Solving the impossible

by Albert Silver
6/24/2021 – Recently an article went up on a colleague’s site that expounded on a series of positions that still eluded, not to mention baffled, modern engines today. The question that mattered was simply: is it true? In this article we take a look at some of these positions and give them to Fat Fritz 2. See the results!

ChessBase 16 - Mega package Edition 2021 ChessBase 16 - Mega package Edition 2021

Your key to fresh ideas, precise analyses and targeted training!
Everyone uses ChessBase, from the World Champion to the amateur next door. It is the program of choice for anyone who loves the game and wants to know more about it. Start your personal success story with ChessBase and enjoy the game even more.

More...

Before addressing the positions in question, it is important to point out in all fairness that the ultimate goal of the article was less about claiming computer fallacies than encouraging the readers to use their own brains, and not be afraid to be creative and fallible. Something we can only applaud and agree with. After all, try explaining to Tal that his chess is flawed. 

No player in chess history contributed more to defense technique than the great Tal. Yes, defense.

Nevertheless, comments such as "Don't ask your engine for help! It can't." should and must be checked. Computer chess and indeed computer hardware have evolved extraordinarily in the last decade alone. Just 4 years ago the idea of enjoying more than four cores in a desktop processor was reserved for the deep-pocketed enthusiasts alone, then along came the AMD Ryzen, breaking the stale status quo of Intel’s monopoly, and suddenly eight cores are barely mid-range. That same year saw the emergence of AlphaZero, on which the Leela project is based. As a result of all this, the number of positions that genuinely elude the wide swath of engines is becoming an exceptionally rarefied group.

The impossible positions

To capture or not to capture

The very first position shared by the author is a remarkable trap posed to a young 16-year-old Arshak Petrosian in 1970:

 

The future GM (and father-in-law of Peter Leko) looked at this flagrant queen offer and actually took it. This egregious blunder allowed Black to create an impenetrable fortress after 1. Nxb6 cxb6 2.h4 gxh4 3. Qd2 h3!! 4.gxh3 h4! and the players soon shook hands. Feel free to play these moves on the dynamic diagram above to see for yourself. 

However, with a cooler more circumspect look at the position, White would have quickly realized that instead of taking the queen, playing Qd2, then the king to a4 and knight to b3, the a5 pawn would have fallen and White would have won the game.

 

So how do the top engines fare? The esteemed writer is entirely correct!

Given the opportunity to capture on b6, they all fail to avoid the trap. Neither Fat Fritz 2, Leela, or the latest Stockfish (June 22 build), do any better after tens of billions of nodes on a solid 32-thread desktop. 

Chalk one up for human pattern recognition.

Thou shalt not enter

In this next position, unidentified, and probably composed, the knight and bishop contrive to keep the enemy queen from ever entering.

 

"Don't ask your engine for help! It can't." the reader is warned, but Fat Fritz 2 gives the solution almost instantly.

After 1. Bb3!! Qc8 (planning Qa8-a6) 2. Nd1! Qa8 3. Bc4!! Black cannot enter, since 3... dxc4 4. Nc3! and the doors are shut for good.

Domino effect

This second position with the same warning is also an example of a fortress, and is a composition by William Rudolph in 1912.

 

For a human, it takes very little time to see the very straightforward solution: 1. Ba4+!! Kxa4 2. b3+ Kb5 3. c4+ Kc6 4. d5+ Kd7 5. e6+ followed by f5 and that is that.

Here it depends on who you consult. Stockfish (June 22 build) stumbles on this one, and even after 11 billion nodes could not figure it out.

However, Fat Fritz 2 was able to produce the goods in under a minute.

You can't catch me

This next position is a fairly cute study by Josef Hašek in 1937.

 

The first moves are fine, but it is the finale that justifies the price of entry. 1. Kb1! (the engines won't even touch this move unless they see the solution) Kg7 2. Rh6!! This astounding move is needed to achieve the stalemate idea at the end. 2...Kxh6 Otherwise the rook will stay on the h-file and the game is an instant draw. 3. Kc1 Rh8 4. Kd1 Kg5 and here is the lovely clincher:

 

5. Ke2!! and now ...Rh1 is stalemate and if 5...Rh2 then 6. Kf1 Kg6 7. Kg1 Kg7 8. Kf1 and Black has no way of breaking in.

Needless to say, the engine is able to figure this one out quite quickly.

The greatest endgame move ever

Shirov's creativity throughout his career is the stuff of legend, and the Latvian genius has given us one of the all-time great games collections, Fire on Board, as well as the greatest endgame move ever.

 

This position arose from the game between Veselin Topalov and Alexei Shirov in 1998 in the classic Linares series. Every titled player worth his salt will know this position and move: 1... Bh3!!! If ever a move deserved three exclamation points, it is this one. The rest goes 48. gxh3 Kf5 49. Kf2 Ke4 50. Bxf6 d4 51. Be7 Kd3 52. Bc5 Kc4 53. Be7 Kb3 0-1

The author then states "chess engines will be baffled". Will they?

So what is the takeaway? Are engines now perfect and unquestioned? Yes, and no. Their ability to resolve the problems in the confines of the chessboard are infinitely superior to ours, but that does not mean you should blindly accept everything unquestioningly. 

"The true sign of intelligence is not knowledge, but imagination."

—Albert Einstein


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, and the content creator of the YouTube channel, Chess & Tech.