An interactive openings board
Learn chess the way masters do — one move at a time.
Play a friend from a link, or play the computer
Play a real game: against the computer, hot-seat on one board with a friend next to you, or online — two people on two devices, paired by a link one of them sends, with no account on either side. Then pick an opening and watch it unfold move by move. Nothing to install.
The link-paired game comes with a chat box, and the chat runs directly between the two browsers when that connection can be established; when it cannot, the board is unaffected and the game carries on without chat. In that game the moves themselves travel through our relay to reach the other player, and the board page spells out exactly what the relay sees.
When the game is over you both stay connected, so a rematch is one button and the colours swap — no second link to send. And you can end a game the way people actually end games: resign, or offer a draw and have your opponent accept it.
The computer plays at three strengths, and you choose. Pick the computer on the board page and a Computer strength control appears beside it with three buttons — Easy, Normal and Hard. It opens on Normal, so a beginner can set it to Easy before the first move, and it can be changed again part-way through a game. The setting is how far ahead the engine searches — one half-move at Easy, two at Normal, up to three at Hard — and also how much slack it gives itself when it picks. Hard always plays its top choice; Normal picks at random from anything it rates within 0.9 of a pawn of that, and Easy from anything within 2.5 pawns. So Easy will play a loose or passive move where Hard plays the sharpest one available. The 2.5-pawn ceiling is deliberate, because it is less than a knight is worth: a move that drops a piece in the exchanges the engine plays out already comes out further behind than that, so no setting hands you a piece just to go easy on you, and all three still find a mate in one. Whichever level you played is written into the PGN header when you export the game, so a saved game records which one it was.
The repertoire
Ten openings worth knowing
Every game starts here. These are the reliable main lines — the ones behind most games you'll ever play. Click any card to load it onto the board above.
Every line on these cards is written in the standard shorthand for chess moves, called algebraic notation — the same shorthand modern chess books, recorded games and the boards on this page all use. It works because every square has a name. The eight columns are files, lettered a to h starting from White's left; the eight rows are ranks, numbered 1 to 8 starting at White's end of the board. A square is its file letter then its rank number, and the board above prints the letters along its bottom edge and the numbers down its left edge. So e4 is the square four rows up the e-column — and it is also a move, because a move is written as the square the piece lands on.
What goes in front of that square is the piece. K is the king, Q the queen, R a rook, B a bishop and N a knight — N, because the king already has the K. A pawn gets no letter at all, just the square, which is why e4, e5 and c6 are pawn moves while Nf3 is a knight going to f3 and Bc4 a bishop going to c4. Everything else is punctuation. An x before the square means the move captures whatever was standing there: Nxd4 is a knight taking on d4, and cxd4 is a pawn taking on d4, where the lone c is the file the pawn came from, since pawns have no letter of their own. A + on the end means the move gives check, as in Bxc3+ in the French Defence below. A # means checkmate, which is why the two mating answers among the tactic cards further down this page are written Re8# and Nf7#. Castling is the one move that names no square: O-O, two capital letter O's, and the Italian Game line below ends O-O O-O with both players tucking their king away.
Two things catch people out. When either of two identical pieces could reach the same square, the notation squeezes in where the mover started, so Nbd2 in the London System below means the knight on the b-file, not the other knight that could also get to d2. And the numbers with full stops are move numbers, not squares: one number covers a White move and Black's reply, so the Italian Game's third pair is written 3. Bc4 Bc5 — White's bishop to c4, then Black's to c5. Read the cards left to right and White always moves first.
Both plans, the branch at move four, and the trap
White's whole idea is the bishop on c4: it looks at f7, the one square near Black's king that nothing but the king defends. Everything else is unhurried. The pawns on c3 and d3 hold a modest centre White hopes to expand with d4 later, and the queen's knight often walks b1–d2–f1–g3 to reach the kingside. Black copies the plan a move behind: the bishop on c5 aims at f2, ...d6 props up e5, and a later ...a6 and ...Ba7 tuck that bishop out of reach before d4 can arrive with tempo.
The structure is what keeps it quiet. e4 and d3 face e5 and d6 with no pawns in contact anywhere, so nothing is forced and both sides simply improve pieces until someone commits to d4 or ...d5. That is why this exact treatment is nicknamed the Giuoco Pianissimo — the very quiet game.
The road forks at move four. 1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. c3 is the slow build. 1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. b4 is the Evans Gambit: a pawn handed over so that c3 and d4 land two moves sooner while the bishop on c5 is chased about. Take the first if you are content to outplay someone over forty moves, the second if you want the position alight by move ten.
The club-level mistake is the greedy one, 1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. Ng5 Qxg5: White throws a second attacker at f7 and the black queen just takes the knight, because nothing defends g5. That raid only works once Black's knight has gone to f6 and blocked the diagonal, which is the position after 1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5.
Why the threat is fake, and what the exchange really buys
The bishop on b5 attacks the knight that defends e5, and the first thing to learn is that the threat is a bluff: 1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Bxc6 dxc6 5. Nxe5 Qd4 forks the knight on e5 and the pawn on e4, and Black takes the pawn straight back. Snatching e5 like that is the most common club mistake in the whole opening.
The bishop's real job is slower. It steps back to a4 and then b3, still pointing at f7, while White plays c3 and d4 and reroutes the queen's knight b1–d2–f1–g3 towards the kingside. Black takes the other wing: ...b5 gains space, ...c5 claims more, and the awkward c6 knight either goes to a5 to hit the bishop or retreats to b8 and d7 to rebuild behind the pawns. The centre stays blocked for a long time, which is why this opening has a reputation for manoeuvring rather than fireworks.
The branch is move four. Retreating to a4 keeps the bishop and the long game. 1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Bxc6 dxc6 is the Exchange Variation, and that trade has a concrete point: Black's doubled c-pawns mean his four queenside pawns can never make a passed pawn against White's three, while White's four kingside pawns against three can. White gives up the bishop pair to bank that endgame. Play the Exchange if you like simple positions and few surprises; keep the bishop if you want the richer middlegame.
The c-file, the hole on d5, and the choice at move six
Black gives up a wing pawn for a centre pawn and gets something permanent in exchange: the half-open c-file. Rooks come to c8, the queen sits on c7, and the pressure down that file never really goes away. The ...a6 on this card is the Najdorf move — it denies b5 to White's pieces and prepares a ...b5 and ...Bb7 expansion of Black's own.
White has more space and a lead in development, so the plans are direct: either the calm setup shown here — bishop to e2, castle short — or bishop to e3, pawn to f3, queen to d2, castle long and run the g- and h-pawns at the black king, after which both sides attack on opposite wings and whoever is a move quicker wins.
The ...e5 push is the structural decision worth understanding. It grabs the centre and kicks the knight, but it leaves a hole on d5 that a white knight can sit in for the rest of the game, and the d6 pawn behind it is backward on a half-open file. Black takes that deal because the c-file gives him something real to play for; an unbalanced position you can actually win is the point of the Sicilian.
Move six is where the game truly forks: bishop to e2 (quiet, little to memorise), bishop to e3 (the English Attack, opposite castling, a pawn race), or bishop to g5 (sharpest, hitting the f6 knight at once). The common club error comes earlier, at 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Qxd4 Nc6: recapturing with the queen looks natural, but the knight hits her immediately, so Black develops a piece while White moves the same one twice.
The pawn chain, the four third moves, and the d4 trap
The French buys solidity with one concession. The pawn on e6 supports the coming ...d5, but it also shuts in Black's light-squared bishop, and finding work for that bishop is the story of most French games. It is the exact problem the Caro-Kann is built to dodge.
Once White pushes to e5 the centre is a locked chain — White's pawns on d4 and e5 against Black's on e6 and d5 — and a chain is attacked at its base. So Black hammers d4 with ...c5, ...Nc6 and ...Qb6, while White uses the space the e5 pawn gives him to play on the kingside. In the Winawer shown here Black trades his dark-squared bishop for the c3 knight: White gets the bishop pair and a broad centre, Black gets doubled white c-pawns to lean on. The catch is that the traded bishop was the one guarding g7, which is why White's queen so often heads for g4.
Move three is the real branch. The knight to c3 invites the pin above. 1. e4 e6 2. d4 d5 3. e5 is the Advance, closing the centre at once and suiting a player who wants one plan and little theory. 1. e4 e6 2. d4 d5 3. Nd2 is the Tarrasch, which sidesteps ...Bb4 at the cost of blocking the c1 bishop for a few moves. 1. e4 e6 2. d4 d5 3. exd5 exd5 is the Exchange: symmetrical, quiet and the easiest road to a draw.
