Arthur Whitney's one liner sudoku solver (2011)

Original Article ↗ Hacker News Discussion ↗

Sudoku: Meditation vs Automation

  • Some treat sudoku as a calm, guaranteed-win pastime; optimizing solvers feels contrary to that.
  • Others find the optimization itself meditative: exploring algorithms, heuristics, and simulation.
  • Solvers are valued to verify puzzle solvability, especially for non-commercial sources where errors are more common (there’s disagreement whether commercial or hobbyist sources are worse).
  • In teaching (e.g., C++ courses) sudoku is praised as a self‑contained, OS‑agnostic project that exercises data structures and algorithms.

Whitney’s K One‑Liner & Algorithm

  • The K solver is identified as breadth‑first search over puzzle states, not heuristic backtracking.
  • It iterates over zero cells and systematically branches possible fillings.
  • Some compare it to short Python/Numpy implementations, noting that high-level array operations can keep other languages concise, though typically not as terse as K.

Array Languages: Power, Style, and Cognition

  • Advocates claim array languages (K, APL, J, BQN, Uiua, Q) enable compact, expressive reasoning over arrays and can feel like a “superpower” for interactive data work.
  • Critics describe the glyph-heavy syntax as “line noise,” hard to remember, and hostile to newcomers; others counter that the core is small and learnable.
  • Debate over whether languages like Julia + Numpy can match array languages semantically, with disagreement about primitives, tables, IPC, and ergonomics.
  • Discussion of first‑class functions and higher‑rank arrays in newer array languages; some see this as a major semantic advance.

Readability, Maintainability, and Metrics

  • Many stress that one-liner golf is a fun stunt, not production style. Real code often uses multiple lines, comments, and more imperative structure.
  • Maintainability concerns: dense expressions are powerful but can be “all or nothing” unless you know the idioms; some report hitting a learning “wall.”
  • Others argue that once fluent, reading dense array code is no harder than reading many lines of traditional code.
  • LoC is widely seen as a bad complexity metric; alternatives like AST node counts or information content (bits) are discussed, but none are viewed as fully satisfactory.

Why People Play or Solve Sudoku

  • Motivations include mental exercise, quick distraction (e.g., on planes), family activity, and the satisfaction of small, repeated wins.
  • Variant sudokus with additional constraints are praised for elegant logical paths, though some find them overly convoluted.
  • Writing solvers is framed as a separate, enjoyable puzzle that teaches general techniques (constraint solving, search, backtracking).