Skype coaching session, 01 July 2018.

The game discussed (53 moves) is given below with annotations and without 
annotations.

Re-structuring of the sessions now offers the option for guests to bring 
their own games for group discussion.
Recordings of some previous coaching sessions can be found at:

http://www.open-aurec.com/Skype/PaulBenson/PaulBenson.htm

Paul Benson.

*  *  *

Annotated Game.

White: E. Iturrizaga, 2653.
Black: S. Sulskis, 2544.
Event: Tromso Olympiad 2014.
Result: 1/2-1/2 in 53 moves.
Opening: Reti Opening, Advanced Variation, A09.

1. Nf3 d5
2. c4 d4
3. b4 g5
4. Bb2 Bg7
5. Nxg5 e5
6. Ne4 f5
7. Nc5

| Varying from the more commonly played, 7. Ng3.
This knight is part of a master-plan to gang up on the cramping black d4 
pawn, take careful note of how white brings 3 minor pieces to bear on the 
target. ||

7. ... b6

| Not just a kick for the sake of it.
Black now has possibilities of organising, pawn c5, to strengthen the 
important d4 pawn. ||

8. Qa4+

| And similarly this is not just a check for the sake of it.
The planned white queenside piece coordination demands the queen find 
somewhere to live other than the seemingly-obvious squares of, c2, or, b3. 
||

8. ... c6

| Black sensibly chooses to limit the activity of the white queen, the 
rejected piece interpositions give white a comfortable game:
(A). If, 8. ... Bd7 9. Nxd7 Nxd7, white has the advantage of the bishop 
pair.
(B). Or if, 8. ... Nd7 9. Ne6 Qe7 10. Nxc7+, costs black at least an 
exchange.
(C). Or if, 8. ... Nd7 9. Ne6 Qc8 10. Nxg7 Kf7 11. Nxf5, white has snatched 
a piece and a pawn.
(D). Or if the somewhat counter-intuitive, 8. ... Kf7, black can castle by 
hand with, Ne7, Rf8/Re8, and, Kg8, while keeping options of, Bb7, both 
pressuring the white g2 pawn and making white think twice about moving the 
f1 bishop. ||

9. Nb3 Ne7

| Perhaps preferred over, 9. ... Nf6, which would reduce the protective 
force of the black g7 bishop up the h8 - a1 diagonal. ||

10. e3

| The white central plan is beginning to take shape.
White builds more pressure on the black d4 pawn, there is still not enough 
fire-power to win it, but what will happen when white has time for, Na3, 
and, Nc2, with 4 attacking units outnumbering the 3 black defenders of d4? 
||

10. ... O-O

| White to play and make a serious decision on the queenside. ||

11. c5

| White is preventing black from organising a supportive, pawn c5, by 
occupying the c5 square, either with a pawn or the knight, according on how 
black replies, all part of the conspiracy against that black d4 pawn. ||

11. ... b5

| Black closes the queenside which actually suits white as well, as there is 
a plan of white king safety involving castling long.
Note that the line-opening, 11. ... bxc5 12. Nxc5, would favour white, the 
black queenside pawn structure would be reduced to long-term targets just 
waiting to be hit. ||

12. Qa5

| Instead, 12. Qa3, would deny the white b1 knight the a3 square meaning the 
plan of, Na3 - Nc2, hitting the black d4 pawn would require further 
shuffling. ||

12. ... Qe8

| Black correctly keeps queens on, the middlegame strategy of constricting 
the white position needs as many units remaining in play as possible.
White on the other hand, would be pleased with every set of piece exchanges, 
the closer the endgame becomes, the more important the gambit-pawn deficit 
becomes. ||

13. Na3 Nd5

| Not just occupying a commanding central square from which it fights for 
control of dark squares in the white half of the board, this piece also 
backwardly guards the c7 square, preventing a potentially annoying, Qc7, 
activation of the white queen. ||

14. Nc2

| There are now 4 white attacking units hitting the twice-defended black d4 
pawn, black had better have a good plan in mind over the future of the 
threatened pawn, but what? ||

14. ... Be6

| Is that it?
Black's whole system embarked upon as early as move 2 demands the d4 pawn 
cramp any white activity in compensation for the gambit g-pawn, and yet 
nothing is being done to prevent the loss of the d-pawn. ||