Most club Frenches are Advance Frenches, and this is the trap that decides them: 1. e4 e6 2. d4 d5 3. e5 c5 4. c3 Nc6 5. Nf3 Qb6 6. Bd3 cxd4 7. cxd4 Nxd4 8. Nxd4 Qxd4 9. Bb5+. White's sixth move looks as though it abandons d4; it does not, because the check on b5 arrives while the black queen still stands on d4, and after every legal answer she is lost. Count the defenders one more time before accepting a pawn that looks free.
What ...c6 costs, what h4 h6 is for, and the one-move-early mistake
The Caro-Kann wants the same pawn on d5 that the French wants, but it props that pawn up with the c-pawn instead, so the e-pawn never blocks the light-squared bishop. The bishop reaches f5 before the door closes, and that is the entire idea. The price is paid elsewhere: ...c6 takes the natural square away from the b8 knight, so Black is a move slower and usually settles for ...e6, ...Nd7, ...Ngf6 and a sturdy, slightly passive position.
Trading on e4 hands White the freer game. What it hands Black is an absence of weaknesses: no doubled pawns, no holes, and endgames that are easy to hold, which is why this defence suits players who would rather defend an endgame than a king. The h4 and ...h6 pair at the end of the card is not filler either: White gains space and asks the bishop where it intends to live, while ...h6 denies g5 to White's pieces and stops the h-pawn running on to h6 itself.
The branch is move three. The knight to c3 or d2 leads to the line on this card. 1. e4 c6 2. d4 d5 3. e5 is the Advance, where Black still gets the bishop outside the chain and the game turns on White's extra space. 1. e4 c6 2. d4 d5 3. exd5 cxd5 4. c4 is the Panov, which turns the most solid defence in chess into an open middlegame, usually with an isolated d-pawn and pieces flying — choose it if you want sharp play rather than a grind.
The club mistake is developing that famous bishop one move too early: 1. e4 c6 2. d4 d5 3. Nc3 Bf5 4. exd5 cxd5 5. Qf3 attacks the bishop and the d5 pawn at the same moment. Black can defend both by dropping the bishop back to e6 — the very square the opening exists to avoid — and otherwise he loses a pawn. Resolve the centre first, then bring the bishop out.
Why the gambit is not a gambit, the problem bishop, and the trap
White's second move offers the c-pawn and it is not a real sacrifice. After 1. d4 d5 2. c4 dxc4 Black has no way to keep it: White plays e3 and Bxc4 and takes it back, and the attempt to prop it up with ...b5 is undermined by a4. What the offer really asks is a trade — a wing pawn for a centre pawn — because the moment Black captures on c4 there is nothing left to contest e4, and White plays it. Declining with ...e6 refuses the trade and keeps a black pawn on d5 for as long as it is doing work.
The pawn on e6 does the propping, and it also locks in Black's light-squared bishop — the identical complaint the French Defence lives with. Freeing that bishop is the black middlegame, and there are three standard routes: ...b6 and ...Bb7 onto the long diagonal, ...dxc4 timed so that ...b5 and ...Bb7 come with tempo, or trading pieces with ...Ne4 until a cramped position has too few pieces left to be cramped. The ...h6 that ends this card belongs to the first plan — it asks the g5 bishop to declare itself, and after 1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Bg5 Be7 5. e3 O-O 6. Nf3 h6 7. Bh4 b6 that bishop is on its way to b7.
White's pieces all point at d5: the c4 pawn, the knight on c3, and the bishop on g5, which pins the f6 knight to the queen until ...Be7 breaks the pin. The most instructive plan comes after 1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. cxd5 exd5, the Exchange Variation. White now has two queenside pawns against three and advances them anyway with b4 and b5, aiming to trade on c6 and leave Black a backward pawn on a half-open file. That is the minority attack — attacking with fewer pawns on purpose, because the object is a weakness rather than space — and it is worth learning even if you never play this opening from either side.
The trap here is Black's, and it punishes exactly the greed the position invites. 1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Bg5 Nbd7 5. cxd5 exd5 6. Nxd5 looks like a free pawn, because the f6 knight is pinned to the queen and cannot recapture. It can. 1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Bg5 Nbd7 5. cxd5 exd5 6. Nxd5 Nxd5 7. Bxd8 Bb4+ 8. Qd2 Bxd2+ 9. Kxd2 Kxd8 and Black is a knight up for a pawn: the queen was worth giving because the check on b4 arrives before White can move again. Before you take a pinned piece's defender, work out what happens if the pinned piece moves anyway.
The provocation, the pawn race it leads to, and the two ways it goes wrong
The King's Indian is a deliberate provocation. Black develops the kingside, castles, and invites White to build the biggest centre he wants — pawns on c4, d4 and e4 — on the theory that a big centre is also a big target. The bishop on g7 is the entire argument: from the moment it lands it looks down the long diagonal at d4 and c3, and every pawn break Black plays later is timed to open that diagonal rather than to win a square.
The break is ...e5, and what White does about it decides the character of the game. The main road pushes past it: 1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Nf3 O-O 6. Be2 e5 7. O-O Nc6 8. d5 Ne7. White's pawns on c4, d5 and e4 now face Black's on c7, d6 and e5 with nothing in contact, so the centre is shut and each side plays on the wing its own chain points towards: White pushes c5 and pours pieces down the c-file, Black steps the f6 knight aside and throws ...f5, ...f4, ...g5 and ...h5 at the white king. It is a race on opposite wings, and it is usually settled by who arrives first rather than by who stands better.
Two things go wrong for beginners, and both are about the tension in the centre. The first is releasing it yourself: 1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Nf3 O-O 6. Be2 e5 7. O-O exd4 8. Nxd4 gives White a free hand in the middle and leaves the g7 bishop staring at a knight that is comfortably defended. Make White resolve it. The second is not noticing when White has taken the attack off the board: 1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Nf3 O-O 6. Be2 e5 7. dxe5 dxe5 8. Qxd8 Rxd8 is a level, queenless endgame, and it is White's most irritating answer precisely because the kingside storm Black bought with all that passivity now has no king to storm. If you play this opening you have to be willing to play that endgame properly instead of attacking anyway.
The other setup worth recognising is 1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. f3, the Sämisch, where White supports e4 with a pawn instead of a piece and often castles long — the same opposite-wing race, with White's attack the faster one. Honest limit: this is not a low-theory choice. Black is objectively worse in the centre for a long stretch on purpose, and the compensation is an attack you have to know how to conduct.
The one move-order rule, the attacking plan, and the poke every London player meets
The London is a setup rather than a variation: pawns to d4, e3 and c3, knights to f3 and d2, bishop to d3, castle short — and the dark-squared bishop out to f4 before e3 goes in. That ordering is the one rule the system genuinely has. Play e3 first and the bishop is serving a life sentence behind its own pawns, which is the same misfortune the French Defence inflicts on Black and the Caro-Kann is built to avoid.
What the setup is for: the knight on d2 supports Ne5, the bishop on d3 looks at h7, and with a rook on e1 or a pawn to f4 White has a kingside attack that needs almost no memorised theory. Black's standard antidote in this card's line is ...Bd6, offering to swap the dark-squared bishops and take the attacker off, and White usually side-steps it — 1. d4 d5 2. Nf3 Nf6 3. Bf4 e6 4. e3 c5 5. c3 Nc6 6. Nbd2 Bd6 7. Bg3 O-O 8. Bd3 keeps the bishop, keeps the diagonal and keeps the plan.
The poke everyone meets is ...Qb6. Once the bishop has left c1, nothing guards b2 — the a1 rook covers a2 and b1, not b2 — and Black's most common try is to go straight after it with 1. d4 d5 2. Nf3 Nf6 3. Bf4 c5 4. e3 Qb6. It is not a refutation, but it is the proof that this system cannot be played without looking. The three replies London players actually use are 1. d4 d5 2. Nf3 Nf6 3. Bf4 c5 4. e3 Qb6 5. Qc1, ugly but solid, with the queen coming back out later; Qb3, offering to trade queens on the square Black chose; and Nc3, which lets Black take on b2 in exchange for a development lead. Which one you pick matters far less than noticing the question was asked.
Against a kingside fianchetto the nuisance is ...Nh5, hitting the bishop. The answer is to tuck it onto h2, and that is why h3 keeps appearing a move early — h2 has to be empty first, so Bh2 is simply not a legal move until the h-pawn steps up: 1. d4 Nf6 2. Nf3 g6 3. Bf4 Bg7 4. e3 d6 5. h3 O-O 6. Be2 Nh5 7. Bh2. Honest limit, because it is the thing London players oversell: this opening buys you a safe, familiar position against almost anything and saves a great deal of study, and that is real. It does not buy an advantage. Expect a balanced middlegame you understand, not a free half point.
