Maximum distance separable code

A maximum distance separable (MDS) code is a linear code that achieves the Singleton bound with equality — meaning it is the largest possible code for its given length and minimum distance. For a code of length n and dimension k over an alphabet of size q, an MDS code has minimum distance exactly n - k + 1, the maximum permitted by the bound. Reed-Solomon codes are the canonical example, constructed using polynomial evaluation over finite fields and widely used in digital communications, storage systems, and cryptography. The MDS conjecture states that for most alphabet sizes and lengths, Reed-Solomon codes are essentially the only MDS codes — a claim that, if proven, would mean the Singleton bound is only achievable through algebraic geometry. The conjecture remains open for many parameter ranges, and its resolution would settle one of the central classification problems in coding theory.

_MDS codes are often treated as optimal solutions to the error-correction problem, but this optimality is bought with a hidden cost: they require large alphabet sizes and are notoriously fragile to burst errors. The Singleton bound is not a recommendation; it is a limit. The fact that most real communication systems do not use MDS codes — preferring LDPC, turbo, and convolutional codes instead — is not an engineering failure. It is evidence that the problem of reliable communication is richer than the trade-off between rate and distance that the Singleton bound captures._