Very little is known about the initiative. That's why I take a baffling position that has confused me for long, in order to see if I can find some governing rules.
You can find the problem here.
First I tried to apply some ideas like CCT in relation to the value of the pieces. But that just doesn't make sense. CCT orders the moves by force. The more forceful the move, the higher in the hierarchy. That might be something than can be easely calculated by a computer, but my brain has problems with that. I soon realized that the problems were caused by the fact that to different types of moves are mixed up.
Three ways to win with a tactic.
How do we win with a tactic? There are three ways:
- Mate the king
- Capture a piece
- Promote a pawn
Direct moves first.
This means that we first must have an idea about the direct moves, before we can study the indirect ones. In the diagram above, mating the king or promoting a pawn is clearly not the theme. So the position is about captures. There is a natural hierchy of captures: the captures where a piece of low value is traded against a piece of high value first.
This means that the first capture that must be considered is Nxg6. That moves gains a value of +5 (a whole rook) since white is outnumbered on g6. After the gain, the initiative is handed over to white. The knight on f4 was shielding the black queen. Black is outnumbered on f4. Qxf5 Bxf5 Rxf5 leads to a cost of -3 (a bishop) for white. So the most logical sequence of capturing leads to an advantage of +5-3=+2 (the exchange) for black.
If you look at other captures that black can make, it soon becomes evident that he cannot improve on the given line. But maybe white can.
Then indirect moves.
The sequence of direct captures of both sides leads to a -/-2 disadvantage for white. Can he improve on that by postponing the direct captures while maintaining the initiative?
Take for instance 1. ... Nxg6 2.Qe3 threatening the black queen. If black plays the natural 2. ... Qe6 to get out of the way of the white rook and white plays 3.Bxh5 than all of a sudden black is in all sorts of problems. His knight is both pinned against the rook and outnumbered. His king is unsafe.
So black too must not play his most direct move 2. ... Qe6 but he must play 2. ...Rxe5. If black cashes in the queen now the score becomes +5 (g6) -/-4 (f5) +3 (e5) = +4 for black (RRN vs Q).
Identifying the manoeuvres.
How do these manoeuvres work?
The problem with cashing in a capture is that it hands over the initiative.
With 1. ... Nxg6 2.Qe3 white abstain from the positive cashing in with 2.Qxf5 Bxf5 3.Rxf5.
In stead of that he introduces two new threats: 3.Rxf5 and Bxh5.
It's very tempting for white to rely on his +6 points (Rp) advantage and to save his queen. The price he pays though is giving up h5, his king safety, and a lot of invasion points for white to use. Alltogether that will proof too much. Especially since black has two undeveloped pieces (Ra8 and Bc8) so effectively he plays with a piece less.
So black has to answer with a manoeuver of himself.
1. ... Nxg6 2.Qe3 Rxe5 3.Rxf5 Rxf5.
Total cost of this manoeuver: +5 +3 -/-5 so black gains a full piece. Since h5 is still protected the black king is still safe enough. The weak point of whites manoeuver is that he hasn't captured something during his previous move. He only threathen. But with Rxe5, only two threats remain for white: capture the queen with Rxf5 or capture the rook with dxe5. Both options are favourable for black, value-wise. Both keep the black king safe enough. Black gives some material back, but he doesn't compromises his king safety and het prevent to hand over invasion squares to white. In the end that is a better option. Since the pawnending is winning for black, white must prevent trades too.