The reversed Sicilian, what the extra move is worth, and the transposition catch
1. c4 fights for d5 from the flank instead of occupying the centre. Black's most direct answer is ...e5, and once that pawn lands you are looking at a Sicilian Defence with the colours swapped. Put 1. c4 e5 2. Nc3 Nf6 3. Nf3 Nc6 4. g3 beside 1. e4 c5 2. Nc3 Nc6 3. g3 and it is the same structure, played by the other player, a move sooner.
That extra move is worth something, but less than people expect, and the reason is worth stating: in a slow position moving first also means committing first, and the opponent gets to choose his setup after seeing yours. The practical value of 1. c4 lies elsewhere — in the fianchetto. The bishop on g2 attacks d5 down an open diagonal and keeps attacking it for the rest of the game, and White's plans are built around it: Rb1 and b4 to take queenside space, or d3 followed by a knight retreat and f4 on the other wing.
The card's line shows the central fight in miniature. After 1. c4 e5 2. Nc3 Nf6 3. Nf3 Nc6 4. g3 d5 5. cxd5 Nxd5 6. Bg2 Nb6 Black has struck in the centre and stepped the knight out of the way of the bishop, which now has nothing in front of it as far as a8. Black's quietest alternative is 1. c4 c5, the Symmetrical, where the mirror is exact and the game turns on who breaks it first: sound, dull, and hard for either side to win.
Two things beginners get wrong. First, chasing a tempo with the e-pawn: 1. c4 e5 2. Nc3 Nf6 3. Nf3 e4 4. Ng5 kicks the knight, but that pawn now stands on e4 with nothing behind it, and Black will spend more moves defending it than White spent being kicked. Second, and far more common, forgetting that 1. c4 is a transposition machine. Play d4 at any point and you may find yourself somewhere else entirely — 1. c4 e6 2. d4 d5 is just a Queen's Gambit Declined, and ...Nf6 with ...Bb4 or ...g6 gives you the Nimzo-Indian or the King's Indian instead. The English is flexible, not low-theory. Before you take it up, decide which of those openings you are content to be steered into.
The defence White can barely avoid, the price of the early queen, and where she should live
Every other defence on this page is a proposal White can decline. This one is close to unavoidable: after 1. e4 d5 2. exd5 Qxd5 the centre pawns have already been traded and Black has the structure he came for. White can sidestep the trade by pushing past with e5 or developing with Nc3 instead of capturing, but both are rare sidelines, so in practice Black reaches this position nearly every game. For someone with an hour a week to spend on chess that is worth a great deal — one setup, one set of plans, no forest of White alternatives to learn first.
The price is a tempo, and it is real. The knight to c3 hits the queen at once and Black is a move behind for the rest of the opening, so the only genuine decision in this opening is where she goes. 1. e4 d5 2. exd5 Qxd5 3. Nc3 Qa5, the square this card uses, keeps her active on the a5–e1 diagonal. 1. e4 d5 2. exd5 Qxd5 3. Nc3 Qd6 is the modern preference: harder to attack, at the cost of standing in front of the d-pawn for a few moves. 1. e4 d5 2. exd5 Qxd5 3. Nc3 Qd8 is simply honest about the lost time and sends her home.
The plan from there is close to the Caro-Kann's, and the move order is the instructive part. 1. e4 d5 2. exd5 Qxd5 3. Nc3 Qa5 4. d4 Nf6 5. Nf3 c6 6. Bc4 Bf5 7. Bd2 e6: ...c6 comes first because it gives the queen b6 and c7 to run to and denies b5 to White's pieces, and only then does the light-squared bishop step outside the pawn chain, before ...e6 shuts the door on it. White's quiet Bd2 in that line is worth understanding rather than copying — while the knight stands on c3 the bishop hits nothing, but the instant that knight moves, to d5 or b5, the bishop on d2 is attacking the queen on a5 and Black is moving her again.
What beginners get wrong is treating the early queen as a developed piece. She is a target: every move that touches her is a free move for White. The light-squared bishop can drift the same way. 1. e4 d5 2. exd5 Qxd5 3. Nc3 Qa5 4. d4 Nf6 5. Nf3 Bg4 6. h3 Bh5 7. g4 Bg6 8. Ne5 — count it: five of Black's seven moves have been made by two pieces, White has gained space on both wings, and the bishop has been herded to the edge of the board. Put the queen on a square nothing attacks, get ...c6 in, then develop everything else before you go looking for activity.
Pattern recognition
Six tactics that decide games
Most games below master level are won and lost on tactics, not strategy. Learn to spot these shapes and you'll win material out of thin air — and stop hanging your own pieces.
These six diagrams are boards you can play, not pictures. Click a piece to select it, then click the square it goes to; the card redraws and names the move it just played. Try an illegal one and the card says so in words and leaves the position exactly where it was. Reset puts a card back to the position published here. Five of the six end in a winning move you play yourself: the fork and the discovered attack are on the board right now, the skewer arrives once the checked king steps aside, and the two mates are published one ply before the finish, so you play Re8 and Nf7 yourself and the card announces the checkmate. The pin is the exception — it is a shape, not a move: try to move the pinned knight and the card refuses it in words, which is exactly the lesson.
01Fork
The knight on c7 hits the king on e8 and the rook on a8 at once. The king must step aside — then the rook falls. A knight can't be blocked, which makes its forks brutal.
Why c7 is the only square, and what the capture is actually worth
A fork is one piece attacking two things with the same move. The knight is the deadliest forker on the board for one reason: nothing can be put in its way. A rook, a bishop or a queen can always be shut out by a piece dropped on the line between it and its target. A knight jumps, so the only answers are to capture it, to move both targets, or to have seen it coming a move earlier. And c7 is the only square on the whole board that attacks e8 and a8 at once — from b6 the knight would hit the rook and miss the king, from d6 it would hit the king and miss the rook.
The cheap way to see forks coming is colour. A knight only ever attacks squares of the opposite colour to the one it stands on, so the two victims of any knight fork are always the same colour as each other. e8 and a8 are both light; c7 is dark. That gives you a one-second scan in your own games: if the enemy king and queen, or king and rook, are sitting on two squares of the same colour, go looking for a knight route to the square that touches both. If they are on different colours, relax — no single knight move can ever hit them both.
The card hands White the move so the payoff is yours to play: 1. Nxa8, and the rook is gone. In a game the check would land first and Black would get a move in between, which changes nothing here, because nothing Black owns defends a8. The one thing that can go wrong is the corner: a knight on a8 has only two squares and can be trapped there. Not in this position — 1. Nxa8 Kd7 2. Nb6+ leaves with check and walks away free.
Be honest about the scoreboard, though. This diagram holds four pieces, and after the capture White has a king and a knight against a bare king, which cannot force checkmate: play it out on the card and it is a draw however you try. The geometry is the lesson, not the result. For the defender the lesson is that the answer always comes earlier than the fork. Once a knight lands with check on a square you cannot attack, the second target has no time to run, because the check has to be answered first.
02Pin
The bishop on b5 pins the c6-knight to the king: it literally cannot move without exposing the king to check. Pin a defender, then pile up on the square it can no longer guard.
The four ways out of a pin, and why only one exists here
The bishop on b5, the knight on c6 and the king on e8 stand on one diagonal with nothing else between them, and that is the whole definition. The knight cannot legally move, because every square it could jump to would leave its own king in check. With a king behind it the pin is called absolute, and absolute means exactly what it says: Black has five legal moves in this position and every one of them is a king move. Select the knight on the card and it will refuse every square you try, in words.
There are four answers to a pin and it is worth knowing all of them before you need one: step the king off the line; block the line between the pinner and the pinned piece; capture the pinner; or defend the pinned piece, so that taking it is a trade instead of a gift. A stripped four-piece diagram only offers the first of those, and even here the choice of king move decides the game.
1... Kd7 steps along the diagonal rather than off it, so the knight stays pinned — but it is now guarded, and that is the fourth answer arriving by the back door. 1... Kd7 2. Bxc6+ Kxc6 is bishop for knight, an even trade and a dead draw. Step off the line instead and the pin is broken but the knight is not saved: 1... Kf8 2. Bxc6 simply wins a piece, because nothing was defending it in the first place. That is the trap hidden in the word — the pin is not what wins material. The attack on the pinned piece is.
