Fat Fritz: Small differences, big effect

by Albert Silver
10/9/2019 – Sometimes seemingly innocuous questions can lead to the most interesting discoveries, and such was the case after deep analysis by ChessBase 15's “Tactical Analysis” with Fat Fritz on one of the best games from the World Cup. The positional masterpiece played by Nikita Vitiugov over Wesley So was compared to Karpov at his best. While the analysis was without question excellent, it was what was not explained that led to fascinating explorations and beautiful moves. | Photo: Kirill Merkuryev, khantymansiysk2019.fide.com

ChessBase 15 - Mega package ChessBase 15 - Mega package

Find the right combination! ChessBase 15 program + new Mega Database 2020 with 8 million games and more than 80,000 master analyses. Plus ChessBase Magazine (DVD + magazine) and CB Premium membership for 1 year!

More...

A tiny change makes a big Fat (Fritz) difference

In some games, such as the spectacular win by Jan-Krysztof Duda over American wunderkind Jeffery Xiong, the correct moves are fairly well-defined and, as a result, both engines Stockfish 10 and Fat Fritz produced similar analysis, noting where the mistakes were and the correct way to proceed. Although the game was undoubtedly one of the most impressive from the World Cup, it did not yield any unusual revelations.

On the other hand, a far more complex positional struggle like the epic battle between Nikita Vitiugov and Wesley So unearthed many pearls just waiting to be discovered. A warm thanks to GM Karsten Mueller for flagging this game for an article.

Because of the strongly positional nature of the game, it is sometimes hard to know what is really a matter of the best move, and what is more a question of taste or style.

A key moment that stood out was at move 27, when not only did Stockfish and Fat Fritz completely disagree on the correct continuation, but the suggestion by Fat Fritz seemed to defy logic. See for yourself below:

 

In this position Vitiugov played 27. Qf1, a move that Stockfish wholeheartedly agrees with.

His idea is clear: he plans to interpose Be2, and pile on the pressure on the b5 pawn. Seems logical enough. The problem is that Fat Fritz thinks the move leads to +1.90, while its choice of 27. Kh2 leads to +3.50! That is no small difference.

You might now be expecting a tale of how Stockfish was wrong here and Fat Fritz was right, but not so. For one thing, as is readily admitted, counter examples no doubt abound, and suffice it to say, yes, after hours of analysis with Stockfish 10 chugging along at 30 million nodes per second as it tried to defend Black, Fat Fritz proved it was able to win with its choice.

Still, it was in one of those lines with a few beautiful continuations that some of the explanations for its mysterious choice came to the surface. It was a lesson in how a small difference can lead to dramatic results.

Let us first look at Fat Fritz's recommended move and the pertinent lines, and then we will examine the move actually played and the continuation that Fat Fritz recommended as So’s best reply. In both examples you will see one beautiful move that actually occurs, but followed by a difference:

 

That is just a preview though. See the full line in the game viewer below. Pay attention to the curious line in which a failed Qc6 is brought up. It will be shown again after, and the second time with an enormous difference.

[Event "FIDE World Cup 2019"] [Site "Khanty-Mansiysk"] [Date "2019.09.20"] [Round "4.1"] [White "Vitiugov, Nikita"] [Black "So, Wesley"] [Result "1-0"] [ECO "C43"] [WhiteElo "2732"] [BlackElo "2767"] [SetUp "1"] [FEN "1r5k/p2rb2p/Pq2b1p1/1pppPp2/5P2/2P2B1P/1RP1QBP1/R6K w - - 0 27"] [PlyCount "29"] [EventDate "2019.09.07"] [EventType "k.o."] [EventCountry "RUS"] [SourceTitle "playchess.com"] [Source "ChessBase"] {[#]} 27. Kh2 $142 {3.55/18} d4 28. Bb7 b4 29. cxb4 cxb4 30. c3 $3 {[#]} b3 31. cxd4 Rxd4 {Forced} ({Not taking the pawn with a move like} 31... Qb4 {fails to} 32. d5 $1 Bxd5 33. e6 $1 Rd6 34. Qe5+ Kg8 35. Bd4 $18 {and Black must give up the queen to prevent mate.}) 32. Rc1 $1 {This lovely maneuver around the back to harass the queen impresses.} Ba3 {This is Black's only attempt to confuse the issue. But....} 33. Rc6 $1 {is decisive.} Qxc6 $4 {would be a major mistake, since it does get Black to queen the pawn, but lose the king in the process (i.e. get mated). This line is presented as the main line, not because it is Black's best choice, but because it is in this continuation that one justification for Fat Fritz's choice of 27.Kh2 will become clear. Play through the line to the end as it is worth seeing.} ({Correct would be} 33... Qb4 34. Bxd4 Qxd4 35. Rb1 {though Black is still in a lot of trouble.}) 34. Bxc6 Bxb2 35. Qxb2 $18 Rc4 36. Bxa7 Rc2 37. Qd4 b2 38. Bxb8 b1=Q {Black gets to realize his dream of a queen, but that is all she wrote, since now White finishes off with} 39. Qd8+ {and Black gets mated.} Kg7 40. Qf6+ Kh6 41. Bd6 {with the unstoppable threat of Bf8+ followed by mate} 1-0

This time we look at the same line, almost, except 27.Qf1 has been played instead and the queen is no longer on e2. Watch what a difference it makes!