15. exd4 e4

| Aha, now that is a really neat trick.
The white minor pieces were arranged on the queenside to win the black d4 
pawn, mission accomplished.
However, now that the threat has been successfully executed, they find 
themselves hampered by their own pawn on d4.
Furthermore, the white d4 pawn has little prospect of advancing, which would 
release the skulking white queenside minor piece trio.
In essence, a potentially dynamic central pawn formation has morphed into a 
static one requiring some manoeuvring from white before the extra pair of 
pawns can begin to make their presence known. ||

16. O-O-O

| So white has achieved king safety and is now ready to launch into an 
opposite-wing attack, right?
Not so simple.
The clutteration of minor pieces on the queenside is in need of a drastic 
feng shui session before any thoughts of counterplay can be undertaken. ||

16. ... Nd7
17. Qa3

| The white queen had no action over on the queenside and so plans a 
transfer over to the kingside.
Black in the game to play and find a concept likely to leave one thinking 
there must be a misprint with the game score. ||

17. ... N7b6

| Yes, the black b6 knight is now en prise to the white c5 pawn.
There must be some method in this apparent madness, but what? ||

18. Na5

| White agrees, there is indeed method in the madness.
Accepting the offered black knight is distinctly inadvisable.
If, 18. cxb6 axb6, then, 19. Na5, would be forced to avoid losing the white 
queen for a rook and knight, not good housekeeping.
Black could then hit immediately with, 19. ... bxa5, but perhaps preparing 
the recapture with, 19. ... Rf7, intending, Rfa7, before picking up the 
pinned white a5 knight might be more professional.
Either way, the white queenside would open up for the black rooks, and all 
the fun will be for black, without any material gains for white in the 
process.
By advancing the b3 knight to a5 there is not only an attack on the black c6 
pawn in need of continual neutralisation, which at this moment is performed 
by the black e8 queen, but the a-file is also plugged against any black 
plans of opening with a random, pawn a5, at any time.
Conclusion: Opening up lines in front of one's king needs very careful 
calculation, and in this instance it will only lead to tears before bedtime. 
||

18. ... Na4

| The black trick of, 17. ... Nb6, was not just a one-mover without further 
meaning.
The b6 square was a stepping-stone for this knight to achieve an active 
location around the white king, nothing decisive yet of course, but if a few 
more black units can get into the white queenside then expect tactical ideas 
to be in the air. ||

19. Ba1 Nf4

| Superficially one might think this knight is preventing the white f1 
bishop from making a move, possibly so, but there is a deeper reason for 
this knight shuffle.
Clue: When a piece moves it is tempting to think the movement of that piece 
is the only reason it has been moved. ||

20. Qg3

| Another point of the white, Nb3 - Na5, shuffle is revealed, the white 
queen finds a 3rd-rank route to the kingside, and with a useful attack tempo 
on the black f4 knight. ||

20. ... Nh5

| Kicking hard at the white queen, but she has a rather inviting aggressive 
invasion point available. ||

21. Qd6 Rc8

| White in the game to play has a dynamic option available which solves an 
annoyance inherent to this opening variation. ||

22. d5

| A game-changing move, the static centre becomes dynamic, pieces previously 
restricted become mobile, and kings will find themselves targets as enemy 
forces head their way. ||

22. ... Bxd5

| So white gives up the one of the harvested pawns in order to gain activity 
for the restricted minor pieces.
This rather suits black as well, the reason for earlier shuffling with, Nf4, 
which got pushed to h5, is also revealed, the bishop on e6 was destined to 
move to d5 to strengthen the pawn structure, which then offered the knight 
chances to probe and prod at the white position.
White in the game to play and demonstrate the difference between a sacrifice 
and a combination... ||

23. Bxb5

| That is certainly one way of solving the development problems of the 
awkward white f1 bishop, and as it happens effectively forces an indirect 
trade of the light square bishops, while also gaining a protected passed 
c-pawn into the bargain.
The tactical tool used here is an overloading, that is exploiting the black 
c6 pawn who has the task of protecting both the b5 pawn and d5 bishop, it 
cannot do both in this position. ||

23. ... cxb5

| Virtually forced, as, 23. ... Bxa2 24. Bxc6, gives white a mobile 
queenside pawn majority just waiting for their chance to expand once a few 
heavy pieces have been exchanged off. ||

24. Qxd5+

| Bobby Fischer informs us: "Obvious Therefore Dubious!", which is not to be 
taken literally but to provoke one to avoid immediate knee-jerk moves based 
on general feelings, instead on should take the whole position into 
consideration before coming to a decision.
With this recommendation in mind, it is black to play and thin 
counter-intuitively to find the game move. ||

