Chess: Science or Art?
Chess demands science-based problem solving skills; a fact borne out by the success of AI chess computers. But can aspects of gameplay also constitute art?
During chess, players draw on two types of thinking. Explicit thinking is a conscious process where we form and test hypotheses in a scientific manner to find solutions in the game. Implicit thinking is an unconscious process of pattern-matching between long term memories, which give rise to hunches and instincts. When either type produces a novel move, it could be argued as a creative leap. Does this make chess a science and an art?
The Science of Chess
Chess is fundamentally a science based on rigid rules and a single objective: checkmate. To play chess, you have to build a scientific model of the game in your mind. It starts with basic principles: how the pieces move, how to attack, how to defend, and how to achieve checkmate. This is a systematic bottom-up process of discovery, just as we employ in the scientific method.
Beginner chess players engage in one experiment after another to learn how this rule-bound system works. Soon, you hit on the repeated observation that it's better to own the centre of the board with pawns. This restricts your opponent's lines of attack and provides a buffer to develop your own pieces.
Gameplay arises from theory, based on a collection of evidence, that guides your exploration of chess to more complex scenarios and deeper levels of analysis.
Such chess theories hold true at all levels—from beginner to grandmaster—revealing the inherent order of the game. Within this order, you can analyse the board, use logical problem solving, and overcome multiple challenges, simultaneously and sequentially. Sometimes the solution is to fail one problem (surrender a piece) to solve another (win a piece).
This is what makes chess so addictive. Every game is a new set of problems to be solved.
As you play more powerful opponents, the problems become harder, allowing you to develop your game with increasing dexterity. All the while, you develop your ability to visualise the chess board, memorising static and mobile positions as you test out the consequences of various moves.
The science of chess is supported by the fact that chess engines now significantly outrank the best chess players in the world. They're evidence that the game can be deconstructed, played, and won at the highest levels without novel creative input. If we find art in AI computer chess, then we have a bit of a problem in how we define creativity—or AI.
First, let's make a short analysis of chess as a science, breaking it down into the opening, the middlegame, and the endgame.
Chess Openings: By The Book
A chess opening is a sequence of initial moves, a little like the obligatory small talk when you meet someone new. They occur somewhat passively and habitually, in order to establish a more meaningful interaction.
There are 1,327 chess openings and variants listed in The Oxford Companion to Chess. The oldest openings were named for their geographical origins, while modern openings take the names of the players who popularise them.
Openings are never arbitrary. They're highly strategic, chosen because they allow you to protect your pieces (defensive play) or challenge your opponent (aggressive play). They set the tone of the game to follow, with each opening defining the next set of problems to be solved.
Here are the top ten chess openings for beginners to learn.
- The Italian Opening
- The Sicilian Defence
- The French Defence
- The Ruy-Lopez
- The Slav Defence
- The Fried Liver
- The Four Knights
- The Queen's Gambit
- The London System
- The Dutch Defence
On average, chess grandmasters know around 1,200 different openings. They study chess books to leverage the work of past players, benefitting from already-established conclusions. Sound familiar? They're like scientific theories: accepted conclusions based on observations and experiments, which provide a platform for deeper exploration.
Many new chess players naturally wonder: what's the fastest way to get a checkmate? Under the right circumstances, it's possible to win a game of chess in only two moves. It's called The Fool's Mate. But as the name suggests, your opponent must be very inexperienced for it to work out. It's more of a party trick (imagine that party...) than a mark of genius.
Devising your own openings (or rather, hitting on established openings without knowing it) is fun for a while, and perhaps necessary to discover the tactical mindset of chess first-hand. But if you want to advance your game, you need to study chess openings. Read a chess book (Bobby Fischer Teaches Chess is an all-time classic) or study interactive openings online. This is where you'll realise how much work has already been done.
All modern chess players need to study openings, just as all modern scientists need to study science. No burgeoning biologist today needs to figure out the mechanism of evolution from scratch, when Darwin and Wallace already laid out the evidence and theory of natural selection before them. Your time and efforts are better spent devouring existing knowledge and applying what you learn to unsolved chess problems.
Such problems emerge in abundance in the delightful middlegame.
Chess Middlegames: Myriad Problems
Chess openings develop into the middlegame when a player goes ”out of book”. In a single move, they deviate from a pre-defined opening algorithm and begin to think independently. This is where the real fun begins.
One of the many consistent rules in chess is that every move counts. If you don't continue to develop your pieces, even for a single turn, you give your opponent a competitive edge. This constant pressure informs your entire strategy.
