Originally posted by Fat Lady
Whenever I plug one of my OTB King's Indian games through an engine to see if my opponent or myself made any outrageous blunders, it always seems to think White is doing fine right up to the point where he loses. This is a typical example from a league game I played tonight:
[pgn]
[Event "Oxfordshire League"]
[Date "2013.03.04"]
[White "NN"]
[Black "Fa Kh1 Bxd4 24. Rxd4 Qxd4 25. Ne2 Qd3
26. Qc1 Nb3 27. Qe1 Nd2 28. Rd1 Nxf3
0-1
[/pgn]
I often have this experience playing the King's Indian Attack OTB.
My study group had a conversation about this two weeks ago, and we decided that it must be the horizon effect. The KIA and KID are the type of opening where the attacks are often deep, and are not readily apparent for up to 15 moves, whereas many openings have very obvious and straightforward threats.
For instance, Bc4 by white in a kingpawn opening is usually a very straightforward threat to f7 and the black king. By contrast, when black plays ...Nf6 and then ...Nfd7 or ...Ne8 in the KID, this seeminly inoffensive idea (a retreat, no less) is actually the beginning of the attack, preparing black's ...f5 pawn break.
I think it is the pawn breaks that are the real culprit. Open games are easier to calculate with a short horizon, whereas closed positions with attacks preceded by piece maneuvering and pawn breaks are much deeper.