Outposts · For club players · 14 min read
Outposts: Kasparov's Octopus Knight v Karpov, Checked by Stockfish
Kasparov's knight on d3 in game 16 of the 1985 match, what Stockfish says about it, and how to build an outpost no pawn can attack.
- plan
- outpost
- weakness
- The knight's reach
In short
- Kasparov beat Karpov in game 16 of the 1985 match with a knight on d3 that no pawn could ever attack.
- The outpost was built by a pawn sacrifice at move 8, which removed the white pawns that would have guarded d3.
- Stockfish says 16...Nd3 was Black's second-best move and that the decisive error was 17.Nab1, not the knight move itself.
- White could have given back the extra pawn with 17.d6 or later challenged the knight with 24.Nb2.
- To use the idea: count the enemy pawns that can ever attack a square before you plan a piece for it.
How we know what we say
- Verified From the record: Stockfish's numbers, the players' own notes or interviews, or a cited source.
- Inferred Read from the moves and the position: a reasoned reconstruction, not the player's words.
- Speculative A guess about mood, intent or practical motives: plausible, not proven.
Part 1The setting: game 16 of the 1985 world championship match
Moscow, 15 October 1985. Anatoly Karpov, the reigning world champion, went into game 16 of the rematch with the 22-year-old challenger Garry Kasparov: the match was level at 7½–7½ after fifteen games. Kasparov won this game, took the lead, and won the match 13–11 to become the youngest world champion in history. [1] Verified
| Fact | Detail |
|---|---|
| Game | Karpov (White) v Kasparov (Black), Moscow 1985, game 16 |
| Score before | 7½–7½ |
| Result | 0-1, Karpov resigned after 40...Re1+ (80 plies) |
| Opening | Sicilian: Kasparov's gambit 8...d5 |
| Strategy | An outpost no pawn can attack |
The question to hold while reading: Black is a pawn down and has no obvious way to win it back. What does Black do with the free move?
Part 2The idea at a glance: what is an outpost?
An outpost is a square, deep in the opponent's half of the board, that no enemy pawn can ever attack, and where one of your pieces is protected by a pawn or another piece. A piece placed on it cannot be driven out by a pawn, so the only way to remove it is to trade a piece for it. Knights benefit most because they cannot be hit by a pawn and do not need open lines. Inferred
- outpost
- target
- Pawnless squares
- pawn chain or line of force
- target
- key square
- Pawns that kick it
A true outpost needs two things: no enemy pawn can ever attack the square, and the knight is supported. Take either away and the knight is only a target.
In this game the square is d3, and the reason nobody can attack it is a decision made earlier: the pawns that would guard it are gone. An outpost is usually the consequence of a structure, not a lucky square. Inferred
Part 3The opening: a pawn for a square
The game is a Sicilian in which Black answers 8.Na3 with the pawn sacrifice 8...d5, known as Kasparov's gambit. The pawn is given up so that White's c- and e-pawns disappear in the exchanges that follow; Chess.com's Stockfish study of the gambit examines the opening phase. [3] Verified
5.Nb5 d6 6.c4 Nf6 7.N1c3 a6 8.Na3 d5 9.cxd5 exd5 10.exd5 Nb4 +0.38White slightly better
The Sicilian with Kasparov's pawn sacrifice 8...d5: White ends a pawn up with a passed pawn on d5 and no c- or e-pawns; Black has the bishop on c5 and a knight aiming at d3. Inferred
- plan
- key square
- target
- No pawns to guard d3
The result is a structure in which White has one extra pawn, a passed d-pawn on d5, and nothing on the c- or e-files. The pawn is real, but the price is a square: d3, where a black knight can sit forever. The pieces would need to be traded off to challenge it. Inferred
Whether the gambit is objectively sound is a different matter: Stockfish's verdicts on the opening moves are close to level, and Chess.com's analysis says the refutation attempts do not give White much. What matters here is the structure it creates. [3] Verified
Part 4The plans: pawn power against piece power
From the position after 15...b5 both sides have a plan. White's is to use the passed d-pawn: push it to d6, activate the bishop on f3 along the long diagonal and use the extra pawn. Black's is to put a piece on a square White's pawns cannot reach and make the extra pawn irrelevant. Inferred
- plan
- outpost
- The pawn's road
- plan
- key square
- What the knight hits
The same position, two plans. White wants to push the d-pawn; Black wants a knight on d3. Whoever carries out the plan first sets the terms.