Spotting pins in your own games costs a second: look down the four diagonals and the two straight lines through your king and ask what of yours is standing on them with an enemy bishop, rook or queen at the far end. Anything there is pinned whether or not your opponent has noticed. On the attacking side, the pin is never the goal — the follow-up is. Pin the piece that guards a square, then attack that square with something else. The defender cannot leave to help.
03Skewer
The bishop checks the king along the long diagonal. The king must move — and the queen sitting behind it drops. A skewer is a pin flipped: the bigger piece is in front.
The king move that saves half the queen, and the one that saves none
A skewer is a pin with the pieces the other way round. In a pin the lesser piece stands in front and is frozen where it is; in a skewer the greater piece stands in front, is forced to step aside, and whatever was sheltering behind it is collected. Same line, same attacker, opposite order — which is why one shape is a nuisance to live with and the other usually just wins material. Here the bishop on b2 checks down the long diagonal, the king on g7 is in the way and the queen on h8 is directly behind it.
Black is in check with six legal moves, every one of them a king move, and which one you choose is the entire lesson. The king has to leave the diagonal, so the queen cannot be rescued by getting out of the way — but it can be defended on the way out. 1... Kf7 walks off the line and abandons the corner, and after 1... Kf7 2. Bxh8 the queen is gone for nothing at all. 1... Kh7 also leaves the diagonal but keeps the king next door to the queen, so 1... Kh7 2. Bxh8 Kxh8 costs White the bishop in exchange. Kg8 does the same job.
Honest limit, because the card will show it to you: this diagram has four pieces on it, so every line above ends in a draw — a lone bishop and king cannot force mate any more than a lone king can. With the rest of a real game still on the board, though, the difference between losing a queen for nothing and losing a queen for a bishop is usually the difference between resigning and playing on. That is why the defender still has a decision here rather than a formality.
The tell, in your own games, is two of your own pieces on one line with nothing between them and the more valuable one nearer the enemy. King and queen on a diagonal is the classic; king and rook on an open file is the endgame version, and a great many rook endings are decided by exactly that. The defence is nearly always earlier than the shape: before parking the queen next to your king, look at what is behind it. Once the check lands there is no spare move left to fix the alignment.
04Discovered attack
The knight on e5 masks the bishop on b2. Play Nxd7 and two threats fire at once — the knight grabs the rook, and the unblocked bishop checks the king. You can only answer one.
Why all eight knight moves are check, and which one is worth playing
A discovered attack is two threats for the price of one move. The piece that moves makes its own threat, and the piece it had been standing in front of suddenly has a clear line to something. Here the bishop on b2 is already aimed down the long diagonal at the king on g7, and the knight on e5 is the curtain. Pull the curtain in any direction at all and the check appears behind it.
Count White's knight moves on the card: Nc6+, Nxd7+, Nf7+, Ng6+, Ng4+, Nf3+, Nd3+ and Nc4+ — all eight arrive with check. That is what makes a discovery so violent. The check is coming whatever the knight does, so the knight is not looking for a safe square, it is shopping. Seven of the eight take nothing. 1. Nxd7+ takes the rook, and after 1. Nxd7+ Kg8 Black has spent the move answering the check while the knight strolls off with it.
The near-miss is instructive. 1. Nc6+ Kf7 is perfectly legal, keeps the knight nearer the centre and looks active — and wins nothing, which means it has thrown away the one free move the discovery was ever going to give. A discovery is a single-use coupon. Spend it on the most valuable thing the moving piece can reach, not on the best square.
The scan runs in that order in your own games too. Look for one of your pieces standing between one of your bishops, rooks or queens and something of theirs, and only then ask what that piece could do if it simply left. Defending is a matter of not letting the alignment happen: once the king sits at the far end of an open diagonal with an enemy knight in the middle of it, there is no defence, because one move cannot answer two threats and the check must be answered first.
05Back-rank mate
Black's king is walled in by its own pawns. Re8 checks down the back rank with no escape. The antidote is luft — nudge a pawn (h6) early to give the king air.
The wall the pawns build, and what luft costs when it is late
Nothing here is clever, which is exactly why it catches so many people. Black's king sits on g8 behind three pawns it has never moved, on f7, g7 and h7. Those pawns are the wall: they were put there to keep enemy pieces out and they end up keeping the king in. A rook or a queen arriving on the empty eighth rank gives a check the king cannot step out of, because the only three squares it could step to are occupied by its own men.
The card is published one move before the finish, with the rook still on e1, so the mate is yours to play: 1. Re8#. Nothing on the board can capture the rook, nothing can interpose between it and the king along the eighth rank, and the king has no square. Play it and the card names it as checkmate.
The cure has a name — luft, German for air. Push one of those three pawns a square early and the king has a hole to step into. To watch it work, give Black the free move White would obviously never grant: 1. Re7 h6 2. Re8+ Kh7, and what was checkmate is now a check the king walks out of. That is the whole trade. Luft costs one tempo when you spend it at a quiet moment and costs the game when you leave it until the rook is already on the way.
Two habits keep this off your own board. Before trading into an endgame, check what defends your back rank other than the rook you are about to swap — a great many back-rank mates are delivered on the move after a trade that looked harmless. And count the squares your king actually has rather than the pieces around it: a king with three of its own pawns in front of it and an open rank behind it has zero, however well defended it looks.
06Smothered mate
The knight lands on f7 with check. The king is smothered by its own rook and pawns and nothing can take the knight. The prettiest mate in chess, often set up by a queen sacrifice.
Why the rook is the jailer, and the capture that throws the mate away
Smothered mate is the one checkmate delivered almost entirely by the defender's own pieces. The black king on h8 is walled in by its rook on g8 and its pawns on g7 and h7. A knight lands on f7 with check, and a knight check can never be blocked, because a knight has no line for anything to interpose on. The king has no empty square, and nothing in the position attacks f7: the rook on g8 moves along ranks and files, and the pawn on g7 captures towards f6 and h6, not forwards. So 1. Nf7# ends it. The card is published one move early, with the knight on g5, so the mate is yours to deliver.
The pattern to keep in your head is the corner plus three filled squares. When an enemy king is tucked on h8 or h1 — or a8 or a1 — with its own rook beside it on the back rank and its own pawns in front, look for a knight that can reach f7, f2, c7 or c2 with check. That set-up is not exotic — it is an ordinary castled king that has not moved a pawn yet — even though the mate itself is rare, and that gap is exactly why it is worth being able to recognise on sight.
The famous version has a traditional name, Philidor's Legacy, and it arrives one move earlier: the attacker puts a queen on g8 right next to the king, where the rook is obliged to capture it, and that capture is what plugs g8 so the knight can mate on f7. It is the most memorable idea in beginner chess literature, and it is the same mate as this one — the jailer has simply been installed by force.
The defence is the back-rank defence again: give the king air. Watch it fail by handing Black a free move — 1. Ne4 g6 2. Ng5 Kg7 and the king has walked out of the corner; the knight may visit f7 whenever it likes now and it is not even check. And the greedy move throws the win away outright: 1. Nxh7 Kxh7 wins a pawn, unplugs h7 and hands the mate back. When a mate is on the board, count flight squares first and take material second.
Converting the win
Three endgames that finish a won game
Nobody wins a game by being a queen up; they win it by mating, and a beginner who is a queen up and does not know how often ends up drawing. These three positions are the technique that gap needs. Take the queen first, because the other two end in it: the pawn card finishes by making a new queen, and the rook card is the same shape done with a piece that cannot do it alone.
These three diagrams are boards you can play, not pictures. Click a piece to select it, then click the square it goes to; the card redraws and names the move it just played. Try an illegal one and the card says so in words and leaves the position exactly where it was. Reset puts a card back to the position published here, and Play this position opens the same position on the full board so you can play it out against the computer. All three are White to move, and White is winning in all three — but in the third one, five of White's six legal moves throw the win away.
01King and queen mate
Black's king is in the corner and the queen already covers the squares it would run to. What is missing is the white king: walk it to b6 and the queen mates on the back rank next move. A queen cannot mate on her own, and that fact is the whole shape of this ending.
Why the king has to come too, and the two moves that draw on the spot
Checkmate means every square the king can reach is attacked, including the one he is standing on. In the middle of the board a king touches eight squares; on an edge he touches five; in a corner, three. A queen attacking from a distance covers three or four of those at once and never eight, so a king loose in the centre cannot be mated by a queen at all — there is always a hole. Driving him to the rim is not decoration. It is the only way to get the number of squares down to something one queen and one king can cover between them.
