Hakoiri Musume / Klotski (1930s) — The Lineage of Blocks That Trapped Cao Cao

Introduction

This is an unusual puzzle: it has no single inventor whose name we can settle on. In the West it is called Klotski; in China, Huarong Dao; in Japan, Hakoiri Musume — the board is the same in every case. Inside a box roughly four by five cells, one large piece sits packed among several smaller ones, and the goal is to slide only the large piece out through a narrow gap in the bottom edge. The rule is simple, yet finding the sequence that clears the smaller pieces out of the way can take close to a hundred moves.

Dig into its lineage, though, and it refuses to resolve into a single origin. A sliding-block puzzle patented in 1890s America, a board Chinese children were cutting out of paper in the 1930s, and a toy sold in a wooden box in Japan around the same time — these converge on nearly identical rules with no trace that any of them referenced the others. What interests me as a historian is not settling on one invention story, but tracing why the same structure arose, as if by agreement, on three continents at once.

Surrounded by smaller pieces with only a narrow gap as an exit — the basic form of Klotski (illustration, AI-generated)

The Context of Its Era

The earliest footprint sits in the US patent office. In 1893, Henry Walton secured US Patent 516,035 for a sliding puzzle combining blocks of unequal size, and in 1900 Frank E. Moss obtained a similar patent, No. 668,386. Lewis W. Hardy's Pennant Puzzle, filed in 1907 and copyrighted in Chicago in 1909, has since been described by later researchers as the closest known direct ancestor of Klotski in design. In this era of American culture, toys and parlour puzzles were routinely secured through the paperwork of patents and copyrights, which is exactly why the inventors' names survive in the record at all.

In China, meanwhile, a different version was growing on paper. Its pieces were cast as characters from Romance of the Three Kingdoms, layered onto the story of Cao Cao fleeing along the narrow escape route called Huarong Dao after his defeat at the Battle of Red Cliffs (208–209 CE). Researcher Lin Dekuan documented children in Shaanxi province playing this paper version in 1938, and a 1949 book by Jiang Changying suggests the game 'was invented in the late 1930s and became popular from the late 1930s into the early 1940s.' In 1932, on top of that, the Englishman John Harold Fleming obtained a UK patent for this very 'Cao Cao's escape' board design — showing that this single board carried both the face of an Eastern folk toy and a Western patented commodity at once.

In Japan, the same structure spread under the name Hakoiri Musume ('the sheltered daughter'), reportedly from the Showa 10s (around 1935) onward. The large piece is cast as a daughter to be freed, rescued out through the bottom gap past the smaller pieces blocking her way — the opposite of Cao Cao's defeat, dressed instead in the language of a happy occasion like a wedding send-off. Patent paperwork, a folk paper craft, a wooden toy on a shopkeeper's shelf: three cultures arrived at the identical combinatorial structure in this era, differing only in their distribution channels and the story-costume draped over it. In fact, according to a historical piece by the Chinese outlet GCORES, the very claim — settled in China from the 1980s onward — that 'Huarong Dao is an ancient traditional Chinese game' is itself a retroactive label, one that obscures this early-20th-century case of parallel, unconnected invention.

A patent drawing, a paper cutout, a wooden box — three origins that never referenced each other (illustration, AI-generated)

Mechanics

The standard board is a 4-by-5 box holding one 2x2 large piece, four or five 1x2 rectangular pieces, and four to six 1x1 small pieces. The goal is to push only the large piece out through a gap in the middle of the bottom edge. Pieces can only slide up, down, left, or right, and the board's edges and the other pieces constrain which direction is open at any moment — a single simple physical rule generates a maze of staggering combinatorial depth.

In February 1964, Martin Gardner, writing in Scientific American under the title 'The hypnotic fascination of sliding-block puzzles,' introduced the shortest solution to this board — 81 moves — discovered by mathematician Thomas B. Lemann. This was not merely the publication of a solving technique. It was the moment a wooden toy's block-shuffling was made public as, precisely, a rigorous search problem: finding the shortest path across a finite graph where each configuration is a node and each legal move is an edge. That reframing anticipated, by nearly half a century, the state-space thinking that now underlies an entire style of grid-based puzzle design built around pushing and sliding pieces as search-space verbs.

Each configuration a node, each move an edge — the moment a wooden toy was revealed as a search problem (illustration, AI-generated)

Lineage to the Present

The digital trail goes back to 1988, to an early graphical computer version made by Jim Bates. But what carried it to the world at an entirely different scale was the version bundled in Microsoft's Windows Entertainment Pack in 1991 — coincidentally, the very same distribution channel and the same era that, as I wrote previously, carried Minesweeper to the world in 1990. It was in this 1991 Windows 3.x release, developed by ZH Computer, that the English name 'Klotski' — derived from the Polish 'klocki,' meaning wooden blocks — is said to have first taken hold, replacing earlier trade names in circulation such as Psychoteaze, Square Root, Intreeg, and Ego Buster. From there it branched into the GNOME desktop environment and countless freeware ports.

That lineage reaches all the way to today's Steam. Blockee - Sliding Puzzle, released in 2021, states directly in its own Steam store description that it is 'a sliding block puzzle inspired by the old Klotski puzzle.' This is a rare case where the lineage is confirmed in the developer's own words rather than through a critic's retroactive genealogy. There is also a related, if not directly descended, physical puzzle worth noting: Rush Hour, conceived in the 1970s by Japanese puzzle designer Nob Yoshigahara and first sold in the US in 1996 by Binary Arts (now ThinkFun). The English Wikipedia classifies it in 'the same family of sliding-block puzzles' as Klotski, even though the two were not created hand-in-hand — a documented case of two lines arriving independently at the same underlying skeleton.

From a wooden box, to a floppy disk, to today's screen — the same skeleton persisting through changing forms (illustration, AI-generated)

References

Sources referenced in this article:

・Wikipedia: Klotski

・Wikipedia: Hakoiri Musume (puzzle, JP)

・Wikipedia: Rush Hour (puzzle)

・Steam: Blockee - Sliding Puzzle

・GCORES: "Let's Talk Puzzle: Huarong Dao and Other Sliding-Block Games" (ZH)

Closing

What this puzzle demonstrates, historically, is that a strong idea is not a single event but a phenomenon that recurs whenever the conditions align. In three places with no evidence any referenced the others — a filing in the US patent office, a scrap of paper in a Shaanxi child's hands, a wooden box on a Japanese toy-shop shelf — the identical skeleton of 'rescue one large piece from a maze of smaller ones' assembled itself independently. The story of Cao Cao's defeat and the story of a daughter's wedding send-off are both costumes applied after the fact. The skeleton itself has not changed shape once, from the 1893 patent drawing to a 2021 Steam store page. What draws me, as a historian, is this quality of being a structure that keeps getting rediscovered — and the suspicion that our very desire to pin one inventor's name to it is what has kept the puzzle's real history obscured.

One last piece resting just short of the exit — the costume changes, the skeleton does not (illustration, AI-generated)