Part 5The critical moments
Four decisions shaped the game. The two outpost moves (16...Nd3 and 17.Nab1) come first and last in the story, so the 17th move gets its own section next; here are the other three, in order, because they show what the knight does to White's decisions. Inferred
Critical moment 1 of 4 · Move 16, Black to play
Where does Black's knight belong?
- weakness
- target
Black is a pawn down, and White's rooks and queen are lined up behind the d5 pawn. Which setup gives Black a lasting grip instead of a short-term fight over d5?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 16...Qd6 17.Bh4 Rad8 | −0.30about equal | Stockfish's choice by a hair: about two-tenths of a pawn better for Black, with the queen eyeing the d-pawn. Verified |
| 16...Rac8 | 0.00about equal | Level: the rook takes the c-file, but Black gains no lasting grip. Verified |
| 16...h6 17.Bh4 Nd3 | −0.14about equal | Slightly better for Black, about a seventh of a pawn: it makes luft but does not deal with White's d-pawn. Verified |
What Kasparov played
16...Nd30.00about equal
Kasparov played the knight from b4 to d3. It cannot be attacked by a pawn, it is guarded by the bishop on f5, and from there it hits the squares b2 and f2 and ties down White's rooks and queen.
Stockfish preferred 16...Qd6 (−0.23); the game move gave up about 0.2 pawns. Depth 27.
What it revealed
The choice shows a plan built around one square rather than one move: Black spent a pawn to remove White's c- and e-pawns, and the knight now uses the hole that the sacrifice created.
What Stockfish adds Stockfish rates 16...Qd6 a shade better (about a quarter of a pawn for Black) and 16...Nd3 as level. The knight move is sound, and it asks harder questions of White.
- Verified The knight on d3 is the move that gave the game its name: Kasparov later compared its power to the tentacles of an octopus.[4]
- Inferred With White's c- and e-pawns gone, the only White pieces that can challenge d3 are the queen, a rook, the bishop (from e2) or a knight (from b2 or c1), and each of those trades leaves Black the better structure.
Critical moment 2 of 4 · Move 24, White to play
The last chance to challenge the knight
- key square
- The octopus's reach
Black has pushed pawns on both wings and the knight on d3 still dominates. White cannot attack it with a pawn. Which piece move at least offers to remove it?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 24.Nb2 24...Nde5 25.Bg2 Qf6 | −2.90Black winning | Stockfish's choice: White is still about three pawns down, but the fight for d3 goes on. Verified |
| 24.a3 24...Nde5 25.Bg2 | −3.64Black winning | Second best, about three and a half pawns down: the knight is left alone for another move. Verified |
| 24.Qc2 | −12.80Black completely winning | Loses heavily, more than twelve pawns down. Verified |
What Karpov played
24.Bg2−4.39Black completely winning
Karpov retreated the bishop to g2 and left the knight alone.
Stockfish preferred 24.Nb2 (−2.30); the game move gave up about 2.1 pawns. Depth 22.
What it revealed
A passive move adds nothing to the fight over d3: the bishop's retreat keeps the long diagonal but does nothing about the one piece that is ruining the position.
What Stockfish adds Stockfish prefers 24.Nb2 by about two pawns: White stays roughly three pawns down, but 24.Bg2 leaves a deficit of more than four and a half. The defence needed a plan, not a waiting move.
- Inferred Nb2 attacks d3 with a piece Black cannot ignore: the knight on d3 must trade itself off or leave.
- Speculative By now Karpov may have felt that trading the knight meant admitting the position was lost; the record does not say so.
Critical moment 3 of 4 · Move 30, White to play
A pawn move that opens a diagonal
- target
- The g1-a7 diagonal
White's d6 pawn is attacked by the queen and Black's g-pawn is on g4. White to play: which move handles both without opening lines near the king?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 30.h3 30...h5 | −5.23Black completely winning | Stockfish's choice: still more than five pawns down, but White keeps the king covered and the d6 pawn defended a little longer. Verified |
| 30.Qa2 | −5.55Black completely winning | About six pawns down: the queen move ignores the g4 pawn and the weak diagonal. Verified |
| 30.h4 | −5.78Black completely winning | About five and a half pawns down: slightly worse than h3. Verified |
What Karpov played
30.f3−7.12Black completely winning
White played the pawn to f3, which loses the d6 pawn and opens the diagonal to the white king.
Stockfish preferred 30.h4 (−5.40); the game move gave up about 1.7 pawns. Depth 22.
What it revealed
The pawn move looks like it challenges the g4 pawn, but it weakens e3 and d4, and the black queen lands on d4 with check.