The reliable way to drive him there is to keep the queen a knight's move from him. From a knight's move she attacks every square beside him except the ones directly away from her, and she cannot be captured, because a knight's move is the one distance a king cannot cross in a single step. He steps back; you move the queen to a knight's move from his new square; the box shrinks by a rank or a file each time. Your own king stays out of it until the box is small, and then walks up to cover the last squares. From the worst legal starting position on the board, king and queen force mate in ten moves.
Here the driving is done and White has thirty-four legal moves. Five of them force mate next move; this is the clearest: 1. Kb6 Kb8 2. Qd8#. After the quiet king move Black has exactly one legal move in the whole position, and the mate then covers a8, b8 and c8 along the eighth rank while the king on b6 covers a7, b7 and c7. 1. Qd7 Kb8 2. Qb7# is the same length by a different route, with the queen landing on the square beside the king — which is only safe because the king on c6 is defending her.
And two of those thirty-four moves draw the game on the spot. 1. Qb6 puts the queen a knight's move from the corner, which is the right instinct one move too late: Black is not in check, and a7, b7 and b8 are all covered, so he has no legal move at all. That is stalemate, an immediate draw with a queen still on the board. 1. Kc7 does identical damage from the other side, the queen on d4 covering a7 while the king takes b7 and b8 away. Both feel like progress. Both end the game at a half point.
So carry two habits out of this card. Give check only when the check drives the king towards an edge; checks that chase him across open board give him room and cost you moves. And never assume a king move must be the safe one — Kc7 here is a king move, and it is a draw. Count the squares the enemy king has left before every move you make, and take material second.
02King and rook mate
The rook on c5 is not attacking anything. It is fencing: no king may step onto a square an enemy piece attacks, so the c-file and the fifth rank are shut and Black has six squares left to live in. Shrink that box, walk your king up, mate on the back rank.
Why the rook cuts instead of checking, and what one loose check costs
A rook covers one rank and one file — a cross. A king standing in the open always has squares off both arms of that cross, so a rook can never mate on its own; it needs the enemy king held against the edge with a friendly king taking the squares the cross misses. What a rook is very good at is fencing. Because a king may not move onto an attacked square, the rook on c5 makes the entire c-file and the entire fifth rank forbidden ground, and with the white king on d6 holding c6 and c7 the black king is left exactly six squares: a6, a7, a8, b6, b7 and b8. That rectangle is the box, and the whole method is to make it smaller without ever letting the fence drop.
The win is five moves, against the most stubborn defence Black has: 1. Rc6 Kb8 2. Rc7 Ka8 3. Kc6 Kb8 4. Kb6 Ka8 5. Rc8#. The first move slides the fence up a rank without leaving the file, and Black is down to three squares. The second takes the seventh rank as well and leaves him exactly one. Now the rook has done everything it can, so the king walks in for two moves — and the mate is the rook checking along the eighth rank, with b8 covered by the same rook and a7 and b7 covered by the king.
Notice that not one of White's first four moves is a check. That is the part beginners get backwards, because a rook check feels like progress and usually is not. 1. Rc7+ Kb6 2. Rc5 is the cautionary line: the check looks far more active than the quiet fence move, Black simply steps out past the rook, and White's best answer is to put the rook straight back on the square it started from. That check costs two moves — with best play afterwards the mate lands on move seven instead of move five. Check when the check pushes the king backwards. Cut when it does not.
Two other things to watch. Keep the rook either protected or well away from the enemy king: a rook that can be taken turns a won game into two bare kings and an instant draw. And stalemate is possible here too, exactly as it is with a queen — a black king on a8 with a white rook on b7 that a white king on c6 is defending is not in check and has no legal move, so the game is drawn. Count his squares before you commit.
One last note on the name. With two rooks this method is called the ladder, and it is quicker, because the rooks fence for each other: one cuts a rank, the other checks the king onto the next rank up, and the attacking king never has to move at all. With a single rook your own king has to supply what the second rook would have done, which is why half the moves above are king moves. From the worst legal position on the board, king and rook need sixteen moves — comfortably inside the fifty-move rule, but not inside it if you wander.
03King and pawn: the opposition
One pawn, and this position is a win or a draw depending on a single move. White has six legal moves here and exactly one of them wins. Pushing the pawn is not it — the king has to go first, and the reason is a rule about kings, not about pawns.
What the opposition actually is, and why pushing the pawn here is only a draw
Two kings can never stand on adjacent squares, and that single restriction is the whole of this ending. When the kings face each other on the same file with one empty square between them, they are in opposition, and the player who does not have to move is the one who holds it: whoever is on move has to step aside and let the other king past. So the opposition is not a square you occupy. It is a position you want your opponent to be forced to move from.
It decides pawn endings because a lone pawn cannot promote if the defending king gets in front of it and stays there — the pawn walks into the king and stops. The attacker's king has to clear the path, and the opposition is the only lever it has. In this position the kings are on d5 and e7: not opposed, White to move, which is exactly the moment to seize it.
The win, played out to the new queen: 1. Ke5 Ke8 2. Ke6 Kf8 3. Kf6 Kg8 4. e5 Kf8 5. e6 Ke8 6. e7 Kd7 7. Kf7 Kd6 8. e8=Q. The first move puts the kings on the e-file with e6 between them and Black to move, so Black must give ground; the second takes the opposition again; the third follows him sideways and takes it a third time, and now the black king has been pushed off the pawn's path for good. Only then does the pawn take a step. Look at how late its first move comes: seven of White's eight moves are made by the king or by the pawn walking in behind it.
The natural move is a draw. Push with 1. e5 Kd7 and Black has found the only move that saves him: his king reaches the square in front of the pawn before White's king can claim it, and after that nothing gets through. The four remaining king moves — Kc4, Kc5, Kc6 and Kd4 — give the win away too, because each of them lets Black take the opposition instead. One winning move out of six, and it is the quiet one.
Two honest limits, because the opposition is a tool and not a law. It settles positions shaped like this one, with one pawn and both kings close to it, and it says very little about endings with pawns on both wings, where a king can be too busy elsewhere to bother taking it. And rook's pawns are the standing exception: with a lone a-pawn or h-pawn, a defending king that reaches the corner square in front of the pawn draws however the attacker manoeuvres, because there is no room on one side of the pawn to walk around him. When the pawn does queen, the mate that follows is the first card in this row.
Counting, and knowing when not to
What the pieces are worth, and when to trade
Every chess book prints the same five numbers, and they are the most useful thing a beginner can memorise: a pawn is worth 1, a knight 3, a bishop 3, a rook 5 and a queen 9. The king carries no number at all — he is never traded, because losing him ends the game, so there is no price to compare him against. Add your own pieces up, add your opponent’s up, and the difference between the two totals is your material count. It is the number the board page keeps for you under the captured pieces while you play, and until now this site showed you that number without ever explaining what to do with it.
These three positions are what it means. One where the count changes hands in a single capture and nothing clever is required; one where a trade is how a lead that already exists gets cashed; and one where the side twelve points behind is the side that wins. All three are White to move. Like every other board on this page they are playable rather than pictures — click a piece, then the square it goes to, and Reset puts the card back to the position published here.
One thing this row deliberately does not do: it never tells you who is winning. A material count is a fact you can check by looking at the board. A verdict on a position is a different kind of claim, and the third card is here to show how far apart those two things can be.
01What the pieces are worth
The numbers every player carries in their head: a pawn is 1, a knight 3, a bishop 3, a rook 5, a queen 9. Here both armies add up to the same total, and one capture changes that — the bishop on b2 attacks the rook that has wandered to a3, and nothing defends it. Three takes five and no piece takes back.
The same five numbers are in the how the pieces move table further down this page, one to a row.
Where the five numbers come from, and the capture on this board that loses two points
Nobody measured these values in a laboratory, but they are not arbitrary either. They are a rough count of how much board a piece controls. On an otherwise empty board a knight attacks two squares from a corner and eight from the middle; a bishop attacks seven from a corner and thirteen from the middle; a rook attacks fourteen squares from anywhere at all, which is why a rook cares so much less than the others about where it stands; a queen attacks twenty-one from a corner and twenty-seven from the middle. Line those ranges up against the values and the familiar shape appears: knight and bishop are close enough to call equal, a rook is worth appreciably more than either, and a queen is worth roughly a rook and a minor piece put together.
Two things the numbers are not. They are not money, so three pawns and a knight are the same number and completely different things to own — the knight can be somewhere useful by next move and the pawns cannot. And they are not a running verdict on the game: they describe what is standing on the board, not what is about to happen to it, which is the subject of the third card in this row.
