Roberto Grau analyzed this game without the benefit of computer proofreaders, and we can't criticize his otherwise elucidating notes, but we think whiteshark's observation is very well spotted.

When we thought about using this as a Sunday puzzle, our computers immediately discovered the 24...Qxb7! line as the critical defense, which cast doubts on the entire combination beginning with 24.Nd6!?. White doesn't gain a piece after 24...Qxb7 although perhaps it is still , we honestly don't know. Even with computer help it is very difficult to analyze.

Two major branches here:

26. Nxd8 Bxd1 and when black disentangles his pieces he is about a pawn ahead. From here 27. Kf1 is -0.66, 27. Re5 is -0.99 and 27. Nf7 g6! is -1.32

26. Re7 (-0.91) with a complicated double rook and minor piece position. Black has more pawns but white has more activity.

So can we really call this game "immortal" if the main talking point (26. Nd6) is flawed?

So to answer the question "what day of the week would this be suitable for?" I am afraid for me the answer is none. CG.com computer checks its puzzles and would spot the holes in this combination within seconds.

Let's be clear. It's a fun game to play through. White pursues his attack with energy and imagination. I certainly wish I had played it! And many other immortal or evergreen games have tactical holes in them which are spotted in the computer age, so perhaps we shouldn't be too harsh.

If <26.Nxd8 Bxe1> and black has still a pawn surplus.> Well after 26.Nxd8 Bxe1 White could play Nf7, winning the rook on h8.>>

...g6 gain the f5-rook and trap the knight. Or lose a knight for a pawn. Not a good idea.

<21.Nb3> Qxc3 22.Bxf5 Qxe1 23.Be6+ Qxf1+ 24.Kxf1 Ke8 25.Qb5+ would have been winning for white.

<21...dxc3+> 22.Kh1 cxd2 23.Be4+ (or Be6+) Nf5 would be winning for black due to the passed pawn threats.

Nd6 is such a cold-blooded move. If I was thinking about it OTB, my temperature would rise (after rejecting capture after capture) at levels that would interfere with the calculations, and would waste much time.

Unfortunately, the line spotted by <whiteshark> with 24.Nd6 Qxb7 25.Nxb7 Bxc3 etc kills it stone dead as a forced win: so the problem is 'cooked'.

It is still a fascinating ending. Engines seem to give black a slight edge -- if he can hold onto the d-pawn and safely play ...c5 then the combined passers are very strong. But it's not clear that black *can* hold the pawns: there are lines where white centralizes his king quickly rather than trying for tricks with Ne6 (or the downright bad Nf7) -- eg, 25.Nxb7 Bxc3 26.Nxd8 Bxe1 27.Kf1 Bd2 28.Ke2 Bf4 29.g3 g6 ... which is highly unclear.

Or white can regain a pawn with something like (in this same line): 29.Ne6 h5 30.Nxf4 gxf4 31.Rxf4 Rh6! -- but black is winning this: he gets some initiative, frees the Rh8, and creates a potential mini-pawnroller. Which looks as though it can survive and prosper.

It is clear that there is no extra tactical finesse which pulls a win out of the hat for white after the queen swap on b7.

And yet White is not simply lost. There are ideas such as keeping rooks and minor pieces on the board, activating the white king, and playing to win one or two pawns down. At the least, white has a few tempi to attack with + before black gets the Rh8 out.

It's not always about counting pawns. To paraphrase Mae West, it's about the life in your pawns rather than the pawns in your life ...

In any case, annotator Grau and both players seem to have missed this resource -- for which Brazil can be grateful.

The classic English sci-fi writer John Wyndham wrote a good book, The Outward Urge, in which Brazil becomes a space-faring 21st century superpower after the northern hemisphere bombs or pollutes itself out of play. Brazil in that book has a slogan which applies neatly to chess as well:

click for larger view

<[-0.48] d=28 1.Nxd8> Bxe1 2.Kf1 g6 3.Rf7 Bb4 4.Rxc7 Ba5 5.Rd7 Bxd8 6.Rxd8 Kg7 7.Rd7 Kh6 8.Rxa7 d3 9.h3 Rc8 10.Rd7 Rc3 11.a4 Ra3 12.g4 Rxa4 13.Rxd3 Ra2 14.Kg1 Rb2 15.Rc3 Rd2 16.Rc7 (6:09.28) 6936548kN

[-0.92] d=27 1.Re7 Rc8 2.Rxg5 d3 3.Rd5 h6 4.Rxd3 Bf6 5.Re4 c5 6.Rc4 Rc7 7.Na5 Kh7 8.Nb3 Rb8 9.Rd2 Rb4 10.Rdc2 Rxc4 11.Rxc4 Kg6 12.Kf2 Be7 13.Kf3 Kf5 14.Re4 h5 15.h3 h4 16.Rg4 (5:55.55) 6679861kN

<[-0.49] d=32 1.Nxd8 Bxe1 2.Kf1> g6 3.Rf7 Bb4 4.Rxc7 Ba5 5.Rd7 Bxd8 6.Rxd8 Kg7 7.Rd7 Kh6 8.Rxa7 d3 9.Rd7 Ra8 10.Rxd3 Rxa2 11.h3 Kh5 12.Rd5 Ra4 13.Kg1 Kh4 14.Kh2 Ra3 15.Rd7 h6 16.Rd4 (82:13.09) 96912622kN

[-0.99] d=31 1.Re7 Rc8 2.Rxg5 d3 3.Rd5 c5 4.Rxd3 Bf6 5.Re2 h6 6.Rc2 Rc7 7.Na5 Kf7 8.Kf2 Rhc8 9.Nc4 Bd4 10.Kf3 Rf8 11.Re2 Kg8 12.Kg4 Rf6 13.Ra3 Kh7 14.Rf3 Kg6 (78:30.50) 92786425kN

I dont know if that is strong. But for sure, taking the knight with any piece on board is a mistake.