resilience rather than payload. For data that must survive scratches, low print quality, or partial obstruction (like a logo overlay), a higher EC level is worth the reduced capacity; for clean, controlled printing conditions, a lower EC level lets you fit more data into a smaller symbol.

Not exactly -- module count grows smoothly and predictably (17 + 4 x version, then squared for the total grid), while data capacity growth reflects the QR specification's codeword allocation tables, which include overhead for format information and finder patterns that doesn't scale in perfect proportion to version.