On the board above the two armies are identical — a rook, a bishop and three pawns each — except that Black’s rook is sitting on a3, on the long diagonal, with no defender anywhere near it. 1. Bxa3 and the count reads eleven to six. Nothing had to be calculated: a piece worth three took a piece worth five and no piece took back. That is the plainest way material changes hands, and it is how most games between beginners are actually decided — not by a combination, but by a piece left somewhere cheaper can reach it.
The same bishop has a second capture on the same board, and it is the mirror image. 1. Bxg7 Kxg7 wins a pawn and loses the bishop that won it, because the king on g8 is defending that pawn. The count afterwards reads eight to ten: White has gone from level to two points behind in a single move, and it felt like taking something. The difference between the two captures is not which piece does the capturing. It is whether the square is defended. So the question to ask before every capture is not “can I take it” but “what takes back, and what is it worth”.
Two honest caveats, because these are guides for trading and not laws of the game. A knight parked on a strong central square that no pawn can ever chase away is worth more than a rook with no open file to work on, and strong players trade down on the count all the time when the position pays them for it. And the bishop pair is a real exception with a number attached: having both bishops when your opponent has only one is generally reckoned worth about half a pawn on top of the three and three, because between them the two bishops cover squares of both colours and a lone bishop is permanently blind to half the board. Neither caveat helps here. On this board a rook is hanging, and hanging pieces come first.
02A trade when you are ahead
White is a rook up and Black’s queen is the only piece left that could change that. Offering to swap it off is how the lead gets cashed: an even trade leaves the gap between the two totals exactly where it was, on a board with less on it. Fewer pieces, same five points, a shorter game to play accurately.
Where this line is walking: the king and rook mate, the second endgame card above.
Why the rule is trade pieces and not pawns, and the ending where getting it backwards turns a win into a draw
A lead in material is a difference, not a total, and the difference is the only part of it that wins games. Here the difference is five, which is a rook. Play 1. Qxd7+ Kxd7 and the two totals read eight and three. Nothing has been gained — the gap is still five — and yet something real has happened, because the queen that just left the board was the one piece Black owned that could generate threats faster than White could answer them. Off the board it generates none.
That is the whole logic of trading while you are ahead. Every pair of equal pieces that comes off makes your extra piece a larger share of what is left, removes one more thing your opponent can attack you with, and shortens the stretch of accurate play you still owe. A rook up with queens on is a game. A rook up with kings, a rook and pawns is a technique, and technique is what the endgame cards above this row are for.
The rule has a second half that beginners drop, and dropping it is expensive: trade pieces, not pawns. Pawns are the only material that can turn into new material, and in a bare ending they are frequently the only reason an extra piece wins anything at all. King and rook against a lone king is a forced mate from any legal starting position, in sixteen moves at the outside. King and bishop against a lone king is a dead draw, and so is king and knight against a lone king — the mate simply does not exist, and the board page will declare the game drawn for insufficient material the moment the last pawn disappears. So a player a bishop up who cheerfully swaps off every pawn has traded a won game for half a point, and will not find out until it is far too late to stop. Being a rook up, as White is here, is the comfortable case: the bare ending is still a win. The habit is worth building on the position where getting it wrong costs nothing.
Playing the trade out a little further: 1. Qxd7+ Kxd7 2. Rb7+ Kd6 3. Rxf7. The rook takes the seventh rank with check, the black king has to step off it, and a pawn falls as well — eight to two now, on a board holding one rook, three pawns and two kings. There is nothing left to go wrong with except carelessness, which is exactly the state a player who is ahead should be steering towards.
And the honest limit, because “trade when you are ahead” is advice and not arithmetic. It says nothing whatever about whether a particular trade is sound. A swap that strips the pawns from in front of your own king, or that hands your opponent a passed pawn, or that leaves you with a bishop of the wrong colour for the corner your last pawn is heading for, is a bad trade however level it looks on the count before and after. Counting tells you the price. It does not tell you whether you wanted the thing.
03When the count is not the answer
Black is twelve points ahead — a queen, a rook and three pawns against a single rook — and Black is lost. It is White to move, and 1. Re8# is mate: the black king is walled in by its own pawns and nothing Black owns can reach the eighth rank in time to stop it or take the rook. A material count is not a verdict on the position.
The shape this position is built from: the back-rank mate, the fifth card in the tactics row above.
What a count cannot see, the three things it is still good for, and the five things it cannot price
Add the board up and Black leads by twelve. Black is also, from this exact position, dead lost, because it is White’s move and 1. Re8# ends the game where it stands. The king on g8 has three squares in theory and none in practice: f8 and h8 lie on the eighth rank the rook has just taken, and f7, g7 and h7 are occupied by Black’s own pawns. Nothing Black owns can interpose or capture either — the rook is stuck down on the second rank, and the queen’s only road to the kingside runs b3, c4, d5, e6 and then stops dead on her own pawn.
So a material count is not a verdict on the position, and it is worth saying plainly because the number is so easy to read. Counting is not judging. A count adds up what is standing on the board. It cannot see whose turn it is, where the two kings are, whether an army is coordinated or scattered across three corners, or whether one side mates next move. All four of those decide this position, and the count is looking the other way while they do it. Turn the tempo round and the same thing happens in reverse: waste one move with 1. Ra1 Qxa1# and it is White who gets mated, because every escape square on the second rank is covered by the black rook and the first rank belongs to the queen.
That is also why the cards on this page stop at counting and go no further. A count is a fact you can check by looking. A verdict on a position is a different kind of claim, and none of these cards makes one.
Three things a count is genuinely good for, and they are worth more than they sound. It settles a trade you are being offered right now — if the pieces coming off are worth the same, the trade is even, and that answers most of the questions a beginner actually faces in a game. It is a running check on your own attention: if you were level ten moves ago and you are three points down now, something went missing and it is worth finding out what before you play on. And it tells you which ending you are steering towards, which is what the previous card in this row is about.
And five things it cannot price, in roughly the order you will meet them. An attack — a rook and a bishop aimed at a castled king may be worth far more than the queen you gave up for them, or absolutely nothing, and only calculation tells you which. The exchange — a rook for a knight or a bishop is a two-point gain on paper and a poor deal whenever the knight has a square nobody can chase it off and the rook has no open file. Three pawns for a piece — level on the count, and decided entirely by whether the pawns can get moving. A passed pawn one square from promotion, which is worth one until the move it is worth nine. And king safety, which has no number at all and is the entire story of the board above.
Back to basics
How the pieces move
A quick, honest refresher. Point values are a rule of thumb for trading — a good position often beats a pawn count.
| Piece | Moves | Value |
|---|---|---|
| King | One square in any direction. Lose it and you lose the game, so it hides early and fights late. | — |
| Queen | Any number of squares in a straight line or diagonal. The most powerful piece — don't bring her out too early. | 9 |
| Rook | Any number of squares in a straight line. Loves open files and the 7th rank. | 5 |
| Bishop | Any number of squares diagonally — forever on one colour. The pair is worth keeping. | 3 |
| Knight | An L-shape: two squares then one across. The only piece that jumps. Strongest in closed positions. | 3 |
| Pawn | Forward one (or two from its start), captures one square diagonally. Promotes on the far rank. | 1 |
Castling
Once per game, slide your king two squares toward a rook and hop the rook to the king's other side. Only if neither piece has moved, the squares between are empty, and your king isn't in, through, or into check.
The four conditions, and the one almost everybody gets wrong
Castling is the only move in chess that shifts two of your own pieces, and the only time a king travels more than one square. The king goes two squares toward a rook and the rook hops over it, landing on the square the king crossed. Short castling, written O-O, puts the white king on g1 and the rook on f1; long castling, O-O-O, puts the king on c1 and the rook on d1. Black's version is the same geometry at the other end of the board — g8 and f8, or c8 and d8. It counts as one move, and you make it by moving the king: the rook is never touched, and on the board below it travels on its own.
Four conditions have to hold at the same time. One: neither the king nor the rook you are castling with has moved at any point in the game. Moving one away and back does not restore the right, and the right is lost per rook — give up the h1 rook's right and the a1 rook's is still intact. Two: every square between king and rook is empty. Short castling needs two squares clear, f1 and g1; long castling needs three, b1, c1 and d1, which is why it is blocked far more often. In the position below the white queen sits on d2 rather than d1, and that is exactly what leaves the long side open. Three: your king is not in check right now. Four: the king may neither cross nor land on a square an enemy piece attacks — f1 and g1 for the short castle, d1 and c1 for the long one.
