Writer’s Note
This essay began with correspondence chess and with Christer Sundqvist, whom I came to know through the games we played and through his thoughtful essay, The Last Human Move: Confessions of a Correspondence Chess Player in the Engine Era.
He asks whether we still see traces of humanity in our opponents’ games and, more fundamentally, why we continue to play.
Those questions stayed with me.
For me, chess has always been something rather simple: a table, a chessboard, dimmed lights and two glasses of wine. There will hopefully be some good moves and some wins, but there will also be mistakes, failures and surprises. That imperfection is part of the attraction.
I do not believe chess should return to an earlier age. Chess evolved because society evolved, and we are all part of that development. Artificial intelligence is changing much of the world around us. There is no reason chess should somehow remain outside it.
The more interesting question is how we remain human within that development.
In the spirit of Nietzsche, though not in his words:
The danger is not that the machine becomes stronger than man, but that man becomes satisfied with becoming weaker in its presence.
—WJJH
Chess as a Human and Imperfect Activity
We humans are imperfect, and so is the chess we play. Perhaps that is part of the reason the game has survived for more than fifteen centuries.
Its beauty does not come from perfection alone, but from uncertainty: two people looking at the same sixty-four squares and seeing something slightly different.
Chess combines memory, pattern recognition, calculation and strategic foresight with experience, understanding and intuition. A move need not always be objectively perfect to be a good move, or even a beautiful one. Sometimes the mistake of one player creates the possibility for the imagination of the other. As Leontxo García observed in El País, much of the beauty of chess is born precisely from this human imperfection.
For all the theory surrounding it, my own conception of chess remains simple: two chairs, a chessboard and two glasses of wine. The game depends upon integrity and fairness. It is serious enough to make us think, but not so serious that we should forget to enjoy it.
Chess has survived kingdoms, wars, borders and changing ideas about what people ought to be doing with their time.
Spain offers a particularly colourful example. The game became deeply established there through al-Andalus and appears prominently in Alfonso X's thirteenth-century Book of Chess, Dice and Tables. Its relationship with the Church was never simply one of approval or prohibition. Clerics sometimes criticised chess when it became associated with gambling or idleness, yet it also entered courtly and educated life.
Ruy López de Segura, one of the founding figures of early modern chess, was himself a Spanish priest.
Chess proved remarkably difficult to suppress.
Muslims, Jews and Christians played it. Kings and priests played it. Merchants and soldiers played it.
The institutions changed.
The game remained.
Correspondence Chess in the Past
Then there is correspondence chess, the form of the game I still play.
Its beginnings disappear partly into legend. Games at a distance have been attributed to medieval rulers such as Henry I of England and Louis VI of France, while another tradition connects the Byzantine emperor Nicephorus with Harun al-Rashid. The evidence is uncertain.
By the seventeenth century, however, there are more substantial reports of merchants playing chess across considerable distances.
The documentary trail becomes clearer in 1804, when Friedrich Wilhelm von Mauvillon, stationed in The Hague, played three games through an exchange of letters with a fellow officer in Breda and later published the moves.
Twenty years later, an old Dutch rivalry found its way onto the chessboard. In 1824 the chess societies of Amsterdam and Rotterdam played two games by correspondence.
Amsterdam won both.
Apparently even postal chess could not escape Amsterdam and Rotterdam.
In the same year, the famous correspondence match between the London and Edinburgh chess clubs began. It continued until 1828, with moves travelling by post and committees discussing the positions at either end.
Correspondence chess demanded something modern players sometimes find in short supply: patience.
Players analysed the board, consulted books, studied earlier games and relied upon memory, judgement and intuition. Correspondence chess was therefore never entirely unaided chess. Research was part of its character from the beginning.
But books and earlier games could show what others had thought and played. The player still had to interpret that knowledge and answer the decisive question:
What should I play now?
The final judgement remained unmistakably human.
An international structure gradually followed. The Internationaler Fernschachbund was founded in 1928, succeeded after the war by the International Correspondence Chess Association, which became the International Correspondence Chess Federation in 1951.
Its motto, Amici Sumus — “We are friends” — expressed something particularly attractive about the spirit of correspondence chess.
The first World Correspondence Chess Championship final began in 1950 and ended in 1953, with the Australian Cecil Purdy becoming the first world champion.
By then correspondence chess had travelled a considerable distance from letters exchanged between acquaintances. Yet its essential character remained recognisable: two human beings separated by distance, each given unusually generous time to think.
That was about to change.
The Machine Enters the Room
The idea of using chess to explore machine intelligence is almost as old as electronic computing itself.
