From a purely positional point of view, 18...Nxb6
would have been a
terrible mistake, as it would have allowed white to then forcefully trade the light squared bishops with the immediate Bg4
move, leaving exposed all the light square weaknesses that black has
created with the d6-e5-f4
setup.
These weaknesses (created holes) would then have been permanently exploited by white's
knight (and queen), in other words, white's minor piece (b6
knight) would have
dominated the game and been virtually untouchable on light squares for the rest of the game.
Notice that next to the knight controlling key light squares from b6
,
it's also for example quite feasible for white to re-route the knight
to e4,
established centrally (permanently) and constantly eying black's
weak d6
pawn.
And as for black, Carlsen would have simply been left with a bad dark
squared bishop hemmed in by his own structure.
Therefore, Carlsen chose wisely to play Qe8
instead of Nxb6
, thus,
preserving his c8
bishop as Bg4
no longer forces a trade thanks to
the well placed knight on d7.
In summary, this is a purely positional decision where black is well
aware that having weakened their light squares, any scenario with the
knight vs dark square bishop would be unfavorable for them, and
arguably,
at that level even objectively lost! Instead having kept both the d7
knight and c8
bishop, black has in fact preserved two pieces that can
provide light square control.
The following situations would have ensued had Carlsen played
18...Nxb6
:
18...Nxb6 19.Nxb6 Ra7 20.Bg4
which would have led to positions of the following nature, infested with light square holes: