Move 19a discussion: Everyone vs Stockfish

[Brian-E]

The what-to-do vote hasn't quite finished but it looks very much as if we're playing this game first so I've spent a little time on it.

I too like the move 19.Nd4 which I think gives us the best chances of a draw. I cannot find any improvement for Black in Gary's analysis.

[WMHalsdorf]

19. Nd4 Nxb2
20. Rfb1 Nc4
21. Rb5 Re5
22. a4 Ra5

[gd_barnes]

Originally Posted by LaurV

Yes, this version is also playable toward a draw, black can't win if we play it right. Here I would argue that white is a bit weaker [edit: compared with 19 Rfd1, and not compared with black], because the sparse pawns. On the other version, if we can keep the three pawns connected there, we are much better off.

I had a long time yesterday to look at the differences between Rfd1 and Nd4. I now agree with LaurV. I think that Rfd1 is better. Nd4 may draw more quickly in some variations but it does create some variations that require much more exact play by us than Rfd1; not something that I think we want to do against SF. The problem is that we have no lines where we can drive the knight away with a timely b3. Also we maintain our nice 3 connected queen side pawns for a while longer. As will be shown below, we can sack a pawn at a point in which I think is more favorable to us. Here are the best Rfd1 variations that I came up with:

19. Rfd1 Re5
20. f4 Re2
21. Nd4 Re4
22. Nf5(!)* Nxb2
23. Rxd5 Rxf4
24. g3 Rc4

Here we can chase the black king around all day and/or put single or double threats on his b7 and f7 pawns and there is nothing that he can do. Eventual moves like Nh6+, Nf5+, and Rd7. One key is that in several lines here we can put a double attack on the f7 pawn threatening to potentially pick off 3 pawns in 3 moves.

If the temporary sacking of a pawn with 22. Nf5 looks unsavory to people, playing the previously discussed 22. b3 also works. It leads to a bit murkier waters. Refer to lines in move 18 for an analysis of that.

*This move had not been previously considered in deep analysis on move 18. It is the point in delaying the sacking of a pawn. Now after 23. Rxd5, you'll see that we will get a 2-pronged attack on the king side with perpetual check threats and on the queen side in potentially picking off both pawns, which threaten to effectively recover our sacked pawn(s) with a very good position.

Other response proposed by LaurV on move 18:
19. Rfd1 Rd7
20. Rd4 Nxb2*
21. Rg4+ Kh8 (a)
22. Rf4 Rd6 (b)
(protecting the pawn this way does not work because...)
23. Nd4 Nc4
24. Nb5 Re6
25. Nd4 Rd6
26. Nb5 Re6
(etc., black must drop the rook back to the 7th rank and give up the f6 pawn or face a 3-time repetition)

*We delay by one move sacking the b2 pawn and our rook is much better placed to mount a perpetual check threat with tempo enough to bring our knight into the foray.

(a)
21.....Kf8
22. Rh4 Rc7 (trying to avoid the perpetual check with a weak queen side attack)
23. Rxh7 Kg8
24. Rh3 Rxc2
25. Nd4 Rd2 or Rc7
26. Nf5
(We are all over the black king. Perpetual check or 3-time repetition are immanent.)

(b)
22.....Kg7
23. Rg4+ Kf8
24. Rh4 Rc7
[ transposes into lines from (a) ]


Because we can so easily force a perpetual check or 3-time repetition if SF responds 19.....Rd7, I feel that SF will respond with 19...Re5.

[gd_barnes]

In an effort to be as thorough as possible, here appears to be a better attempt by black to avoid quick drawing lines from the (a) variation shown in my last post:

19. Rfd1 Rd7
20. Rd4 Nxb2
21. Rg4+ Kf8
22. Rh4 Rc7
23. Rxh7 Ra4 (!) (a)

This stops us from bringing our knight into things, which is key for us. Things have gotten murky again. Maybe we should not be so hasty to take the h7 pawn.

(a) Let's bring the knight into play, which creates a lot of problems for black:
23. Nd4 Ra4
24. Rb1 Nc4 (b)
25. Rxh7
(Looks good for us here. We'll eventually play Nf5 with many threats on the king side.)

(b)
24.....Rcc4
25. c3 Rxa2 (c)
26. Rxh7 Rxc3
27. Nf5 Kg8
28. Rg7+ Kf8
29. Rh7
[black must give up the f7 pawn (and we are all over his other pawns) to avoid mate or 3-time repetition]

(c) We can swap down to a drawn end game a pawn down but we must play a little more exactly with:
25. Rxb2 Rxd4
26. Rxd4 Rxd4
27. Rxd6 Rd1+
28. Kf2 Rd2+
29. Kg3 Rxc2

I think that 25. c3 is better.

[LaurV]

Originally Posted by gd_barnes

Because we can so easily force a perpetual check or 3-time repetition if SF responds 19.....Rd7, I feel that SF will respond with 19...Re5.

Do you think he has a so-deep intelligence? I would be scared to play against it in that case! hehe..
I mean, SF can evaluate positions, very deep and wide, on the moves' tree, but I don't think it implements strategy, behind a very basic level, and for sure it does not implement "philosophy" . From the opening it chose I would have assumed that it always plays the "safer path", i.e. toward the draw. Unless it considers it has advantage (material or positional) and in this case we are surely missing something important. I don't think that is the case.

Anyhow, excellent analysis for sure you put a lot of time into it. I will spend some time tonight (in just few hours, here is lunch break now) to go through all your tree, I already have the position in my mind, but with a board in front it should be much clear.

If nothing comes up, my move is 19. Rfd1, for sure.

[gd_barnes]

Originally Posted by LaurV

Do you think he has a so-deep intelligence? I would be scared to play against it in that case! hehe..
I mean, SF can evaluate positions, very deep and wide, on the moves' tree, but I don't think it implements strategy, behind a very basic level, and for sure it does not implement "philosophy" . From the opening it chose I would have assumed that it always plays the "safer path", i.e. toward the draw. Unless it considers it has advantage (material or positional) and in this case we are surely missing something important. I don't think that is the case.

I think it has tremendous brute force power both far and wide; not intelligence per se. I would expect that it would be programmed to avoid a draw if it thinks it has a slight advantage or to play for a draw if it thinks it has a slight disadvantage (if it can do so without increasing its disadvantage too much). Perhaps that's why it played 17...Qb6 allowing the queen swap instead of the more tactical 17...Qd6. It likely perceived that it was still at the slight opening disadvantage and was maybe playing towards more drawish lines.

All of that said, in a nearly equal game like I think we have now, I think it would be programmed towards more complex positions that are not easily drawn and would allow its opponent to blunder more easily. This is why I think it will play 19...Re5 over Rd7 in response to our 19. Rfd1.

[richs]

I agree with 19. Rfd1. I think the line will proceed as Mr. Barnes notes in his first line in post #14.

19. Rfd1. A considered and valid analysis. I'm down for it also.

Has anyone read "Kleines Lehrbuch des Schachspiels?"

I'm in Freeport, Bahamas for the month. If anyone knows of chess players in the area I'll be interested in playing a few games.

[Brian-E]

I prefer 19.Nd4 on the positional grounds that I feel that we need to contest the e file when we can.