Alan Turing was already thinking about computer chess in the late 1940s, when the machines needed to run such programs scarcely existed. At roughly the same time, Claude Shannon approached the problem more systematically. His 1950 paper, Programming a Computer for Playing Chess, treated the game not merely as entertainment but as a laboratory for machine reasoning.
Shannon understood the central difficulty. A computer could not sensibly examine every possible continuation. It would have to search selectively, evaluate positions and distinguish promising moves from unpromising ones.
The question then was whether a computer could play chess at all.
Half a century later, the question had changed.
When IBM's Deep Blue defeated Garry Kasparov in 1997, it demonstrated that enormous calculating power, sophisticated evaluation and accumulated chess knowledge could overcome the strongest human player.
It was a psychological threshold as much as a technological one.
But Deep Blue was not the end of the story.
The later emergence of neural-network systems such as AlphaZero represented another change. AlphaZero learned chess through self-play and developed ideas that even strong human players sometimes found surprising.
The machine had progressed from learning how to play, to defeating us, to discovering ways of playing that could teach us.
It could calculate tactics, compare candidate moves, reject inferior continuations and assign evaluations to positions. Instead of merely helping the player remember more, it increasingly offered an answer to the question that had traditionally belonged to the player:
What should I play?
That is where assistance begins to become something else.
The human still makes the final move. But if the machine has identified the candidates, calculated the principal variations and ranked the choices before we decide, how much of that decision is genuinely ours?
The machine has not taken our chair away.
It has simply pulled up another chair beside us.
The Subtle Change
The most important change may not have occurred when a computer defeated a world champion. That was the spectacular moment, but not necessarily the most consequential one.
After losing to Deep Blue, Kasparov did not turn against the machine. He moved in the opposite direction. In 1998 he introduced what became known as advanced chess: human and computer working together, combining human intuition, experience and strategic judgement with the machine's memory and calculating power.
On the face of it, this was entirely reasonable.
Why compete against a tool that can make us stronger when we can learn to work with it?
Kasparov subsequently argued that the strongest combination was not necessarily the strongest player working with the strongest computer. What mattered was the quality of the process connecting human and machine. More recently he has suggested that machines may provide something like 95 or 96 per cent of the decision-making capability, leaving the final few percentage points for human judgement and purpose.
Perhaps they do.
But there is an important qualification.
Kasparov speaks from the perspective of one of the greatest chess players who ever lived. A Kasparov, Karpov or Fischer may possess the experience, intuition and originality to make those remaining few percentage points decisive.
Most chess players do not.
For the ordinary player, the imbalance between human and machine is greater still. We may formally make the final choice, but our capacity to improve upon — or even meaningfully challenge — the machine's judgement is much more limited.
That does not make human-machine collaboration meaningless. But it makes the question of agency more difficult.
Somewhere within that apparently sensible collaboration, something subtle begins to change.
At first the computer stored our games and helped us search what had been played before. Then it checked our calculations. Next it suggested alternatives. Eventually it began ranking those alternatives and telling us which move deserved our attention.
Each development was useful.
Yet the relationship had changed.
When I consult a database, I am asking what others have played.
When I consult an engine, I can ask what I should play.
The distinction appears small.
It is fundamental.
Engines became stronger, cheaper and widely available until computer analysis ceased to be an experiment conducted by elite players and became part of the ordinary landscape of chess. What once required grandmasters, assistants, libraries and considerable time could increasingly be done on an ordinary computer.
Once a tool becomes normal, refusing to use it begins to look less like independence and more like choosing a disadvantage.
There is obvious merit in these developments. Artificial intelligence can enrich analysis, uncover possibilities and deepen our understanding of positions. Modern engines have unquestionably strengthened our understanding of chess.
But the collaboration is no longer between equals.
If the machine has already selected the candidates, rejected inferior alternatives and established what counts as a reasonable choice, it has also defined much of the intellectual territory within which our decision is made.
Nothing has been taken from us by force.
Quite the opposite.
We willingly accepted more knowledge, greater accuracy, fewer mistakes and better decisions.
The computer did not demand authority.
We gradually gave it authority because it was so often right.
And once that happens, the question is no longer merely whether the machine helps us play better chess.
The question becomes what remains of the player.
Agency
Agency sounds like an abstract word, but the idea is simple.
It is our capacity to judge, to choose, to act and to accept responsibility for the consequences.
For most of human history, our limitations forced us to exercise it. We did not know everything. We could not calculate everything. We relied upon memory, experience, intuition, conversation and judgement.
