Thinking Is Like a Game, It Does Not Begin Unless There Is an Opposite Team

In March 2016 the Go program AlphaGo played its second game against the champion Lee Sedol and made a move that no human expert would have played. Commentators first took it for a mistake. It was move thirty seven, a placement so unconventional that Lee left the room for several minutes before he could answer. In the following games the human player produced a brilliant move of his own, one that analysts called unforeseeable, which only a machine's challenge had made possible. The episode illustrates the claim in the topic, that thinking is like a game and does not begin until there is an opposite team.

The claim has a long ancestry. Socrates did his philosophy in the Athenian marketplace by cross examining anyone who claimed to know something, and his method, the elenchus, proceeded by drawing out consequences until the other person's belief collapsed or was refined. In the Indian tradition the Nyaya school set out the rules of debate, and the practice of shastrartha made public contest the way truth was settled. According to tradition the young Adi Shankara travelled the country defeating rival teachers, including the ritualist Mandana Mishra, in debate.

John Stuart Mill put the argument in modern terms. In On Liberty in 1859 he wrote that he who knows only his own side of the case knows little of that. Even a true opinion, if it is never challenged, is held as a prejudice, without understanding its grounds. The opposing team supplies what the home team cannot supply for itself, namely the strongest objection, the awkward example and the view from outside. Mill wanted dissenters protected not only for their sake but for ours, since without them our beliefs go slack.

Psychology has begun to agree. Hugo Mercier and Dan Sperber argued in 2011 that human reasoning did not evolve to find truth alone, but to persuade others and to evaluate their arguments. This explains a puzzle. People reasoning alone are poor at spotting flaws in their own views, yet in a group where members disagree and must justify themselves, accuracy improves sharply. Reasoning behaves like a sport that is played well only against opposition.

Reasoning behaves like a sport that is played well only against opposition.

Science has built the rivalry into its machinery. Peer review, replication and the culture of challenge at conferences are formal ways of supplying the opposite team. Daniel Kahneman, who often disagreed with the psychologist Gary Klein about the value of intuition, arranged a joint paper with him in 2009 to find where each was right, which they called an adversarial collaboration. The paper stated their agreements more clearly than either would have alone. Competitive pressure on rival laboratories has also driven discovery, as when two groups raced each other to decode the structure of DNA in the early 1950s.

Chess offers a neat illustration. After losing to the computer Deep Blue in 1997, Garry Kasparov argued that the best results come from a human and a machine playing together, and a few years later amateur teams with good procedures beat grandmasters and the strongest programs. The opponent had changed the way the game was played, and the loser became a better thinker for it. Games, like arguments, have rules that let the contest proceed without destroying the players, and this is the condition that makes the contest productive.

Law courts show the structure plainly. The adversarial system in common law countries assumes that truth is likelier to emerge when two advocates, each bound to present the strongest case for a side, argue before an impartial judge than when a single investigator decides alone. The system has well known faults, including inequality of resources, but its premise is the one in the topic, that a contest disciplines inquiry.

But the metaphor has limits. Solitary thinkers have produced some of the greatest ideas. Descartes shut himself in a heated room in Germany in the winter of 1619 and began the reflections that led to modern philosophy. Ramanujan filled notebooks with results in a small house in Madras before anyone read them. Yet even here, the opposite team may have been present in the mind. Descartes imagined an evil demon deceiving him so that he could test his certainties, and Ramanujan was a lifelong student of the problems that mathematicians before him had left unsolved. The solitary thinker often plays against an imagined opponent.

A second limit is that contest can end thinking as readily as it begins it. Cass Sunstein has shown that groups of like minded people who debate among themselves tend to grow more extreme, because each member hears only his own team's arguments. A game in which the supporters shout at each other from separate stands is not thinking at all. Television debates in which each side repeats its slogans, and social media quarrels in which opponents are mocked and not answered, offer rivalry without inquiry. For the contest to produce thought, the teams must be playing the same game and must want to find the answer.

The practical lesson is that anyone who wishes to think well should seek opposition deliberately. A good teacher assigns the student the side he disagrees with, a judge hears both parties before ruling, and a wise leader keeps at least one honest dissenter close. Some institutions formalise the role, as the Catholic Church did with its devil's advocate, who was appointed to argue against the canonisation of a candidate. The device recognises that the mind, left to itself, tends to agree with itself.

It remains unsettled whether the opposing team must be real. If thinking can begin against an imagined opponent, then the game may be played in solitude, and the opponent is only a part of the thinker that he has learned to treat as a stranger. If the imagined opponent is never as strong as a real one, then the solitary thinker has only been practising. The topic asks us to decide whether the other side of the board is a person or a habit, and those who think hardest rarely have time to look up and check.