e5 is not really a threat yet, and many kings indian experts used to castle at move 4. the reason is that the white centre collapses after 4.0-0 e5?! Ne8 followed by d6 and c5.

?! Premature. b4 or Ne1 were better.

The point of this knight move is, of course, to free the f-pawn for the ...f5 break move. If Black wants to keep a line open for his c8-bishop, then, 10...Ne8 is in my view a reasonable alternative. Naturally, White would prefer to keep the knight active, if he could, by ...Na5, but, though it is good in fairly similar lines, it won't do here owing to White's Q+B hold on the h5 square.

The other downside - and possibly the more important consideration - against the bxa5 capture, is that it renders White's thematic break move c4-c5 very difficult to carry out effectively. The purpose of the break move is to weaken Black's pawn chain at its base, to wit, the point d6. Black, meanwhile, hopes to weaken White's base at e4, OR to allow the transfer of the base to f3 (by now advancing ...f4) where it can be attacked much closer to the castled White king with the advance of the g-pawn. This kind of attack can be very dangerous indeed.

This move seems to mark down 14.Nb3 as an error, where it did seem to achieve very much.

This is the thematic break. Now both sides are striking at their opponent's bases. But the proximity of Black's attack to White's king makes his attack the more dangerous, as more often than not this attack on the pawn base metamorphoses into a direct attack on the king. Possibly White ought to have considered whether or not 19.h3 inhibited Black's g-pawn advance, and/or how serious was the price of weakening the dark squares around his king..