24. ... Rf7

| The power of the pin is to paralyse, and here we have black voluntarily 
entering into rook-paralysis, why?
The answer is not easy to appreciate at first glance.
Black is anticipating a dark square bishop trade somewhere, most likely on 
g7, as white is unlikely to allow it to occur on a1, which would 
unnecessarily decentralise the c2 knight.
If this is so, then the black king can capture with, Kxg7, escaping the 
self-induced pin, meaning the f7 rook is then freed to swing across the 
black 2nd rank, the most likely square being d7, bringing pressure to bear 
up the semi-open d-file.
Isn't, "Grand Mastery", grand? ||

25. Bxg7

| Bobby Fischer warns us: "Obvious Therefore Dubious!", black in the game to 
play should sit on hands for a minute or so until the creativity juices have 
a chance to stir themselves into action... ||

25. ... Nf4

| Known in the trade by the term, "Zwischenzug".
This is essentially postponing an immediate recapture in favour of an 
attacking move which the opponent cannot ignore, resulting in a positional 
or tactical gain from the in between move.
Here the white queen must be careful where she goes, there are some sneakies 
just waiting to ensnare greed.
Had black foolishly ignored the advisory Fischerism, "Obvious Therefore 
Dubious!", and made the recapture of, 25. ... Kxg7, then, 26. g3, would have 
prevented the black h5 knight from enjoying the game continuation about to 
unwind. ||

26. Qb3

| The tempting material-retaining tries get their just desserts:
(A). If, 26. Qe5 Nd3+ 27. Kb1 Nxe5 28. Bxe5 Qxe5, white has lost the queen 
for a piece, an overwhelmingly decisive material plus for black, not to 
mention the little threat of mate on b2.
(B). Or if, 26. Qd4 Ne2+ 27. Kb1 Nxd4 28. Bxd4, white has lost the queen for 
a couple of black minor pieces, a long-term win for black. ||

26. ... Nd3+

| And now the relevant term is, "Zwischenschach", also known by some as, 
"Zwischen-check".
Warning: This black knight has jumped from obscurity on h5 to be of 
considerable annoyance on d3, and it is far from finished yet. ||

27. Kb1 Kxg7

| Black sensibly resists, 27. ... Nxf2 28. Ba1 Nxh1 29. Rxh1, when the 2 
white minor pieces plus a pawn should eventually prove of greater value than 
the black rook, there would also be many problems for black on the dark 
squares, not to mention the awkwardly placed black a4 knight which at 
present is stranded without any means to re-enter the game.
And as if the above was not enough, why should black give up a 
beautifully-placed d3 knight for an unmoved white rook with at present few 
prospects of finding middlegame activity?
Conclusion: Pawn count piece values of, rook = 5, and, knight = 3, are 
general guidelines which will fit well in most positions, however there are 
always exceptions, as should be observed here. ||

28. Rhf1

| What a picture!
One black knight on d3 is totally restricting the white d1 rook and forcing 
the other white rook to guard the f2 pawn.
Here is a fantastic demonstration that piece values using pawn-count, that 
is knight = 3, and 2 rooks = 10, need re-evaluation depending on the 
relative positions of all units concerned.
Of course, the 2 white rooks are not going to be so easily controlled for 
ever, so it is up to black to take full advantage of the plusses in the 
position before white breaks out of the prison. ||
28. ... Rd7
29. f4

| White wishes to fix the black f5 pawn, which strongly hints at a plan in 
mind.
Instead, 29. f3, intending, 30. fxe4, but after, 30. ... Qxe4, black will 
have an more active position than the game continuation. ||

29. ... Ndb2
30. Ne3

| The white d1 rook must sit tight as, 30. Rde1 Rxd2, to be followed by 
black increasing the pressure up the d-file with the other rook by doubling 
with, Rcd8, and possibly including the queen with,Qd7, looks very painful. 
||

30. ... Rd3

| Setting off a sequence of moves where neither side seems to have little 
choice other than the game continuation.
Instead backing off with a defensive move such as, 30. ... Rf7, or, 30. ... 
Qf8, would offer white the opportunity to organise, pawn h3, and, pawn g4, 
initiating some kingside activity. ||

