The Mechanical Turk
A chess-playing automaton that toured Europe and America for eighty-five years, beat Napoleon and Benjamin Franklin, and provoked the first serious public argument about whether a machine could think. There was a man inside the whole time.
Why this verdict
Debunked conclusively in 1857, when the son of the machine's last owner published a full account of the mechanism in *The Chess Monthly*: a sliding seat and a hinged interior let a concealed human player move with the cabinet doors open, one section at a time.
- Period
- Region
- Reading time
- Sources
01The claim
That the automaton built by Wolfgang von Kempelen and presented to Empress Maria Theresa in 1770 — a robed figure seated at a chest of clockwork, which played chess at a strong amateur level against all comers — was a machine, and that the cabinet's interior, displayed to audiences before each exhibition, contained only mechanism.
02Where it came from
Kempelen, a Habsburg court official, built the Turk after seeing a conjuring performance and claiming he could do better. It toured Europe, then passed after his death to Johann Nepomuk Maelzel, who took it to Britain and, from 1826, to the United States. Along the way it played Benjamin Franklin in Paris and Napoleon at Schönbrunn. Strong players were hired as the hidden operators, among them Johann Allgaier and William Schlumberger — which is why the machine played well.
03What the record shows
✕ DEBUNKED — Investigated and refuted. Here is what actually happened.
The trick was in the architecture, not in any single concealed compartment. The visible clockwork occupied only part of the cabinet; a sliding seat let the operator move behind whichever section was open for inspection, and a pantograph linkage let him move the Turk's arm. A candle's smoke vented through the turban.
The exposure was gradual. An anonymous London pamphlet of 1819 and then Robert Willis's analysis of 1821 worked out the essentials of the concealment. Edgar Allan Poe's 1836 essay 'Maelzel's Chess-Player' reasoned to the right conclusion from the wrong premises — he argued that a true machine would necessarily win every game, which is false, but observed correctly that the Turk's play varied in strength the way a person's does, and that a particular attendant was never visible at the same time as the machine's operation.
The definitive account came after the Turk burned in a Philadelphia fire in 1854. Silas Weir Mitchell, whose father had bought the machine when Maelzel died, published the full description of the mechanism in The Chess Monthly in 1857. There has been no dispute since; a working replica built in the 1980s reproduced the method.
04What it cost
The Turk's importance is not that it fooled people but what it made them argue about. For eighty-five years, educated audiences on two continents debated in public whether a machine could perform an act of judgement — and the debate was serious, because the alternative to a thinking machine was a deception nobody could locate. Charles Babbage played the Turk in 1819, lost, and recorded the encounter; his own calculating engines followed within a few years. The name was revived by Amazon in 2005 for a service in which human beings perform tasks that appear to a user to be done by software, which is either a very good joke or an admission.