How many turns do you get in a chess game? On average, it's 38 moves. The longest chess game in history was 269 moves (Nikolic vs Arsovic, 1989) and ended in a draw. The game took 20 hours.
The combinations of possible moves and countermoves in a chess game makes for astounding complexity. According to Claude Shannon's 1950 paper, Programming a Computer for Playing Chess, by the seventh move there are already 3.2 billion unique games that could have been played.
Shannon's Number posits that a 40-move game of chess presents 10120 possible games.
Aggressive chess games are often won sooner, either when the aggressor's risk pays off, or they become trapped in a losing scenario. But defensive games are slow burners. Players become locked in a war of attrition where neither is willing to risk losing material.
While chess openings rely on memorisation of algorithms, the middlegame is driven by independent problem solving using the scientific method:
Observation. All my pieces are developed, but I'm yet to launch an attack.
Question. Which of my opponent's pieces are undefended?
Hypothesis. My knight can take the pawn without risk of capture.
Experiment. Visualise! Don't actually perform the move yet.
Conclusion. Nope, if I do that, the queen can fork my knight and bishop.
As the middlegame increases in complexity, the challenge is to perform this process with increasing breadth and depth to out-think your opponent. Breadth means expanding your scope of available moves, while depth means thinking more moves ahead.
Beginners start out thinking just one or two moves ahead, yet grandmasters often look eight moves ahead.
This ability to imagine many moves ahead is why so many grandmasters have well-developed memory and visualisation skills.
Novice chess players aren't only disadvantaged by lack of foresight and book knowledge, but also by failure to adhere to the scientific method. Beginners routinely fail to falsify their own hypotheses, convincing themselves that risky moves will still work out in their favour. They're more likely to focus on countermoves that benefit them, and overlook how their opponent might take advantage. Fortunately, this can all be rectified with training.
From the order of the opening, chaos emerges in the middlegame. You have more material in the centre, affording more legal moves, more chances to attack, and more to necessarily defend.
Chess players have developed numerous ground rules to improve their middle game. Simple principles that give a distinct advantage include:
Avoiding Pawn Islands. When pawns defend each other in a line, they're formidable. But once you lose your pawn structure, you have to spend valuable resources plugging that hole in your defence.
Keeping Your Bishops. While knights shine in the middlegame, bishops are superior in the endgame. So don't sacrifice a bishop for a knight or pawn without good reason.
Connecting Your Rooks. Castling your king allows your rooks to protect each other on the back rank. They can also defend pieces out in the open and put pressure on your opponent.
Trading Strategically. When you sacrifice a piece, make sure it supports your overall strategy. In games with strong pawn structures, defensive play helps you retain material. Else, a bold sacrifice can expose a fatal weakness in your opponent's strategy.
Chess Endgames: Beautiful Moves
Play proceeds to the endgame when few pieces remain on the board and/or at least one queen is out. Since there are so few pieces, there's also a decline in the variety of available moves.
Like openings, there are common endgame tactics you can exploit. For instance, the Back Rank Mate punishes an opponent who fails to activate their king in the endgame.
Here are five more types of checkmates to use in the endgame:
- Two Rooks Checkmate
- King and Queen Checkmate
- King and One Rook Checkmate
- King and Two Bishops Checkmate
- Arabian Mate
Are endgames easier than middle games? There are certainly fewer pieces to forecast, but this is also true for your opponent. The game becomes laser focused on protecting the king and promoting pawns, so don't ease up your scientific analysis, nor stop reaching for creative solutions.
Besides learning about book checkmates, you can improve your chess endgame with certain principles:
Simplify The Board. If you have extra material in the endgame, drive at trades to simplify the game while remaining ahead.
Promote Your Pawns. Keep your opponent busy by threatening to advance pawns for promotion. A pawn race—where both players have pawns making a beeline—can decide the game.
Beware of Stalemate. A chess game ends in an anti-climactic draw when your opponent's king isn't in check, but they have no legal moves available. On the other hand, if you're down material, you can force stalemate on yourself and deprive your opponent of the win.
Chess Computers vs Grandmasters
Chess engines were originally developed as a way to understand artificial intelligence. It was Turing, Von Neumann, and Shannon who asked: can machines think like humans? They each investigated systems of machine chess, an enquiry that culminated in the development of MANIAC which, in 1956, was the first chess computer to defeat a human opponent.
Chess algorithms got smarter, and by 1985 achieved ratings of well over 2000, which put them in the 98th percentile of human chess players.
In 1997, Deep Blue (rated 2700) famously became the first chess computer to beat a reigning world champion: Garry Kasparov (rated 2863).
Kasparov attributed his critical loss to a single blunder and insisted it proved nothing. However, Kasparov would later introduce Advanced Chess, in which human players explore candidate moves with chess bot assistance, but ultimately control the game themselves.
In the years since the famous Kasparov showdown, chess engines have advanced their ratings to over 3500; far beyond even the top grandmaster ratings. With a peak rating of 2882, Magnus Carlsen is the highest rated chess player of all time, yet he won't take on chess computers for official matches. Commentators say there would simply be no contest.
In fact, chess computers are so good they now have to play each other to determine their ratings.
Here's an example of a middlegame between AlphaZero vs AlphaZero. Though not yet officially rated, this chess bot is thought to have a rating of around 3500.
Chess engines are built on decision tree algorithms, analysing all the options at each turn and selecting the move with the highest probability of winning the game. This offers a major advantage because computers can analyse many more moves ahead than the human brain. (Although no computer can calculate all 10120 possible scenarios of a typical 40-move game—yet.)
Chess engines also have the advantage of more gameplay. There are databases of billions of chess games played online, by both humans and machines, on which chess bots can perform statistical analyses.
Although Carlsen has been playing chess since he was five years old, there's no way he, nor any other human, can compete with that.
But there is one thing that chess computers lack: intuition. Chess bots don't play chess; they merely simulate it.
When humans play, we use explicit and implicit problem solving processes to reach decisions. Implicit thinking—that unconscious, unknowable type—prevents us from explaining why one move feels better than another. Kasparov used to say that certain moves "smelled right". Computers can't make these kinds of leaps.
Is Chess an Art?
So far, I've presented chess as a science: a systematic process of observation, analysis, experimentation, and conclusion, serialised to the point of checkmate. It requires order, logic, and problem solving. But there are angles from which we might consider chess as a window onto art.
The definition of art is broad: any activity that expresses imaginative or technical skill.
Right away, this definition lends itself to the middle and endgames of chess, when players use imagination to form novel hypotheses and solve the board. This answer alone would satisfy some that chess is an art, but let's probe a little harder.
By necessity, chess extends beyond conscious problem solving. Even the world's best chess players have a finite working memory to visualise moves and countermoves into the future. At almost every move in the game, players have to commit themselves to a decision without knowing if or how it will lead to checkmate. This is where implicit thinking lends a hand.
The intuition that arises from implicit thinking provides conclusions beyond the realm of conscious reasoning. Psychologists believe it relies on pattern-matching, where the unconscious mind sieves through long term memories for similar scenarios. If any matches are found, we feel driven by instinct.
In the same way, on and off the chess board, your unconscious mind works at solving chess puzzles. It's searching for solutions, some of which may very well be novel. This doesn't necessarily equate to creativity, but it does put us in the right domain.
Implicit processing is frequently attributed by artists and scientists alike to some of their greatest works. Whether a stroke of insight occurs in the shower or asleep in a dream, history is replete with stories of sudden creative solutions.
Dreams—drawn from implicit processing—were cited as the spark of inspiration for the structure of the atom by Bohr, the speed of light by Einstein, and the chemical nature of nerve impulses by Loewi.
If creativity is the combination of two or more separate ideas to reach a new conclusion, then art is the end product that expresses it. Art can be a performance or a display (which chess can be), and it can evoke an emotional response (which chess can do).
Above all, art strives to portray a novel perspective. And when we watch grandmasters play chess, we're itching to see what novel solutions they can apply to new problems.
So there's a case for describing chess as a science and an art, although the argument for art is somewhat subjective. Indeed, that's the very nature of art. If we put science in the box of order and logic, then art lives in the box of chaos and creativity, sometimes escaping the box just to have a look around. That's what art does.
If we do perceive art in chess, we have to admit it sits within a very narrow frame.
Designing chess pieces—now that's art. The Queen's Gambit (the show, not the opening)—that's art. Art delivers best when it isn't constrained by heavy limitations or rules. Nonetheless, this doesn't stop chess fans describing certain games as "beautiful", for beauty is in the eye of the beholder.
The idea of chess as a science and an art has ramifications on how we view art and/or artificial intelligence. It's thought that broad AI will outstrip human thinking during the 21st century, and those who doubt this prediction point to a lack of two human functions: consciousness and creativity. But the rise of AI suggests our idea of creativity isn't so well defined. If chess computers are deemed capable of producing beautiful games, we have to rework our definition of art, or artificial intelligence, or both.