31. Nxf5+ Kh8
32. Nd6

| Instead the tactically intriguing, 32. Qc2 Nxd1 33. Rxd1 Qh5 34. Nd6 Qxd1+ 
35. Qxd1 Nc3+ 36. Kc2 Nxd1 37. Nxc8 Nf2, leaves black an exchange up, but 
with problems of how to cope with 2 active white knights floating around the 
queenside, in particular aiming to force through the passed c5 pawn, 
unsurprisingly this line is not the best for black.
After, 32. Qc2 Nxd1 33. Rxd1 Qh5 34. Nd6, black takes advantage of the 
c-file pin with, 34. ... Rxd6, going a whole rook ahead, white must then 
accept the continuation, 35. cxd6 Rxc2 36. Kxc2 Qd5, black has the material 
plus of a very mobile queen against a white rook plus 3 pawns, but some of 
these white pawns must fall soon, everything indicates that black has the 
far better winning prospects.
A different white idea of, 32. Qc2 Nxd1 33. Rxd1 Qh5 34. Ne3 Qxh2, trying to 
hold the position an exchange down should not work in the long term, hence 
white in the game must settle for an indirect trade of queens. ||

32. ... Rxb3
33. Nxe8 Rxb4

| Black is still 2 pawns down, but now has 3 active units in the region of 
the white king, there is in particular the immediate black threat of a 
material-winning discovered check with the b2 knight in need of 
neutralising.
White on the other hand has 4 units scattered around the edges of the board 
with no coordination worthy of mention other than a passed f-pawn ready to 
sprint for the 8th rank.
Furthermore, the previous white trump-card of the protected passed c-pawn 
now has no protection.
And if this is not enough, white has the problem of the d1 rook and e8 
knight under attack, and no one move by either piece can solve their 
dilemma. ||

34. a3

| The only move to avoid material loss.
The idea is if, 34. ... Nxd1 35. axb4 Rxe8 36. Rxd1, and white has 
consolidated the 2 pawn advantage and can look forward to converting the 
extra material into the full point.
Instead, 34. Kc2 Rxc5+ 35. Kb1 Nxd1+ 36. Ka1 Rc1+ mate is cute. ||

34. ... Rd4
35. Nd6 Rxc5

| As the white potential danger-pawn on c5 is eliminated so is the defence 
to the white d6 knight, it seems white is running out of retaliations.
Instead defending allows white to break out, 35. ... Rb8 36. Rc1 Rxd2 37. 
Nxe4 Rxg2 38. Rc2, and white can soon think in terms of making something 
with either passed pawn. ||

36. Nb3

| What an reversal.
Until this knight fork, it was black who was hitting vulnerable white units, 
now it is black who must tread carefully. ||

36. ... Rxd6

| Instead, 36. ... Rcd5, leaves both players having to sift hard to 
determine which is wheat and which is chaff, some example lines:
(A). If, 36. ... Rcd5 37. Nxd4 Nxd1 38. Rxd1 Rxd4 39. Nxb5 Rd5 40. Nxa7 Nc3+ 
41. Kc2 Nxd1 42. Kxd1 Ra5 43. Nc8 Rxa3 44. Nd6, and it will be black under 
pressure to hold the draw, but black can improve on this.
(B). Or if, 36. ... Rcd5 37. Nxd4 Nxd1 38. Rxd1 Rxd4 39. Nxb5 Rd5 40. Nxa7 
Nc3+ 41. Kc2 Nxd1 42. Kxd1 e3 43. Ke2 Rxd2+ 44. Kxe3 Rxg2 45. h4 Rh2, black 
wins the white h-pawn creating a passed h-pawn of his own, white will find 
it impossible to keep the passed a-pawn, leaving matters to be decided on 
the kingside with white an exchange down, black should win this one.
(C). Or if, 36. ... Rcd5 37. Nxd4 Nxd1 38. Rxd1 Rxd6 39. Nxb5 Rb6 40. a4 a6, 
and as the pinned b5 knight is falling off, the pressure is on white to hold 
the draw.
(D). Or if, 36. ... Rcd5 37. Nxd4 Nxd1 38. Rxd1 Rxd6 39. Nb3 Nc3+ 40. Kc2 
Nxd1 41. Kxd1 Rf6 42. g3 Rh6 43. h4 Rg6, and the black rook gets in amongst 
the white kingside pawns, combine this with a black mobile queenside pawn 
majority and black should win.
(E). Or if, 36. ... Rcd5 37. Nxd4 Nxd1 38. Rxd1 Rxd6 39. Nf5 Rd5 40. Ne3 
Rd3, and black is winning the white a3 pawn as white cannot afford, 41. Ka2 
Nc3+, winning the white d1 rook.
The above 5 variations all have black playing, Nxd1, in response to dropping 
a rook on d4.
It is possible for black to consider, 37. ... Rxd4, or, 37. ... Rxd6, 
leaving white to decide the fate of the d1 rook while the other d-file 
knight is attacked.
Aren't middlegames wonderfully complex when pieces find themselves hanging 
in fresh air without supporting pawns to help them? ||

