Which are better, Knights or Bishops?

Question

I have always thought Bishops were better - I know that a KBB vs. K endgame is easily won, whereas KNB vs. K is difficult, and KNN vs. K is impossible; but then again, some people prefer Knights. I realize there is a reason for each receiving the same point value (3) - they are so similar in strength, the Bishop on open lines and the Knight in close quarters.

But is there any concrete study on the point, such as someone analyzing thousands of games in which one player began the game with two Bishops and no Knights, and the other had two Knights and no Bishops?

Answer 1

It depends on the board position. Both have equal points. But in a closed position, i.e, when the mobility of our pieces are blocked by the opponents pieces, knights have advantage. On more open board positions, bishops have advantage.

Answer 2

Usually in open positions the bishops are better. When the position is blocked up the knights are generally better. In an endgame with pawns only one one side of the board, the knight is usually better. In endgames with asymmetrical pawns on both sides of the board, the bishop is usually better. With a symmetrical endgame pawn structure you might like the have the knight and place your pawns on the opposite color of the opponent's bishop. Possessing the 2 bishops is also generally overrated and I have not really seen anyone < 2000 USCF really able to finesse the position to take full advantage of that.

Remember all these are general principles and the specific position must be evaluated always before general considerations.

Answer 3

None of them are "better" in a general sense, but they are different. As bishops are long-ranged and can theoretically control more space than a knight (up to 13 squares against 8), they tend to become more powerful in situations where there are open diagonals (space-aspect) or during endgames when pawns are at both sides of the board (long-range-aspect). On the other hand, knights have a great maneuverability, can attack more different pieces without being attackable from them in return and support pieces with different movement rules very well (like attacking the same spot, but from a different square). So they are different tools, becoming very powerful or useless depending on the board situation.

Your question can't be answered generally. You could ask as well, whether a spoon or a fork is the better eating tool. It all depends on whether soup or chips is the dish of the day ;)

Answer 4

Steve Mayer wrote an excellent book on the subject, Knight versus Bishop: The Verdict. You could do worse than have a look at that. It explains the circumstances under which one is better than the other, and the fact that he wrote a well-regarded book about the topic suggests that it's not easily addressed in a forum reply.

Answer 5

My answer comes years late but differs from the others.

A bishop can often reach a square from which it can control, restrict or bind an opposing knight. See the diagram, which gives an unspectacular but typical example.

[fen ""]
[startflipped "0"]
[startply "58"]
[title "S. Mamedyarov v. M. Carlsen, Wijk an Zee, 2018"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.Qc2 d5 5.cxd5 Qxd5 6.e3 c5 7.Bd2 Bxc3 8.Bxc3 cxd4 9.Bxd4 Nc6 10.Bc3 O-O 11.Nf3 Rd8 12.Be2 Qe4 13.Rc1 Qxc2 14.Rxc2 Nd5 15.Ne5 Bd7 16.Nxc6 Bxc6 17.Bd2 Ne7 18.f3 Rac8 19.e4 f6 20.Be3 a6 21.Kf2 Bb5 22.Rhc1 Rxc2 23.Rxc2 Bxe2 24.Kxe2 Nc6 25.b4 Rc8 26.Rc5 b5 27.f4 Kf7 28.a3 Ne7 29.Rxc8 Nxc8 30.Bc5 Ne7 31.Bxe7 Kxe7 32.Kf3 Kd6 33.Ke3 e5 34.f5 Ke7 35.g4 Kf7 36.h4 Kg8 37.Kf3 h6 38.h5 1/2-1/2

White moves 30.Bc5. In this particular game, Black is Magnus Carlsen, so Black has foreseen the trouble and has advanced his king into a position from which the king can help to rescue the knight from its bind. The point, however, is that the knight is indeed in a bind. The bishop has done this to the knight.

Admittedly, knights can take away squares from bishops, too, but a bishop can often escape by advancing past the knight out of the knight's range. On average, bishops have the upper hand over knights in positions like these.

Nor are positions like these rare. Such positions are common, in fact. Indeed, such a position occurs again earlier in the same game. (Easy exercise: find the position. The same white bishop restricts Black's other knight for six full moves.)

This bishop-versus-knight dynamic, in which the bishop has slightly the better of the struggle, is a significant reason to regard the bishop as being—on average—slightly stronger than the knight.

The knight can of course reach all squares of the board. This is a significant point in the knight's favor, but the bishop has other, subtle advantages. For instance, a bishop can three-step.

Consider: a knight can move away from its square and back in two moves but, interestingly, cannot do so in three moves. A bishop by contrast can indeed use three moves to return to the square from which it starts. In an endgame, this three-step dynamic can allow the player with a bishop to put his opponent, who has a knight, in Zugzwang.

There is also the bishop pair. One of the other answerers opines that non-expert players can seldom effectively use the bishop pair. Actually, there is some truth in that opinion, but I have (occasionally) seen non-expert games decided by a pair of bishops, so you have that factor to consider as well.

On the edge of the board, bishops are often happier than knights.

A bishop and a pawn can mutually protect one another—yet, with a quick push of the pawn, the position can be converted into one in which the pawn covers squares of one color and the bishop squares of the other.

All factors considered, on average, the bishop is probably very slightly the better piece. But yes, of course, as others have pointed out, it depends on the specific position at hand.

Answer 6

We could definitively answer this. You'd use Shapley Values and simulation with Stockfish to reveal the value of the various pieces. You'd have to think about how to set up the simulations, if you'd only simulate games from starting positions, removing 1 or 2 knights or Bishops, but basically it's a quantitative question which could be answered.

Intuitively, I'd rather have one knight and one bishop than no knights and both bishops or vice versa. This could also be tested, and various other conditions, via simulation.