As noted by <FlashinthePan>, 27. f5!! does not involve any immediately spectactular (dare I say "flashy") tactics. Indeed, the idea behind the move is almost strategic as 27. f5!! visualizes the transition from a complicated middle game position to a won endgagme.

27.f5 seems obvious both to get the pawns going but also to prevent loss of e pawn and advance on queenside. White should not worry about a1 rook.

27...Nc2 28.Kf4 Nxa1 29.Rxa1 and the game is as good as over because white will also win the g4 pawn with the help of the knight (via f1) or even with the help of the rook.

D Rogozenko vs Nguyen Anh, 1994 (27.?)

White to play and win.

Material: Up a P. The Black Kd7 has 5 legal moves, but is secured from all checks but 27.e8=Q+. The salient feature of the position is the White Pe7 and the other passed Pf4. The advanced passed Ps invite sacrifice of material for tempi, because Black threatens 27…Nc6 28…Nxe7. Black also threatens 27…Nc2+ forking Ra1 and Ke3, but the White Ke3 is secured from all other checks, except the pointless 27…Nf5+.

Candidates (27.): f5

27.f5 Nc2+ [else, the connected passed Ps are stable without compensation]

Candidates: Kf4, Kd3

[28.Kf4 Nxa1 29.Rxa1 looks like difficult progress]

28.<Kd3> (keeping the f-file clear for a R)

<[At 15 plies, Toga evaluates the game variation 28.Kf4 at about +1.2 P; my 28.Kd3 at about +0.7 P. I delete the rest of my analysis for brevity.]>

This is one occasion where the normal piece values don't really apply. The Nd4 is arguably stronger than the Ra1, especially as the position is closed. And white's passed pawns are arguably stronger than either of them!

So, 27...Rg7 instead. Now, if 28 f6, then 28...Rf7.

click for larger view

From the Merriam-Webster Online Dictionary [ http://www.merriam-webster.com/dict... ] :

Etymology: trade name, from Latin, higher, comparative of excelsus high, from past participle of excellere.

A) 27... Nc2+ 28.Kf4 Nxa1 (otherwise 29.Rac1 and White made progress without any compensations) 29.Rxa1

A.1) 29... a6 30.e5 dxe5 (30... Bxg2 31.e6+ and 32.f6 + -) 31.Kxe5

A.1.a) 31... Bxg2 32.f6 Ke8 33.Ke6 + -.

A.1.b) 31... b5 32.Ne4 (threatens 33.Nf6+, 33.Rd1+ Ke8 34.Nd(f)6+, etc.) Bxe4 33.Rd1+ Kc7 (33...Ke8 34.Ke6 Bxf5+ 35.Kf6 Rf8+ 36.exf8=Q+ Kxf8 37.Kxf5 + - [B vs P]) 34.Kxe4 followed by f6, f7, etc.

A.1.c) 31... Rac8 32.Rd1 Ke8 (32... Kc7 33.f6 + -; 32... Bxg2 33.Ne4+ + -) 33.Ne4 Bxe4 34.Kxe4 looks unclear. For example, 34... b5 35.Ke5 bxc4 36.Rd6 c3 37.Ke6 c2 38.Rd8+ Rxd8 39.exd8=Q#.

A.2) 29... b5 30.cxb5 and White has two pawns for the exchange and his knight can move to c4 now.

A.3) 29... Ke8 30.g3 with a plan similar to A.1.c.

Playing around with the position after 27. f5 Nc2+, I think 28. Kd3 is fine and appears to be more efficient than 28. Kf4. After 28...Nxa1, even thought the knight is trapped, I'd play 29. Rxa1 immediately, so that the rook can attend to more important matters, i.e. getting behind the passed pawns. That should lead to 29. Rxa1 Rae8 30. f6 Ke6 31. Rf1 Kf7.

This all comes from the famous "Excelsior" chess problem by Sam Loyd:

http://en.wikipedia.org/wiki/Excels...

The story goes that Sam used to bring his compositions to the chess club for vetting, and one of the players there claimed he could always pick out the principal mating piece in a chess problem. So Loyd composed this position and bet him that he couldn't pick out the man that DIDN'T deliver mate.

White to play (27.?) "Insane"

White is ahead a pawn (strongly poasted on e7), but Black is threatening 27...Nc2+ winning the exchange, and if White avoids that, then Black plays 27...Nc6 and 28...Nxe7 removing the strong pawn and restoring the balance. It was obvious to me that the idea here was pushing the f-pawn, since two connected pawns are worth more than the exchange. I guess the "insanity" is in the details!

27. f5! Nc2+

There is a choice here; Black can decline the rook by 27...Nc6 28. f6 Ne5, where the knight has a decent chance of blockading on f7 if necessary. This may be a better defense, but White still has targets of pawns on g4 and d6 which he can pursue, and if he wins one of those, his two extra pawns are on f6 and e7! That should be a winning advantage.