37. Nxc5 Nxd1

| This time the white d1 rook must be taken, instead, 37. ... Nxc5 38. Kxb2, 
black is an exchange and pawn down, white is easily winning. ||

38. Nxa4

| Instead, 38. Rxd1 Nxc5, and white is a piece down for a pawn.
With the departure of the black a4 knight the d1 knight is stranded without 
a flight route, but black has an answer. ||

38. ... Rxd2

| Black who has had a material deficit since move 5 at last not only 
achieves material equality but has the more active position, sometimes 
gambiteering requires a little patience, right? ||

39. Kc1

| The simpler route for white is to head for a rook and pawn ending.
Instead moving the endangered a4 knight needs to be handled with a little 
care:
(A). If, 39. Nc5 Nc3+ 40. Ka1 Ra2+ mate is unlikely to happen.
(B). Or if, 39. Nc5 Nc3+ 40. Kc1 Ra2 41. Re1 Rxa3 42. Nxe4 Ra1+ 43. Kd2 Rxe1 
44. Kxe1 Nxe4, white has lost a piece, black wins.
(C). Or if, 39. Nc5 Nc3+ 40. Kc1 Ra2 41. Re1 Rxa3 42. Re3 b4 43. Nxe4 Nxe4 
44. Rxe4 a5 45. Kb2, and despite the black pawns being more advanced white 
can easily hold the draw. ||

39. ... Rxg2

| And shock-horror, black has actually gone a pawn ahead, gambiteering 
actually works!
Instead, 39. ... Rd4 40. Rxd1 Rxa4 41. Kb2, should be comfortably drawn.
Black should not fall into, 39. ... Rd4 40. Rxd1 Rxd1+ 41. Kxd1 bxa4, as the 
white king marches to gobble up the black e4 pawn for a simple winning pawn 
ending. ||

40. Kxd1 bxa4

| Aha, black might be a pawn up but the doubled a-pawns are in effect worth 
only 1 pawn. ||

41. Re1

| Now the game enters a curious phase where each player sets about 
guaranteeing they can hold the draw.
White must first eliminate the potentially dangerous passed black e4 pawn, 
though as it happens black has no real means of using it as the h8 king is 
too far away to assist. ||

41. ... Ra2

And now black need only eliminate the white a3 pawn and there is a 
well-known draw available, white cannot win with rook, f-pawn and h-pawn 
against rook, and this assumes that black is somehow going to conspire to 
lose all remaining pawns. ||

42. Rxe4

| Again white takes the simple solution.
Instead, 42. Re3 Kg7, and the black king given enough time will advance to 
the d4 square pushing the white rook around, the big threat would then be, 
pawn e3, Kd3, and Ra1+ mate. ||

42. ... Rxa3
43. Ke2 Ra1

| If black is going to make anything with the a4 pawn then this rook needs 
to be on the black 8th rank.
Of course, white has a few moves to set up the correct defence to neutralise 
the combined efforts of such a black rook and advancing passed edge-pawn. ||

44. Kf2 Kg7
45. Re6

| A double purpose move.
Firstly, the white rook prevents the black king from advancing any further 
up the board.
Secondly, the rook will transfer to a6 placing itself behind the black a4 
pawn, this is absolutely vital to prevent pawn promotion. ||

45. ... a5

| At this standard one can be certain each player knows what the result will 
be, they are probably going through the motions to satisfy the demands of a 
team match by continuing until there is absolutely no chance of finding a 
win.
Black could have left the pawn on a7, white will still put a rook on a6, 
designed not to win the a7 pawn but solely to hold the black advanced a-pawn 
from the rear. ||

46. Ra6 a3
47. Rxa5 Kg6

| So by pushing, pawn a5, and letting the white rook gobble it up, the black 
king has managed to sneak a rank closer to the centre of the board.
All very good, but there is no sensible means of getting the white a5 rook 
off the white 5th rank, or is there? ||

48. Kg2

| White now has the necessary conditions to ensure the black passed a-pawn 
cannot promote, these are:
(A). The defending rook must be sitting behind the attacking passed pawn, 
keeping it under attack from the rear.
(B). The king must be sitting on either of the g2 or h2 squares, and please 
note the mirror-image position of someone trying to promote the h-pawn would 
demand the defending king sits on either of the b2 or a2 squares. ||