[Event "FIDE World Cup 2019"] [Site "Khanty-Mansiysk"] [Date "2019.09.20"] [Round "4.1"] [White "Vitiugov, Nikita"] [Black "So, Wesley"] [Result "1-0"] [ECO "C43"] [WhiteElo "2732"] [BlackElo "2767"] [SetUp "1"] [FEN "1r5k/p2rb2p/Pq2b1p1/1pppPp2/5P2/2P2B1P/1RP1QBP1/R6K w - - 0 27"] [PlyCount "16"] [EventDate "2019.09.07"] [EventType "k.o."] [EventCountry "RUS"] [SourceTitle "playchess.com"] [Source "ChessBase"] {[#]} 27. Qf1 {1.98/21 This position is nearly identical to the previous line analyzed above, with the key difference being the queen and king.} d4 {1.98/21 is now much stronger than before as will be seen.} 28. Bb7 b4 29. cxb4 cxb4 { and here too White has the spectacular looking} 30. c3 $1 {which exploits two pins on the black queen. And just as before Black continues} b3 {Except here, the continuation} 31. cxd4 {does not work here due to that seemingly small but vital difference: the queen is on f1 and not e2. See how the line ends compared to the one above!} (31. Qd3 dxc3 32. Qxc3) 31... Rxd4 32. Rc1 Ba3 33. Rc6 {And here is where we see the difference. Now *Black* wins with} Qxc6 $1 34. Bxc6 Bxb2 {and the Queen is not on e2 to recapture.} 1-0

While the point of the main lines might seem to be far deeper than any subtlety you would encounter in your own games, remember the next time you are studying a position that small differences can sometimes have powerful consequences.

Links




Born in the US, he grew up in Paris, France, where he completed his Baccalaureat, and after college moved to Rio de Janeiro, Brazil. He had a peak rating of 2240 FIDE, and was a key designer of Chess Assistant 6. In 2010 he joined the ChessBase family as an editor and writer at ChessBase News. He is also a passionate photographer with work appearing in numerous publications.
Discussion and Feedback Join the public discussion or submit your feedback to the editors


Discuss

Rules for reader comments

 
 

Not registered yet? Register

macauley macauley 11/4/2019 08:59
@Nezhmetdinov1919 - See all stories on Fat Fritz
Nezhmetdinov1919 Nezhmetdinov1919 10/12/2019 07:30
If I missed something I apologize in advance, but as far as I'm concerned before we can really enjoy any of those articles, Chessbase still needs to do some explaining regarding 'his' (?) so called Fat Fritz. Silence in face of the accusations that have been made looks suspicious and -- I must admit -- although I really respect the Chessbase staff I'm a bit disappointed about their handling of all this hype. All in all, I hope you guys can came clean, remember that we're all a big family and that we want to keep liking Chessbase as member of our chess family.
LuckyLuke LuckyLuke 10/10/2019 01:14
27. Qf1 seems to force d4 because of the Thread Be2. After 27.Kh2 black has better options than d4. What is the big deal after 27.Kh2 Kg7. Now after 2. Qf1 d4 seems not be +3 because you can later sac the exchange on b7. Maybe Qe1?
chinguaro chinguaro 10/9/2019 11:12
When Fritz was a chess program with no possibilities and Stockfish was number one, nobody talked about Stockfish on this website, only about Houdini and Komodo that were programs for sale. Now instead there is interest in continually talking about how bad Stockfish is and how great Fat Fritz is, quite pathetic all this.
ortsac2014 ortsac2014 10/9/2019 07:39
Running Lc0 wT160 (2 cores) and Stockfish 10 (10 cores) simultaneously on my Aquarium, both found 27.Kh2 after about 10 min. 27.Qf1 was considered 2nd best.
maac2002 maac2002 10/9/2019 03:36
My Stockfish (20190801) choose 27. Kh2 as best at depth 42 (albeit with only +1.73 [+1.50 for Qf1])
thirteen thirteen 10/9/2019 12:57
I do think it would have been some 'nice touch' to also actually show the Fat Fritz phot0 shoot, with its analysis and numbers, at the move 27 stage. Just saying. Details guys, they're the stuff of interest!
1