I. Three Apples

Three apples are often made to carry a history.

One belongs to Newton: a falling fruit, a question joining earth and sky. It was not a magical blow that delivered the law of gravitation. Another belongs to a computer company's logo, with a bite that its designer says helps identify the fruit, not commemorate Turing.

The third was in Turing's room after his death. An apple had been partly eaten. Cyanide poisoning was established, but the apple was not tested. We cannot acknowledge that uncertainty and then begin our story by confidently describing a fatal bite.

Legends draw a life into one memorable shape. An apple fits on a commemorative poster. A paper tape, a logical argument, years of collective work, and a criminal prosecution are harder to put in the same frame.

Alan Turing helped explain computation and contributed to wartime codebreaking. His country later convicted him for a sexual relationship with a man and subjected him to hormone treatment. The contrast is appalling, but his rights did not need to be purchased with wartime service.

Had he invented nothing and helped no nation, the persecution would still have been wrong. Unless we remember that, our tribute risks reducing him once again to a certificate of useful genius.

II. The Tape

In 1936 Turing wrote On Computable Numbers, with an Application to the Entscheidungsproblem. Its machine was not a device on a desk. It was a way of making calculation by rules precise enough to investigate.

Imagine a tape that can keep extending, divided into squares bearing symbols. A head reads the current square, writes, and moves left or right. A finite collection of states and rules determines the next action from what is presently being read.

The symbols need not be just zeros and ones; that is a familiar simplified picture. What matters is the local, explicit, repeatable step. The machine does not skip a rule through inspiration. It need not understand what its calculation means to us.

Then comes universality. Encode the rules of another such machine and give that description to a universal machine: it can simulate the other. Doing a different job need not require building a fundamentally different device. Change the instructions.

A theorem and a thesis must be distinguished here. The simulation of these formal machines is a mathematical result. The identification of all effective mechanical calculation with such models is the Church–Turing thesis, not a proof that every process in the universe is a tape machine.

Bach makes elaborate music from limited materials; Newton relates falling bodies to the heavens. Turing reduces calculation to a few actions so that a new question can finally be asked precisely: how far can those actions go?

III. Halting

The decision problem, stated carefully, asks for a mechanical procedure deciding whether any first-order logical formula is universally valid. It belongs to Hilbert's larger ambitions for mathematics, but is not simply a request for a machine to answer every imaginable question.

Turing's 1936 paper gives a negative result; Church independently obtains a related result that year. Today's most accessible route is the halting problem: could one program always decide whether any given program, on a specified input, eventually stops?

Suppose there were an infallible judge. Build a contrary program: if the judge predicts that it stops, it runs forever; if the judge predicts endless running, it stops. Arrange the appropriate self-reference, and either answer given by the judge is overturned.

This is not a shortage of speed or tape. Within the specified computational framework, no halting decider works for every program and input. The argument in Turing's original paper is not identical to the modern textbook presentation; the latter is an explanation, not a transcription.

We can still establish whether many particular programs halt. Restricted classes of problems can have reliable algorithms. No universal decider does not mean nothing is computable, and it does not prove that human intuition must always outperform machines.

The boundary matters because it is exact. Turning it into an all-purpose slogan about life, society, or the universe would erase the precision that made the achievement possible.

IV. War

In 1939 Turing joined the wartime work at Bletchley Park. Enigma resembled a lock with an overwhelming number of possible arrangements. The task was not to admire that complexity but to discover which apparent possibilities could not actually occur.

A letter could not encrypt as itself. A message might contain a predictable phrase. Connections had to remain consistent. These constraints could work together: the Bombe rejected large numbers of impossible settings rather than magically understanding every German message.

It was not a war won by a solitary genius. Polish mathematicians' prewar breakthroughs and their sharing of results were indispensable. Turing supplied crucial British Bombe design ideas, Gordon Welchman made a major improvement, engineers built the equipment, and operators kept it working.

Intercepts, linguistic analysis, captured material, naval operations, and the people who turned intelligence into decisions also mattered. Machines depended on people. People's tasks depended on one another, across stretches of repetitive work that a heroic montage easily leaves out.

The contribution was enormous. Claims that it shortened the war by a precise number of years and saved a precise number of lives are counterfactual estimates, not totals anyone can count directly. Recognition does not require pretending otherwise.

After the war Turing worked on computing at the National Physical Laboratory and later at Manchester. He did not simply return to Cambridge from a secret estate and wait for tragedy. There were designs, disagreements, programs, and questions still unfinished.

V. The Imitation Game

His 1950 paper Computing Machinery and Intelligence begins with whether machines can think. That question can imprison a discussion inside definitions: define thinking, argue over what truly qualifies, and then return to the definition.

Turing changes the approach. Through written exchanges, an interrogator tries to identify interlocutors; the question becomes how a machine might perform in such an exchange. The paper starts with an imitation game between people before introducing a machine. It does not merely proclaim that deception proves a soul.

The sharp move is to ask us to explain our standards. Why does a colleague's response count as understanding while a machine's counts as mere mechanism? If the evidence in conversation is comparable, are we rejecting one participant in advance because we know its material?

Passing a conversational test, however, does not establish consciousness or automatically confer moral status. Duration, conditions, questions, and the range of performance affect what can be concluded. Indistinguishability in one exchange does not establish identity in every respect.

Hume offers a useful comparison: we often move from something observable to a judgment larger than the observation. But Hume did not abolish all causal knowledge, and Turing did not declare thinking a fraud. A comparison is a footbridge, not proof that two philosophies are identical.

What did you observe? What did you infer? What remains untested? Turing did not end the argument for future generations. He moved parts of it into a space where people could inspect one another's claims.