48. ... Ra2+

| For a little extra rook ending theory please imagine the position altered 
as follows:
White: imagine an extra pawn on g3.
Black: Imagine 2 extra pawns on g7 and f7.
The kingside is now equal, 3 pawns each.
Black can try for a win with the following method:
Step 1: March the king along the 3rd rank to the b6 square, attacking the 
white rook.
Step 2: White should keep the rook on the a-file, so, Ra8, seems logical.
Step 3: March the black king up the b-file to find a hiding place on the a2 
square.
Step 4: Create shelter for the black king by placing the rook on the b3 
square.
Step 5: Play, Kb2, followed by, pawn a2, and, pawn a1 promoting.
Quite a long-winded process which white can thwart if the king can land on 
the c2 square, preventing black, Kb2.
There would also be ideas of white trying to snatch at least 1 if not 2 of 
the black kingside pawns, return to the a-file anticipating having to give 
up the rook when black plays, pawn a2.
This subtlety is necessary as after black plays, pawn a2, black has the 
option of, Ra3, and attempts to perpetuate from the black back rank with 
checks will soon run out as the black king simply runs backward toward the 
checking white rook, when contact is made the advanced pawn will promote 
next move.
Black having successfully won rook for the a-pawn must then get both the 
king and rook over to the kingside before white can make something with 
whatever pawn majority was established before the white rook had to be given 
up.

This process demands careful calculation, if the white pawns can get rolling 
before black wins the white rook it could be that white might find a draw as 
black is forced to give up his rook for a promoting pawn, or even worse, 
white might win the black rook and have sufficient resources to win the king 
and pawn ending.
Well, there is the theory, now all one needs to do is analyse in real-time 
should it arise and judge whether a win is on or one is risking a defeat. ||

49. Kg3 Ra1

| Having just given some theory on promoting edge-file passed pawns, black 
judges any attempts to get the king over to the a2 square will cost heavily, 
white will use his king to shepherd the f-pawn up the board, costing black a 
rook while the black a-pawn costs white a rook, then the white king will mop 
up the black h-pawn freeing the file for white to promote the h-pawn. ||

50. Kg4

| White is demonstrating there are several ways to draw.
The simplest method was to shuffle the king on the g3 and g2 squares in 
response to black playing,Ra2+, and, Ra1. ||

50. ... Rg1+
51. Kh4 Rf1
52. Rxa3 Rxf4+

| Perhaps the players have a sponsored move side-deal going on? ||

53. Kg3 Rb4
Draw agreed, 1/2-1/2

*  *  *

## Unannotated Game.

White: E. Iturrizaga, 2653.
Black: S. Sulskis, 2544.
Event: Tromso Olympiad 2014.
Result: 1/2-1/2 in 53 moves.
Opening: Reti Opening, Advanced Variation, A09.

1. Nf3 d5
2. c4 d4
3. b4 g5
4. Bb2 Bg7
5. Nxg5 e5
6. Ne4 f5
7. Nc5 b6
8. Qa4+ c6
9. Nb3 Ne7
10. e3 O-O
11. c5 b5
12. Qa5 Qe8
13. Na3 Nd5
14. Nc2 Be6
15. exd4 e4
16. O-O-O Nd7
17. Qa3 N7b6
18. Na5 Na4
19. Ba1 Nf4
20. Qg3 Nh5
21. Qd6 Rc8
22. d5 Bxd5
23. Bxb5 cxb5
24. Qxd5+ Rf7
25. Bxg7 Nf4
26. Qb3 Nd3+
27. Kb1 Kxg7
28. Rhf1 Rd7
29. f4 Ndb2
30. Ne3 Rd3
31. Nxf5+ Kh8
32. Nd6 Rxb3
33. Nxe8 Rxb4
34. a3 Rd4
35. Nd6 Rxc5
36. Nb3 Rxd6
37. Nxc5 Nxd1
38. Nxa4 Rxd2
39. Kc1 Rxg2
40. Kxd1 bxa4
41. Re1 Ra2
42. Rxe4 Rxa3
43. Ke2 Ra1
44. Kf2 Kg7
45. Re6 a5
46. Ra6 a3
47. Rxa5 Kg6
48. Kg2 Ra2+
49. Kg3 Ra1
50. Kg4 Rg1+
51. Kh4 Rf1
52. Rxa3 Rxf4+
53. Kg3 Rb4
Draw, 1/2-1/2

*  *  *