@tpr you're right about that but either side can make error during the ongoing battle. White cannot afford to make minor inaccuracies simply because black is ahead in material. White has to keep the tension. Black can afford to do some riskier moves simply because he is up material but he can't go nuts.
«In open positions where the knight has no outposts in the centre, the strength of two bishops and a pawn is no less (and tends to be greater) than that of a rook and knight»
Glenn Flear wrote:
The defender probably has more chances of successfully defending with the pair of bish-
ops than "with other minor piece combinations. This seems to hold true even when the board
isn’t opened up for bishops. Indeed, there is a certain school of thought that claims that ’a
pair of bishops equals a rook and knight in the ending’. I’m not sure that this assertion can be
proven, but I decided to look at it from a statistical point of view.
From a sample of 68 high-level games in the NQE of rook and knight vs two bishops (this
time with players rated 2550 and above), there were 30 wins for the rook< and knight duo, 20
for the bishop pair, and 18 draws.
The number of wins achieved by the bishops seems quite high, which may reflect the
double-edged nature of this imbalance, but the overall percentage is a respectable 57% to
43%. We can interpret this as such: a pair of bishops shouldn't be underestimated, but the
rook and knight are still superior.
Glenn Flear wrote:
Although I fortuitously saved this game it gave me the lasting impression that, ceteris paribus, rook and knight are superior to two bishops. Here, even with an extra pawn, Black was always somewhat worse. I’m also willing to bet that White’s chances are better than 57% (see the introduction to this chapter) in the initial position, but everyone is entitled to their opinion! My impression is that, in cases where the pawn structures are intact, the bishops can’t find enough targets and are thus inferior.