We were often wrong.
But the decision remained ours.
Increasingly capable machines change that relationship because accepting their assistance is rational. They remember more, calculate faster and often produce a better answer than we could reach unaided.
We remain formally in control. We can reject the recommendation, choose another move or close the program.
But formal control is not necessarily the same as meaningful agency.
If a system has already framed the problem, selected the relevant possibilities and established what counts as the most reasonable answer, much of the intellectual work of deciding has taken place before we make the final choice.
We may still press the button.
We may still make the move.
But increasingly we may be approving a decision rather than originating one.
There is no conspiracy in this, and no coercion. The machine earns our trust one accurate answer at a time.
Chess makes the process unusually visible because the board is finite and the consequences comparatively harmless.
At first we think, and then use the machine to check whether we have overlooked something.
Gradually the order can reverse.
We ask the machine first.
Our intuitive move now requires approval.
The result may be better chess. It may even approach perfection.
But perfection has a price.
When players consult the same databases and increasingly powerful engines, choices tend to converge. Experiment becomes harder to justify. Surprise becomes rarer. The slightly eccentric move, the personal preference, the willingness to take a risk because a position simply feels right must all defend themselves against a machine that can immediately tell us another move is fractionally better.
The chess becomes more accurate.
It may also become more standardised.
And with that standardisation, part of the personality of the player risks disappearing from the board.
Perhaps that is the real cost of perfection: not that the machine becomes more capable, but that we become less willing to be imperfect on our own terms.
Which brings me back to Sundqvist's original question.
If so much of the intellectual work can now be shared with the machine, what remains of the human player — and why do we continue to play?
Sundqvist Asks What Remains of the Game: A Personal Answer
Chess is rather like an old friend who has changed with the times.
How we experience that friend depends greatly upon why we play.
I have never regarded chess primarily as a matter of rating or victory. I notice my rating, naturally, and I prefer winning to losing. But neither determines whether I have enjoyed the game.
I am more interested in whether I enjoy the game and can get the best out of myself than in whether I can extract the maximum possible result from it.
For me, chess remains entertainment, exercise for the mind and, at its best, an encounter with another person.
My earliest memories of chess are uncomplicated. As a child I played with my father on the sixty-four squares of an old wooden board that I still have. Later I played with my younger brother. Max Euwe's books helped us understand the game, and several of those books still remain on my shelves.
They were tools.
But they did not play for us.
We looked at the board, argued with ourselves, trusted our judgement and occasionally made dreadful moves.
That was part of the attraction.
The person on the other side could surprise you because people are capable of being surprising.
Chess changed during my lifetime, as almost everything else did. Fischer and Spassky turned a world championship into a global event at the height of the Cold War. Later Kasparov and Deep Blue made the relationship between humans and machines part of the chess story.
Databases, engines and online platforms subsequently brought extraordinary amounts of knowledge and analytical power within reach of almost everyone.
The strongest players had always carried a database of sorts in their heads: thousands of remembered games, structures and patterns.
The modern database externalised that memory.
The engine went further.
It began to interpret it for us.
Much of this has been beneficial. Chess has become more accessible, more international and easier to study. Millions of people who might never have entered a chess club can now find an opponent within seconds.
Yet something has also become less certain.
When I play, there can be a small question at the back of my mind:
How much of the person on the other side am I actually meeting?
In over-the-board chess, where engine assistance is prohibited, the answer should be straightforward. Another fallible human being is making the decisions.
In correspondence or freestyle forms where computer assistance is permitted, the distinction becomes less clear. The player may make the final move, but the ideas, calculations and evaluations behind it may increasingly have been shaped by the machine.
That does not make the game dishonest.
If the rules permit assistance, the player is doing nothing wrong. Different forms of chess have different rules. In ICCF correspondence chess engine assistance is permitted; elsewhere it may be prohibited.
But the distinction matters.
Where engines are permitted, we must ask how much judgement remains with the player.
Where they are prohibited, we depend upon something technology cannot itself provide: trust that the person on the other side is actually playing the game.
Style has not disappeared. The aggressive player and the cautious player are still there. So are tactical imagination and positional understanding.
But when the same machine can tell both players what a position objectively requires, the space in which those differences can express themselves becomes narrower.
Perhaps machines have made chess objectively stronger while making it a little less mysterious.
For some players, that is progress. They enjoy the search for greater accuracy and the challenge of working effectively with increasingly powerful tools.
There is considerable merit in that view.