28. Kf4

This was my original plan, but for awhile I was looking at 28. Kd3, thinking that there was a reason to delay capturing the knight (since it would be trapped), but the king placement is more important (and the only reason to delay capture is if the f-pawn could advance to f7, which it can't).

28...Nxa1 29. Rxa1 Rae8 30. f6 Ke6 31. Nf1

Heading for e3, where it will attack the pawn on g4 and have the option of going to f5 or d5 (from where it can go to c7). The knight move also clears the d-file for the remaining rook to attack d6.

31...Bc8

Black can try giving back some material to win the advanced pawns, but White comes out on top after 31...Rg5 32. Bxg5 hxg5 33. Kxg5 Rg8+ (not 33...Bxe4? 34. Re1) 34. Kf4 Kxf6 35. Rd1 and 36. Rxd6+.

32. Ne3 Kf7 33. Rd1 Be6

There is no point in 33...Ke6 34. Nf5, with numerous threats.

34. Rxd6 and White will continue with 35. Ke5 with devastating effect, e.g. 34...Rb8 35. Ke5 Bc8 36. Nf5 Bxf5 (or 36...Rh8 37. Rd2, forcing the capture anyway) 37. Kxf5 and Black is helpless again 38. e5, 39. e6+ and 40. f7+.

http://en.wikipedia.org/wiki/Excels... >

Hi, <dzechiel>. Being of a restlessly inquisitive nature, I checked your Wikipedia article. It states that the name of the problem was based on the poem, Excelsior, by Longfellow. The Wikipedia article on the poem

[ http://en.wikipedia.org/wiki/Excels... ]

gives the first few lines:

The shades of night were falling fast,
As through an Alpine village passed
A youth, who bore, 'mid snow and ice,
A banner with the strange device,
Excelsior!

To come full circle, I recognized the openings lines, not from the poem, but from a parody by AE Housman that I first heard more than 25 years ago. It is also in Wikipedia

[ http://oldpoetry.com/opoem/62860-A-... ]

The shades of night were falling fast
And the rain was falling faster,
When through an Alpine village passed
An Alpine village pastor;

In this opposite-colored bishop endgame, white is a pawn up, but it is a very important, protected pawn that sits on the verge of promotion at e7. In fact, white has connected passed pawns and black can't stop the f4 pawn from marching to f6. Hence, my first impression on seeing the board was that white should play f5 immediately, even before I noticed that black can win the exchange with Nc2+. If white plays 27.Rac1?, Nc6 wins that critical pawn. So the analysis of this position centers around the theme of white's effort to promote the pawns and break the inevitable black blockade on the light squares. White has powerful grip on d5 that effectively shuts down the Bb7, a key factor in establishing the superiority of the white position. White starts with

27.f5!

This is a low-risk sacrifice if you consider the old rule-of-thumb "2 connected passed pawns on the 6th are at least equal to a rook". Pawns on the 6th and 7th should be even better. But can black establish an unbreakable blockade on f7? Let's consider acceptance of the sacrifice to be the main line, given that the black position is very passive anyway:

A) 27... Nc2+ 28.Kd3 Nxa1 29.f6

White is in no hurry to grab the Na1 (it is going nowhere) and black can not allow f7, winning two rooks.

29... Ke6 (Ke8 leads to the same position) 30.Rf1 Kf7 31.Bg3

White has two ways to break the blockade at f7: Maneuver the N to attack it (difficult) or advance the e4 pawn to e6 (easier). Now black faces a difficult choice:

A.1 31... Kd6 32.f7 Kxe7 (else 38.f8/Q wins both rooks) 33.fxg8/Q Rxg8 34.Rxa1 (finally) and white has an extra minor piece for a pawn with an easily won position, thanks to the isolated pawn targets at g4 and h6 and the grip on d5.

A.2 31... Rg5 (going for broke) 32.Bxd6 Rxh5 33.e5 Bxg2 34.Rf4 (34.e6+? Kxe6 35.f7 Rh3+ 36.Ke2 Bxf1+ 36.Nxf1 Kxf7) Rh3+ 35.Ke2 Bc3+ 36.Nxc3 Rxc3 37.e6+! Kxe6 38.f7! Rxf4 39.e8/Q+ Kxd6 (Rxe8 40.fxe8/Q+ Kxd6 41.Qb8+) 40.Qxa8 Rxf7 41.Qd5+ Ke7 42.Qe5+ followed by Qxa1 leaves white with Q vs R+3 pawns with winning chances but not a clear win.

White seems to be able to improve the above with a tempo gaining move:

A.2.1 33.Rf4! Rg8 (Rh1/h2?? 34.Rxg4 wins) 34.e5! Bxg2 35.e6+ Kxe6 36.f7 Rh3+ 37.Kf2 g3+ 38.Kxg2 Rh2+ 39.Kg1 and black's counterattack falls a tempo short in view of 40.fxg8(+)

28. Kd3 Nxa1 29. Rxa1 Bc6 30. f6 Ke6 31. Rf1 Be8 32. Rf5 g3 33. Nf1 Rg4