I read many times in books that when the bishop on a fianchetto is removed, like in some variants of the Sicilian Dragon, the squares of its color around are weaker, which is obvious.
What I don't understand is why this is said even when that bishop is exchanged by the opponent's one.
From my point of view, those squares are weaker, but the chances of converting that into something concrete are smaller, since we now have one less piece to attack them.
Trading dark squared bishops (DSBs) does weaken the kingside dark squares. It is also true that without a DSB, white has a slightly harder time attacking the dark squares. But the queen is an important attacker that CAN hit the dark squares, and the two knights (+ two rooks on the open h file) mean that there is still serious attacking power against black's weakened king.
One famous, representative example of how you can build a kingside attack on the dark squares after a DSB trade is Karpov - Korchnoi, 1974. Notice how the open h file (and therefore h7) and ultimately the g7 square were the eventual targets in this vicious attack.
Here's a similar, extremely unfamous example by me where I target g7 more directly than Karpov. Notice how the idea of Nf5 only comes up once you see how hard it is to defend g7 - the knight has to be taken.
When the fianchettoed bishop is gone, the opponent can do this and it's not easy to dislodge the queen:
[FEN "6k1/4pp1p/3p2pQ/8/8/8/8/8 w - - 0 1"]
When the fianchettoed bishop is still around, ...Bg7 ejects the queen. White's DSB is not a factor, because it's not needed to put the queen on h6.
