Mystery Archive / Exhibits & mechanisms
The Chess Automaton: Who Moved the Pieces?
Around 1770, a figure that moved chess pieces attracted attention. Does a visible move show that a machine chose its own next move?
First, what was this chess-playing device?
Around 1770, inventor Wolfgang von Kempelen presented a figure that sat at a chessboard and moved its pieces. Known as the Mechanical Turk, the device appeared to play chess by mechanical means.
Chess requires a player to respond to an opponent and choose a move. A moving arm could therefore suggest more than a physical action. The historical apparatus, however, had a concealed human operator.
We will separate the visible action from the control behind it. Our linked activity uses an original box-and-arm model in which different control paths produce the same small movement; it is not a reconstruction of the Turk.
We can see a piece move one square. Can that same view tell us who decided on the movement, or how?
The automaton sat at a chessboard
Kempelen’s apparatus combined a cabinet, a chessboard and a human-shaped figure. The Computer History Museum dates its presentation to 1770. We begin the story with the approximate date of around 1770.
The moving arm was part of what spectators could see. The force moving it, the means of conveying the position of pieces, and the agent choosing the next move each require separate explanation.
The device appeared in more than one demonstration
The apparatus was exhibited in different places. An 1834 Philadelphia playbill in the museum’s online exhibit is one record of its presentation. It helps document how the attraction was advertised and shown.
An advertisement records a presentation; it does not independently establish every claimed capability. A record of the device being exhibited as a chess player does not by itself show a wholly autonomous interior.
Matching movements, different controls
A person directed the mechanical arm
The Deutsches Museum describes a person inside the apparatus controlling its mechanism. The HNF’s functional reconstruction likewise uses a mechanism to transmit an internal player’s movements to the arm. Human decisions and mechanical movement worked together.
This need not become the opposite claim that the apparatus did nothing. There was a mechanism for moving pieces, and a person directing it. Identifying the role of each gives a more useful explanation.
An open door showed a limited view
The Deutsches Museum describes demonstrations in which doors of the cabinet were opened to show mechanical parts. Displaying some of the interior was itself part of the presentation.
A space visible through one door at one moment is different from a complete account of control throughout a performance. That distinction does not tell us what every spectator could see or believe. It helps us keep the limits of a record in view.
- A chess-playing apparatus appears
Kempelen’s device appears to play chess. The Computer History Museum gives 1770 for its presentation.
- An exhibition playbill
A Philadelphia advertisement for the Automaton Chess Player survives in the museum’s online exhibit.
- The original is destroyed
The apparatus is destroyed by fire. Later reconstructions and explanatory models must be distinguished from the original.
A modern reconstruction is not the original apparatus
The original device was destroyed by fire in 1854. In describing its own reconstruction, the HNF explains that research and experimentation were needed because detailed construction plans and precise descriptions were unavailable.
We should therefore avoid treating a familiar historical cutaway or a functioning modern model as the original design. The simple arm, pawn and screen on this page are our own examples. They do not reconstruct the original seating or linkages.
Different controls can produce the same one-square move
In the activity, first watch an arm on a screened box move a pawn. Opening the screen reveals a path from the model’s human-operated button to the arm. This adds information about control to the visible action.
The second step lets a human button and a preset signal trigger identical movements. Neither chooses a chess move or plays a game. The comparison shows why matching outputs need not identify matching controls.
Ready to inspect the control behind the move?
Watch an action, then open the screen. Compare two controls that can produce the same one-square pawn movement.
Explore the box modelSources and scope
We consulted the Computer History Museum’s introduction and playbill record, the HNF’s account of its own reconstruction, and a Deutsches Museum curator’s article. These four records come from three museum institutions; the playbill is held by the Library Company of Philadelphia.
- 01Mastering the Game: A History of Computer Chess Computer History Museum · Checked 2026-10-22
- 02Playbill for the Automaton Chess Player Computer History Museum · Library Company of Philadelphia 제공 자료 · Checked 2026-10-22
- 03The reconstruction of the HNF’s Chess Turk Heinz Nixdorf MuseumsForum · Checked 2026-10-22
- 04Schach, die Mensch-Maschine und die Anfänge der KI Deutsches Museum · Carola Dahlke, 정보학 큐레이터 · Checked 2026-10-22
The wording and diagrams are original. We reproduce no historical illustration, playbill or reconstruction photograph. The museum accounts of an internal human operator are separate from our invented external-button and preset-signal model.