VI. Turing and Gödel

Gödel published the incompleteness theorems in 1931; Turing's paper followed in 1936. They did not work in the same year or prove exactly the same statement twice.

For consistent, effectively axiomatized theories strong enough to express adequate arithmetic, Gödelian results disclose limits of proof. Turing's undecidability concerns whether general algorithms can solve specified classes of problems. The connections are profound, but the assumptions, objects, and conclusions must not be mixed.

For orientation, one might say that Gödel shows why certain proof rules cannot exhaust arithmetic, while Turing shows why no general program settles every halting question. That comparison helps a reader get started; it is not a formal definition.

Nor should Gödel become a Platonist who merely gazes beyond a boundary, while Turing becomes an engineer who merely builds. Gödel's proofs require intricate construction. Turing's machine begins as abstract mathematics. Both make limitations visible through creative work.

A limitation is not a command to cease research. Without a universal answer, we can still devise useful programs, prove theorems, and learn which methods fit which problems. The distinction between impossible in general and possible here is itself productive.

Knowing a tool cannot do everything is far from believing it can do nothing. Turing did not settle beside his negative result. He kept designing things that could work and that other people could use.

VII. With Socrates

Place Turing beside Socrates and a related question becomes audible: you say you know, but how are you making that judgment?

Socrates tests a definition against people and cases, looking for the point where it fails. Turing sets up a conversational experiment in which apparently obvious exclusions of machine intelligence have to face examination.

These are not the same method. Socratic ethical questioning is not a statistical test. The imitation game is not designed merely to leave everybody knowing nothing. Turing also proposes constructive research possibilities, including machines that learn.

What they share is resistance to letting a familiar word complete the thinking. Calling something only a machine requires explaining what only excludes. Calling something a person must not become merely the recognition of a performance profile.

Here Turing's own life makes the issue hurt. A society was willing to recognise value in the work he could perform, while using the identity of his sexual partner to restrict his life. It understood the usefulness of codebreaking and failed to respect the codebreaker's private existence.

His dignity is not a consequence of the halting theorem. No mathematical result must first vouch for him. A person cannot be reduced to a list of contributions: that is where our ethical reading of the history should begin.

VIII. The Conviction

In 1952 Turing reported a burglary, and the investigation brought his sexual relationship with Arnold Murray to police attention. At his conviction in March he was thirty-nine, not forty-one. The law then criminalised certain consensual sexual acts between men.

The charge of gross indecency had also been used against Oscar Wilde. Turing accepted probation conditional on hormone treatment rather than imprisonment. Calling this a choice requires adding that both options were imposed by the state, not offered to a freely consenting patient.

The treatment, commonly called chemical castration, injured his body and violated his autonomy. He lost access to secret cryptographic work but retained his Manchester research position. His remaining years should not be written as a period without friends, activity, or anything but approaching death.

Chinese readers may recall Sima Qian enduring punishment while trying to continue his historical writing. The circumstances, penalties, and people differ. We cannot make research Turing's sole proven motive, still less celebrate the noble choice extracted by a coercive institution.

His work on morphogenesis had begun before the conviction. An important paper appeared in 1952, and the investigation continued afterward. How could reaction and diffusion produce patterns? How did living shapes develop? These questions were not consolations invented only after punishment.

He continued working, meeting people, and joking. That does not diminish the persecution. It warns us against writing an unverified psychological diagnosis for the dead. A harmed person remains larger than the record of what was done to him.

IX. That Apple

Turing was found dead at home on 8 June 1954; his death is dated 7 June. He was forty-one. Cyanide poisoning was confirmed and the inquest returned a verdict of suicide. His mother believed an experiment had caused an accident. The untested apple cannot supply a verified reconstruction of the act.

Acknowledging this does not erase the official verdict or excuse the persecution. It refuses to fill gaps in evidence with a perfectly composed tragic shot. What he finally thought cannot be settled by what our narrative needs.

The British government apologised in 2009. A posthumous royal pardon followed in 2013. In 2021 his portrait appeared on the fifty-pound note. Public remembrance had changed direction, but none of these acts returned the years he lost.

Redress should not be reserved for the most famous man. Those who never broke a code or appeared on a banknote should not have been punished for consensual intimacy either.

At the imaginary bridge of this series, Turing stands among the others. Hume puts away his backgammon; Kierkegaard turns; Schopenhauer's dog approaches. We will not make Turing bite a poisoned apple before an audience. It is neither established history nor the price of understanding him.

Let him hold a sheet covered in symbols instead. There is something he has not finished saying, and somebody makes room. The invented scene cannot repair history, but at least it does not turn him once more into nothing but a symbol of death.

He was not only a national asset finally recognised, or a fable in which genius must be miserable. He was someone who should have been able to keep working, loving, and growing older.

We do not know whether the apple contained cyanide. We know that he is gone. Remembering the difference is part of remembering him honestly.

Notes

The account distinguishes universality, the Church–Turing thesis, halting, and first-order validity. The diagonal argument uses a modern explanatory presentation, not a transcription of the 1936 paper. The Gödel comparison neither equates different theorems nor derives ethics from mathematics.

Codebreaking was collective work. Turing was thirty-nine at conviction, and morphogenesis preceded it. Cause of death, inquest verdict, and the untested apple are distinct evidential questions. The bridge is an invented scene, not a reconstruction of his death.

• Turing: On Computable Numbers, archive B/12
• Turing: Computing Machinery and Intelligence (1950)
• National Museum of Computing: Turing–Welchman Bombe
• Andrew Hodges: Turing, 1952–1954, with archival links
• Interview with Apple logo designer Rob Janoff
• UK government: posthumous pardon, 2013
• Bank of England: Turing banknote, 2021