What Stockfish adds Stockfish scores 30.f3 as a further loss of nearly two pawns. Even the best move, 30.h3, leaves White more than five pawns down: the point is how many more ways the knight offered Black.
- Inferred After 30.f3 Qxd6 31.fxg4 Qd4+ the queen reaches d4 with check; without the f2 pawn nothing covers the diagonal to g1, and White's queen on d2 cannot cover d4 because the knight on d3 blocks its file.
Part 6The turning point: 17.Nab1
Chronologically the decisive moment comes right after the knight lands, which is why it is treated apart from the others. Inferred
Critical moment 4 of 4 · Move 17, White to play
Give the pawn back, or block the knight in?
- target
- key square
- Hit by the knight
The knight has landed on d3 and cannot be driven out by a pawn. White is still a pawn up. Which plan keeps White afloat: hold the extra pawn, or hand it back to break the bind?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 17.d6 17...Qxd6 18.Bxf6 Qxf6 19.Bxa8 | −0.03about equal | Level: the pawn goes back and White keeps a sound position. Verified |
| 17.b4 17...Bxb4 18.Nc2 | −0.31about equal | About a third of a pawn behind: playable, it also breaks Black's grip a little. Verified |
| 17.Nc2 17...Nb2 | −1.18Black clearly better | About a pawn behind: it challenges the knight but allows Nb2 trades that favour Black. Verified |
| 17.Be3 | −1.30Black clearly better | About a pawn and a half behind: the bishop leaves the long diagonal. Verified |
What Karpov played
17.Nab1−2.66Black winning
Karpov moved the knight a3 to b1, planning to reroute it and cover d2 and c3.
Stockfish preferred 17.d6 (0.00); the game move gave up about 2.7 pawns. Depth 24.
What it revealed
The move holds the extra pawn and keeps the pieces together, but it commits the knight to a slow route and lets Black choose the next phase. The engine says this was the decisive mistake.
What Stockfish adds Stockfish: 17.Nab1 drops from level to nearly three pawns behind, while 17.d6 keeps the position level. This single move cost close to three pawns.
- Verified Chess.com's Stockfish analysis names 17.Nab1 as White's critical error, calls it too passive, and says White needed to give back the pawn with d5-d6 on move 16 or 17.[3]
- Inferred Nab1 keeps the extra pawn and keeps White's pieces away from the knight, which is its appeal and its flaw.
- key square
- The d-file
The point of 17.d6 is that it breaks the bind before it forms. After the trades White has given back the extra pawn, but the black knight no longer has a steady diet of weaknesses to attack, and Stockfish finds the position balanced. Verified
- threat
- key square
- target
34.Qxd3 Nf2+ 35.Rxf2 Bxd3 36.Rfd2 Qe3 37.Rxd3 Rc1 38.Nb2 Qf2 39.Nd2 Rxd1+ 40.Nxd1 Re1+ −#2Black mates by force
The octopus is removed, but at the price of the queen. Black's pieces take over every square White has left. Verified
Part 7What White could have done
There were two chances. The first, at move 17, was to return the pawn with d6. The second, much later, was to challenge the knight with a piece: the knight on d3 cannot be attacked by a pawn, but it can be attacked by a knight. Inferred
- plan
- key square
24.Nb2 Nde5 25.Bg2 Qf6 26.Kh1 Qb6 27.Nxd3 Nxd3 −2.51Black winning
Challenging the knight with Nb2 forces Black to find a retreat or trade, and 27.Nxd3 Nxd3 removes the octopus. Verified
Part 8A different angle: the outpost was built eight moves earlier
The usual account credits Kasparov's 16...Nd3. Stockfish's account is different in two ways. First, 16...Nd3 is level on the engine's scale and 16...Qd6 is a shade better, so the big swing is White's reply, not Black's move. Second, the outpost was paid for at move 8, when Black gave up a pawn to remove the c- and e-pawns. By move 16 the hole existed; the knight only occupied it. Verified
| Moment | What happened | Stockfish's view |
|---|---|---|
| Move 16 | 16.Rad1 Nd3 | Level; 16...Qd6 a shade better for Black |
| Move 17 | 17.Nab1? | White drops from level to nearly three pawns behind; 17.d6 stays level |
| Move 24 | 24.Bg2? | About two pawns more lost; 24.Nb2 was the fight |
| Move 30 | 30.f3? | Nearly two pawns more lost; 30.h3 was the best try |
| Move 38 | 38.Nb2 | Black has a forced mate |
This changes the lesson. An outpost is not a thing you land on at the last moment; it is a structure you choose, and defending against one is a structural decision as much as a tactical one. Inferred
Part 9How to use this in your own games
The recipe
- Find the target squareLook for a square on the opponent's side of the board, and check which pawns could ever attack it.