The mistake almost everyone makes is to apply condition four to the rook's journey as well. It does not reach that far. b1 has to be empty, but it does not have to be safe: if an enemy rook or bishop covers b1, long castling is still completely legal, because it is the rook that crosses that square and a rook is allowed to be attacked. The rook you castle with may itself be under attack as you castle. Two smaller traps go with it. You can never castle out of check, so castling is never the answer to a check. And condition three is about your king alone — an attacked queen, rook or knight elsewhere on the board does not stop you.
The board below is a mirror: both sides hold the identical pieces on identical squares, both keep every castling right, and White is not in check, so both castles are legal here and so are Black's. Click the king on e1 and then g1 and watch two pieces move for one click pair — the king arrives on g1, a white rook appears on f1, and e1 and h1 are left empty. Press Reset, then click e1 and c1 instead: king on c1, rook on d1, a1 empty. Nothing else on the board shifts either time, and the pawn wall on f2, g2 and h2 that the short castle tucks the king behind is the whole reason the move exists.
Do it: White to move, both back-rank gaps clear. Click the king on e1, then g1 for the short castle or c1 for the long one — the rook moves by itself.
En passant
If an enemy pawn springs two squares and lands beside yours, you may capture it as if it had moved one — but only on the very next move.
Why the board seems to eat a pawn, and how wide the window really is
En passant is French for "in passing", and it exists to close a loophole. Pawns were once allowed only one square at a time; when the two-square first move was added, a pawn could use it to sprint straight past an enemy pawn that would have been able to capture it had it stepped one square. The rule takes that back. If an enemy pawn moves two squares and lands directly alongside one of yours, you may capture it exactly as if it had moved one: your pawn steps diagonally onto the square it skipped, and the enemy pawn is lifted off the square it is actually standing on. It is the only capture in chess where the piece you take is not on the square you land on, which is why it looks like a glitch the first time it happens to you.
Three things have to be true. Your capturing pawn must be on its fifth rank — rank 5 for White, rank 4 for Black — because that is the only rank where the geometry lines up. The enemy pawn must have just played a two-square move and landed on a file directly beside yours. And you must take immediately: the right lasts exactly one move. Play anything else first, even a check, and that pawn is safe on that square for the rest of the game. Taking is optional, not forced — on the board below White's other legal reply to the spring is to push e6, walking past the black pawn and letting the moment go.
Two beginner errors sit either side of the rule. The first is expecting the window to reopen later; there is no way to get it back. The second is the mirror image — assuming that any enemy pawn arriving beside yours can be taken this way. It cannot. Reset the board below and play d6 instead of the two-square spring, and White's pawn can still capture with exd6 — but that is an ordinary diagonal capture onto an occupied square, it is not en passant, and unlike the real thing it stays available for as long as that black pawn sits there.
Read the demo as two click pairs. Black moves first: click the pawn on d7, then d5, and it springs past White's pawn on e5 and lands beside it. Now click e5 and then d6. The white pawn appears on d6, behind where the black pawn stood, and d5 is left empty — that empty square is the whole rule in one picture. To watch the window shut instead, press Reset, play d7 to d5, then make any king move with White; after Black replies, exd6 is refused and the board does not budge.
Do it: play Black's pawn d7 to d5, springing past the white pawn. Then take it with e5 to d6 — the white pawn lands behind it and the black pawn comes off.
Promotion
Push a pawn to the far rank and it becomes any piece you choose — almost always a queen. Yes, you can have two.
Why the choice is free, and the one position where a queen is the wrong answer
A pawn that reaches the far rank — the eighth for White, the first for Black — stops being a pawn on that same move. You take it off and put a queen, a rook, a bishop or a knight of your own colour on the square instead, and you choose which. Promotion is compulsory: there is no option to leave a pawn sitting on the eighth rank, and a king is the one piece you may not choose. The new piece is live the instant it lands, so a knight that promotes onto a square next to the enemy king is giving check immediately, before the opponent gets a move.
The choice is genuinely free and has nothing to do with what has been captured. If your queen is still on the board, you now have two; in theory you could have nine. Beginners often think they have to lose a queen first, or that they somehow pick the captured piece back up off the side of the table — neither is a rule anywhere. Almost every promotion should be a queen, and the honest advice is to take the queen unless you can name the concrete reason not to.
There are two real reasons. A knight reaches squares a queen never can, so a knight promotion that arrives with check, or that forks king and queen, is occasionally worth more than the strongest piece on the board. And a rook is sometimes right when a new queen would stalemate a cornered king on the spot: a rook covers fewer squares, and here that is the point. The position below shows the trap from the other side. c8=N is a legal move in it and the engine will accept it — but a king and a knight cannot force checkmate against a lone king, so that choice turns a won game into a dead draw. c8=B is the same mistake. Only c8=Q and c8=R keep the win.
In the demo the white pawn stands on c7 with the square in front of it clear. Click c7, then c8. A white queen appears on c8 and c7 is empty: the pawn was replaced, not moved. This board always takes the queen, so it never asks which piece you want; a real game does. Notice too that the new queen does not check the black king on h4. That is normal — a promotion is not a check by default, and expecting one is a common way to miscount a pawn race by a move.
Do it: push the white pawn c7 to c8 and watch what stands there afterwards. This board always takes the queen; a real game lets you pick rook, bishop or knight instead.
Checkmate vs. stalemate
Checkmate: the king is attacked and has no legal escape — game over. Stalemate: the side to move is not in check but has no legal move — it's a draw.
The single question that separates them, and why this ending is drawn however White plays
Checkmate and stalemate look identical from where you are sitting: it is your turn and you have no legal move. One question separates them, and it is the only question — is your king attacked right this second? Attacked, with no legal move, is checkmate, and you have lost. Not attacked, with no legal move, is stalemate, and the game ends there as a draw, half a point each, no matter what is left on the board. A player a queen and two rooks ahead who stalemates the opponent takes the same half point as the player who was stalemated.
Two details get missed. First, stalemate is not a king-only rule: the stalemated side can still own pawns and pieces, provided every one of them is blocked or pinned and the king has nowhere legal to step. Second, "no legal move" is the entire test on the checkmate side too — a king in check that can capture the checking piece, block the line, or walk away is simply in check, not mated. There is also no requirement to announce a check out loud; that is a convention between friends, not a rule.
The expensive mistake is a won game thrown away in its last five moves: a huge material lead, the enemy king herded to the edge, every escape square quietly taken away — and no check. The habit that prevents it is one question asked before every move in a winning ending: if I play this, does my opponent still have a legal move? If the answer is no, then that move had better be check, or it should not be played. Driving the king to the edge is right. Sealing it in without checking it is how half a point disappears.
The board below is more interesting than its label. Black's king is on a8 with White's pawn on a7 right underneath it, and White's king on b5 has eight squares to choose from. Play the one the card asks for, b5 to b6, and Black is stalemated on the spot: b7 is covered by the king that just arrived, b8 is covered by White's own pawn on a7 — pawns capture diagonally, so a7 guards b8 — and a7 itself is now defended by the king, so it cannot be taken. The pawn has no move either, because a8 is occupied. Then look at the alternatives. Ka6 stalemates Black just as instantly, and each of White's six other king moves leaves a7 undefended, so Black replies Kxa7 and king against king is a dead draw. Every legal move White owns here draws. So the stalemate here costs White nothing — the position was already drawn, which is what a rook's pawn is worth once the enemy king is sitting in front of it.
Do it — the stalemate half. Black's king is stuck on a8 behind White's a7 pawn, and White's pawn cannot move either. Step the white king b5 to b6 and Black has no legal move and no check.
The three draws the board declares for you
Stalemate is the draw everyone hears about first, and the “Checkmate vs. stalemate” card covers it. It is not the only one. The board on the play page ends games with three more, and it ends them itself — it never waits to be asked — so the first one you meet arrives as a verdict nobody taught you. Here are those three, in the order the board tests them: repetition, then material, then the counter. A position that satisfies two is announced under the earlier name.
Threefold repetition
The same position on the board three times and the game is over. The same position, not the same moves — and the three appearances need not be consecutive.
What counts as "the same position", and why perpetual check ends up here
This rule counts positions, not moves. If an identical position stands on the board three times during a game, the game is drawn, and the three appearances do not have to be in a row or arrive by the same route. It takes four moves each from the opening position to prove it: 1. Nf3 Nf6 2. Ng1 Ng8 3. Nf3 Nf6 4. Ng1 Ng8. The starting array has now stood there three times — before White's first move, after 2...Ng8 and after 4...Ng8 — and our board ends the game "Draw — threefold repetition" the instant that last knight lands. Nothing was captured, nobody agreed to anything, and neither player claimed anything.