Sundqvist himself has adapted remarkably well to the engine era. He approaches correspondence chess partly as a problem-solving exercise, progressively closing possibilities and searching for precision, even while questioning what this development has done to the human character of the game.
There is a certain irony in that. His reservations are not those of someone left behind by technological change, but of someone who has learned to operate successfully within it.
My own relationship with chess is less ambitious and therefore perhaps less accommodating.
My answer is not that machines have ruined chess.
They have not.
They have made chess stronger, opened new analytical possibilities and allowed us to understand positions with a precision previous generations could hardly have imagined.
But strength and meaning are not necessarily the same thing.
Some of my most enjoyable games were followed by post-mortems that lasted longer than the games themselves — occasionally helped along by the wine.
We replayed positions, defended questionable decisions, laughed at mistakes and discovered that the other person had often been thinking about something entirely different.
That was part of the game.
An engine can tell us that a move was inaccurate. It can usually tell us what should have been played instead.
What it cannot reproduce is the encounter between two imperfect minds trying to understand the same position in different ways.
So perhaps the future of correspondence chess depends not upon resisting machines, but upon preserving enough space for the human player to remain visible within the game.
If that space becomes too small, correspondence chess may continue to produce extraordinarily strong chess.
Whether it will still produce the same kind of encounter is another question.
Chess, the Watchers and the Future
Chess has evolved because society has evolved.
We should not pretend otherwise, nor complain as though technology were something imposed upon us from outside. We built the databases, improved the engines, embraced their convenience and learned to trust their results.
We are all part of the process.
That is precisely why chess is such a useful mirror.
We begin as players. Then we become assisted players. Increasingly, we may become supervisors of systems that calculate, compare and recommend more effectively than we can.
Eventually there is a danger that the player becomes a watcher.
This is not only a chess problem.
Across modern life, machines increasingly help us search, remember, compare, predict and decide. They save time and reduce mistakes. Relying upon them is often perfectly rational.
But delegation has consequences.
The more often we allow systems to determine what deserves our attention, which option is best and which course is most efficient, the less often we are required to exercise judgement ourselves.
Nothing needs to be taken from us by force.
We may surrender it because the alternative appears slower, less accurate and increasingly unnecessary.
Nietzsche might have recognised the danger: not because machines become stronger, but because human beings may become content to exercise less of their own will in their presence.
There is no returning to an earlier board.
Nor should there necessarily be.
The machine entered the room because it was useful and powerful. Players adapted. Institutions adapted. Rules adapted. What once seemed exceptional became normal.
The interesting question is therefore not whether artificial intelligence should leave chess.
It will not.
Nor is the question whether human beings can still defeat machines. In any meaningful sense, that argument has already been settled.
The more important question is what kind of relationship we want with them.
Perhaps the future of correspondence chess is not the disappearance of the human player, but a renegotiation of the boundary between player and machine.
For some, this will represent progress. Chess will become more accurate, more sophisticated and analytically richer. Human beings working with machines may continue to discover ideas that neither would have reached in quite the same way alone.
There is nothing inherently wrong with that.
Agency in such a world cannot simply mean doing everything unaided.
That world has already gone.
It may instead mean knowing what to delegate and what to retain; what to question and what to accept; when to trust the recommendation and when to accept the possibility that our own imperfect judgement remains worth exercising.
Correspondence chess has always adapted. Letters overcame distance. Books expanded knowledge. Databases externalised memory. Engines added calculation and evaluation. Artificial intelligence will take that development further.
Why should chess be different from the society of which it is part?
It should not.
But adaptation need not mean surrender.
The machine has no ambition, no vanity and no desire for the chair.
It does not need them.
We may simply offer it the chair because it plays better than we do.
That is why the final question is not whether machines will become more powerful.
They will.
The real question is what human beings choose to retain when they do.
The final move may still be ours.
The responsibility may still be ours.
The purpose must remain ours.
William J J Houtzager, September 2026
📌 Blog Excerpt
Correspondence chess has always changed with technology: letters overcame distance, books expanded knowledge, databases externalised memory, and engines added calculation and evaluation. But when the machine increasingly decides what deserves our attention and what move is best, a deeper question appears. What remains of the player? A reflection on chess, artificial intelligence, human imperfection and the meaning of agency.
Keywords:
Chess, correspondence chess, Christer Sundqvist, artificial intelligence, AI, chess engines, Garry Kasparov, Deep Blue, AlphaZero, Claude Shannon, Alan Turing, Cecil Purdy, ICCF, human agency, memory, judgement, technology, correspondence chess history, future of chess
No comments:
Post a Comment