One of the things I mentioned when I introduced my new column last week was the paradox of the column itself. It deals specifically with positions without the queen, but is not necessarily about endgames. There tends to be differing methods of play in endgames versus middlegames - in endgames, structural weaknesses and long-term factors tend to be more important, while king safety and dynamics become less important. As a result, though, it becomes easy for players to underestimate dynamics and dangers to the king in positions without the queen. Indeed, the players' assessment of the position as an "endgame" versus a "queenless middlegame" affects their way of play. For example, look at the following game:
The players had a greatly different opinion of what kind of game of chess they had on the board after move 13. Mukhin clearly saw it as a placid endgame, and he was probably looking forward to making a draw against the former World Champion. Tal, on the other hand, saw the game as a queenless middlegame, where the white king was fair game.
White's assessment of the position as of the "quiet" type induced him to make certain decisions. First of all, the move 13.b4 - seemingly a natural, space gaining move which prepares the development of the bishop on b2 - actually was an inaccuracy, since it leads to some weaknesses on the c-file. It's hard to imagine that White would not have the time to consolidate the c4 square, but Tal shows that time is very valuable in this position.
Second, White made an important decision to leave his king in the center. This is a crucial difference between middlegame and endgame positions. As you know, in the endgame the king is usually best placed in the center, while in the middlegame it should be hidden away somewhere. White clearly saw the game as an endgame and didn't imagine that the king could come to any harm on the "safe" e2 square, but Tal showed that it actually could.
Attacking the king in the center, with lots of spectacular sacrifices, only happens in some kind of sharp opening (with the queens of course remaining on the board), perhaps an open Sicilian or something such? Isaias Pleci shows that it can begin with a queen trade:
Of course, not all queenless middlegames in which the king is displaced are dangerous. The following variation is well-known as good for Black - the displaced king is not as important as White's unhappy pawn formation - the c4 pawn restricts the white bishop and creates weaknesses on the d-file and queenside:
Naturally, in a true ending king safety can also become an important factor. Clearly you cannot call the following example a middlegame. And you could be forgiven for expecting that Black should stand well, with a bishop against a knight and an outside passed pawn. But it turns out that White's direct attack on the king gives him all the chances. Black defended well and the game had an exciting conclusion.