"The same position" is stricter than "the same picture", and four things have to match: every piece on every square, the same side to move, the same castling rights, and the same en passant capture available. Castling rights are the one that bites. Play 1. Nf3 Nf6 2. Rg1 Rg8 3. Rh1 Rh8 4. Ng1 Ng8 instead and every piece is back on its starting square with White to move for the second time — but it is not the same position, because those rook trips destroyed both sides' right to castle kingside and that right never returns. The position's castling field went from KQkq to Qq, so the count for the picture you are looking at stands at one, not two, and the engine is right to play on.
Perpetual check is not a rule of its own; it is this rule. A player who is lost but can check the enemy king forever draws, because the checking sequence repeats a position three times. 6k1/7p/7P/4Q3/r2r4/q7/8/5R1K w - - 0 1 is that idea in miniature: White is a whole rook down and losing in any normal sense, and White draws with 1. Qe6+ Kh8 2. Qe5+ Kg8 3. Qe6+ Kh8 4. Qe5+ Kg8. Black is not cooperating — at each of those four checks the king move shown is Black's only legal move, because g7 is covered by the white pawn on h6, f8 by the rook on f1, and h7 is blocked by Black's own pawn. After 4...Kg8 the diagram position has appeared three times and the extra rook is worth nothing.
So: if you are winning, treat the second appearance of a position as an alarm. Improve a piece, push a pawn, change something — repeating once to gain time on the clock or to see the position again is normal and legal, and repeating a third time by accident is how won games evaporate. If you are losing, perpetual check is the commonest save in club chess: hunt for a check the enemy king cannot walk out of, then count that your checks last longer than its shelter takes to improve. One honest difference from over-the-board play: FIDE lets a player claim the draw at the third occurrence and makes it automatic at the fifth. Our board does not wait to be asked. It stops at three, so you cannot play on hoping your opponent fails to notice.
Insufficient material
Neither side has enough force left to checkmate by any legal sequence at all, so the game stops where it stands — usually on the capture that took the last real piece off.
The exact material that ends it, and the two endings that surprise people
This one is not about who is winning; it is about whether checkmate is reachable. The board tests it after every single move, so it fires the moment the material drops below what a mate needs. Four balances trigger it here: king against king, 8/8/8/4k3/8/8/4K3/8 w - - 0 1; king and bishop against king, 8/8/8/4k3/8/8/4K3/7B w - - 0 1; king and knight against king, 8/8/8/4k3/8/8/4K3/6N1 w - - 0 1; and one bishop each when both stand on squares of the same colour, 8/8/8/3bk3/8/8/4K3/7B w - - 0 1, where d5 and h1 are both light squares.
What those four share is stronger than "hard to win". There is no arrangement of those pieces — none, however the loser helps — that is checkmate, so no legal series of moves can reach one. That is why the rule can be applied automatically instead of left to a claim. Note what is missing from the list: a single pawn. 8/8/8/4k3/8/8/4K1P1/8 w - - 0 1 is not drawn here and never will be on material, because that pawn can become a queen. While one pawn survives anywhere, this rule cannot apply.
Two cases surprise people, and both stay alive. Two knights and a king against a lone king is not on the list: 8/8/8/4k3/8/8/4K3/5NN1 w - - 0 1 plays on. Two knights can mate, but only if the defender walks into it, and no one can force that, so the position is a theoretical draw that is still legally a game. What actually ends it is the counter in the next card. One bishop each on opposite colours is not on the list either: 8/8/8/3bk3/8/8/4K3/6B1 w - - 0 1 plays on, because a mate can be constructed — here is one, 6bk/8/7K/8/8/8/1B6/8 b - - 0 1, where Black is checkmated by a dark-squared bishop on b2 while holding a light-squared bishop on g8. "Can be reached" is the test, not "can be forced". Set that white bishop on h1 instead of g1 and both bishops are light-squared and the same board calls the draw at once — and since a bishop can never change square colour, which of the two positions you are in was settled the moment those bishops were the last pieces left.
The practical use is defensive, and it is worth real points. If you are losing, steer at it: trading your last rook for your opponent's last rook when the only survivor is a knight converts a loss into half a point, and giving a bishop for the last enemy pawn does the same job. The trade that matters is whichever one removes the final pawn, because until then no material list applies. If you are winning, the mirror habit is not to grab a piece just because it is free — count what would be left first, and keep a pawn on the board if the alternative is a bare bishop.
The fifty-move rule
Fifty moves by each side with no capture and no pawn move, and the game is drawn. Any capture by either side, and any pawn move by either side, puts the counter back to zero.
What resets the counter, where to read it, and the endings that really do run out
"Fifty moves" means fifty by each player — a hundred half-moves, which is the number this board counts. The counter is shared between you: it advances on every half-move by either side, and exactly two events reset it to zero. A capture of anything by anyone, and a move of any pawn by anyone. Nothing else. Not a check, not castling, not a repeated position, not a draw offer declined. Promotion resets it because a promotion is a pawn move, which is why a pawn race never gets anywhere near the limit.
You can read the counter directly: the fifth field of a position's FEN is the number of half-moves since the last capture or pawn move. 8/8/8/4k3/8/8/R7/4K3 w - - 99 120 sits one half-move short — that 99 is the counter. White is a rook up with a forced mate a few moves away, and if White plays the aimless Ra3 the counter reaches 100 and the board ends it "Draw — fifty-move rule" with the extra rook still standing there.
Which endings genuinely run out? King and rook against a lone king is mate in at most sixteen moves with best play, so the limit is never the problem: lose that one on the counter and the honest diagnosis is that you were not making progress. King, bishop and knight against a lone king needs up to thirty-three moves and a specific technique most players have never learned, and it is the classic ending where fifty is not enough for someone improvising. And king with two knights against a lone king — the ending the card above leaves legally alive — cannot be forced at all, so the fifty-move rule is precisely how it ends.
That gives both players something to do. Defending an ending that looks lost, count: every half-move that is not a capture and not a pawn move is a half-move closer to half a point, and the worst thing you can do is push a pawn out of impatience and hand your opponent fifty fresh moves. Attacking, make the counter part of the plan — if you still have a pawn, a pawn move at the right moment buys fifty more, and if you have none, know roughly how many you have left before you start manoeuvring. One more difference from over-the-board rules: FIDE lets the defender claim at fifty and makes the draw automatic at seventy-five. Our board does not wait for a claim; it ends the game at fifty.
Questions
Chess, briefly answered
What's the best opening for a beginner?
Start with the Italian Game (1.e4 e5 2.Nf3 Nc6 3.Bc4) or the London System (an easy d4 setup). Both give you clear, repeatable plans instead of memorised traps, so you actually understand your positions.
How do the chess pieces move?
The king moves one square any direction; the queen any distance straight or diagonally; rooks straight; bishops diagonally; knights in an L and they jump; pawns forward one (two on their first move) and capture diagonally. See the “How the pieces move” table above for the details. If your browser runs scripts, the board under that table marks every move the rules engine allows whichever piece you pick.
What is castling and when can I do it?
Castling tucks your king to safety and activates a rook in one move. It's legal only if neither the king nor that rook has moved, the squares between them are empty, and your king is not in check, does not pass through an attacked square, and does not land on one.
What is en passant?
If an opponent's pawn advances two squares and lands directly beside one of your pawns, you may capture it on the very next move as though it had moved only one square. Miss the chance and it's gone.
What's the difference between checkmate and stalemate?
In checkmate the king is under attack and has no legal way out — that side loses. In stalemate the side to move is not in check but has no legal move to make — the game is an immediate draw. Watch for it when you're winning.
How do I get better at tactics?
The six tactic boards further up this page are playable, not pictures: click a piece, then click the square it goes to. Five of the six end in a winning move you play yourself — the fork and the discovered attack are there straight away, the skewer arrives once the checked king steps aside, and the back-rank and smothered mates are published one move short so you deliver Re8 and Nf7 — while the pin is a shape rather than a move: try to move the pinned knight and the card refuses it in words, which is the lesson. Try a wrong move on any of them and watch it get refused without the position moving. Then, in a real game, ask one question before every move: what is my opponent threatening? That single habit prevents most blunders.
Is DankChess free?
Yes. DankChess is a free, static study board with no account, no ads and no tracking wall. Pick an opening, step through it, and go. The rules engine runs in your browser, so the openings, the tactics and a game against the computer involve no server round-trip. The one exception is Play a friend on the board page: those moves travel through our relay so they can reach the other person's browser, and the relay sees which two addresses paired and when. Chat in that mode goes browser to browser instead, so each player's browser learns the other's IP address.