- Check the supportThe piece needs support from a bishop or a pawn so a trade does not simply win it.
- Plan the routeCount the moves a knight needs to reach the square; keep the moves you need unforced.
- Add the second weaknessOnce the knight sits there, open a second front so the defender cannot defend everything.
- Pawns first: can any pawn ever attack the square?
- Is the piece supported by a bishop or pawn?
- What does the piece attack from there?
- What is the opponent's counterplay while I spend moves on it?
Your turnPractise it
Set each position up, decide your plan before you look, then play it out against Stockfish to see whether it holds.
White to play (move 16): Black's knight on b4 wants to land on d3. Which move deals with it best?
Hint
Think about the extra pawn on d5. Is it better as a pawn or as a bargaining chip?
A good answer
16.d6! gives the pawn back and breaks the structure the knight needs; the quiet 16.Rad1 allows Nd3.
White to play (move 17): the knight is on d3. What is the most resilient move?
Hint
There is no way to attack the knight with a pawn.
A good answer
17.d6! The pawn returns and the position stays balanced; 17.Nab1 drops two pawns.
White to play (move 24): the knight on d3 still dominates. Which piece move fights back?
Hint
Which piece can attack d3 without being exchanged for nothing?
A good answer
24.Nb2, challenging the knight; the quiet 24.Bg2 loses more.
Black to play (move 15): White's knights are on a3 and c3. Which pawn move stops the knight reaching c4 and prepares the plan?
Hint
It gains space and takes a square away.
A good answer
15...b5, which stops Nc4 and keeps the queenside pieces active; Stockfish agrees.
Part 10What to take from this game
- An outpost is a structure, not a square: count the enemy pawns that could ever attack it.
- Sacrifices can create outposts: Black gave a pawn to remove two White pawns.
- Attack the piece, not the pawn: if no pawn can hit it, use a piece.
- Passive moves lose to outposts: 17.Nab1 and 24.Bg2 let the knight stay.
- Give back the pawn when it breaks the bind.
Questions readers ask
What is an outpost in chess?
An outpost is a square in the opponent's half of the board that no enemy pawn can attack, where a supported piece cannot be driven away by a pawn. Knights are the best occupants because they do not need open lines.
What is Kasparov's octopus knight?
It is the black knight on d3 in Karpov-Kasparov, game 16, 1985. Kasparov compared its reach over the white pieces to the tentacles of an octopus. It could not be attacked by a pawn and tied down White's rooks and queen.
Was 16...Nd3 the best move?
Stockfish rates 16...Qd6 a shade better, but 16...Nd3 is level and creates much harder problems for White. The decisive error was White's 17.Nab1, not the knight move itself.
How do you fight an outpost knight?
Attack it with a piece, since no pawn can, or change the pawn structure so a pawn can reach the guarding squares. Here White could have returned the extra pawn with d6, or challenged the knight with Nb2 on move 24.
How do you create an outpost for your own piece?
Look at the opponent's pawn structure and find squares no pawn can ever attack, usually after the opponent's neighbouring pawns have been exchanged. Then plan a route for a knight and a support for it before you commit.
Sources and method
The game score was checked against two independent sources. Every position, line and candidate in this article was replayed for legality, and every engine number was produced by Stockfish 19 WASM Multithreaded for this article (depth 20 to 38, deepest at the critical moments), not copied from elsewhere. What the players thought is reconstructed from the moves, their published notes and later analysis; it is labelled as verified, inferred or speculative, and nothing is presented as a quotation unless a source is cited for it.
Read next
Kramnik, the new world champion, turned the two bishops into a win at Wijk aan Zee 2001. Stockfish 19 shows what the pair was worth and where both players let the advantage slip.
Bishop vs Knight: Fischer's Famous Exchange, Checked by StockfishIn game 7 of the 1971 Candidates final, Fischer traded a dominant knight for a bad bishop and won a textbook bishop-against-knight ending. Stockfish 19 says the trade was not best. Here is why it worked anyway.
Prophylaxis: Karpov's 24.Ba7 Against Unzicker, Checked by StockfishKarpov's quiet bishop move at Nice 1974 is the textbook case of prophylaxis. Stockfish 19 says the move was the only good one, and that the game was decided later, by Black's 31...g5.
Turn reading into skill
Strategy sticks when you use it. Practise real positions with Stockfish's verdict at once, or play the plans out from any diagram above.