New Toy

by admin on January 26, 2020

Now I have both of my new toys — my new laptop and my new edition of Fritz 17 (together with ChessBase), so I am back to normal for blogging purposes.

I might as well use this post for a quick review of Fritz 17. I have to say that the change I was looking forward to the most is a complete non-starter for me. The new version of Fritz comes bundled with “Fat Fritz,” a new version of the engine that is roughly based on the open-source version of Google’s Alpha Zero, Leela Chess Zero (Lc0). I was totally looking forward to seeing variations evaluated in human terms: not as “White is ahead by 0.76 pawns,” but “White has a 48 percent chance of winning, 37 percent chance of drawing, and 15 percent chance of losing.”

However, Fat Fritz comes with some steep hardware requirement — you need a super-fast Nvidia graphics card. It looked to me as if a laptop equipped for Fat Fritz would cost at least $1500, maybe closer to $2000, and I couldn’t justify the expense.

The description of Fritz 17 on the website made it seem, however, as if Fat Fritz might run on a weaker graphics card, but with a dramatic reduction in performance. Even that would have been okay with me. But when I booted up Fritz 17, it gave me a choice of three engines (Fritz 17, Fritz 13 SE, and Stockfish 10 64-bit). Notice that you do not see Fat Fritz on this list. As far as I can tell, the program must have looked at my system, determined that it doesn’t have good enough hardware, and so it didn’t even give me the option to select Fat Fritz. This was a disappointment.

Aside from that, Fritz was pretty much the way I remember it. I don’t like it as much for analysis as I liked Rybka, because its evaluations are not stable enough. When it’s White’s move, Fritz’s evaluations tend to shift in White’s favor, and when it’s Black’s move they shift in Black’s favor. In my opinion a chess engine should somehow factor in the advantage of the move. If White is up by 0.25 pawns and Black makes a reasonable move, the evaluation should stay at 0.25 pawns instead of increasing to 0.50 pawns just because it’s now White’s turn.

I also don’t like playing against Fritz as much as I liked playing against Shredder. Of course I have to “dumb it down” a little bit to have a chance of winning, but somehow when I dumb down Fritz, its play becomes strategically horrible. So if I beat it, I don’t get the same pleasure as I would from beating a human (or even Shredder). And if I lose to Fritz, then I feel twice as bad because I’ve lost to an opponent that played a horrible strategic game.

An example is the very first game I played against Fritz in “Rated game” mode. I set its rating to 2006. Time control was 10 minutes plus 2 seconds per move.

Dana – Fritz 17

1. e4 d5 2. ed Qxd5 3. Nc3 Qe6+ (Ewww….) 4. Be2 Nc6 5. d3 …

Mouse slip! Actually, my new laptop has a touch-screen, so I don’t need a mouse any more. I can just touch the piece with my finger and drag it. This is a huge improvement! But this was my first game playing in this fashion, so I didn’t move the d-pawn quite far enough. Literally a fingerfehler (finger slip).

5. … g6 6. Be3 Bh6 7. Qd2 Nd4 8. O-O-O Bg7 9. Re1 h5?! (what’s this?) 10. Bd1 Qa6 11. a3 Qf6?

Up to now the computer and I have played equally badly, thanks in part to my accidental fifth move. But Fritz shows no comprehension of opening principles, moving its queen back and forth for no apparent reason. After 11. … Bf5 or 11. … Bg4 Black is okay.

12. Nf3 Ne6 13. d4 b6??

Position after 13. … b6. White to move.

FEN: r1b1k1nr/p1p1ppb1/1p2nqp1/7p/3P4/P1N1BN2/1PPQ1PPP/2KBR2R w kq – 0 14

If only I had an angel to tap me on the shoulder and say, “White to play and win!” In fact, White has at least two completely winning ideas. But it’s speed chess, so I rushed and played the first reasonable move that occurred to me.

14. Ne5?? …

A perfectly decent move under most circumstances. But much stronger was 14. d5! with the idea 14. … Nc5 15. Bd4, with a mighty skewer. And if that isn’t enough for you, there’s more. White also wins with 14. Nd5! with the idea of trying to trap Black’s queen with 14. … Qf5 15. c4 followed by Bc2, etc.

14. … Bb7 15. Bf3 O-O-O 16. Bxb7+ Kxb7 17. Qd3! …

A very strong move, threatening Qe4+. Fritz tries to stop this with trickery, but its position is too bad.

17. … Nc5 18. Qc4 …

I didn’t even see what Black could play here, as both the knight on c5 and the pawn on f7 are hanging. Well, the knight on c5 isn’t really hanging, but something much worse is: the Black king.

18. … Nh6

Typical for Fritz, taking care of the short-term problem but missing the long-term one.

19. dc Qxe5 20. c6+ …

I can’t even give this an exclam, because every human player would see it.

20. … Ka8 21. Qa6 Rb8

Position after 21. … Rb8. White to move.

FEN: kr5r/p1p1ppb1/QpP3pn/4q2p/8/P1N1B3/1PP2PPP/2K1R2R w – – 0 22

This time, even though I didn’t have a friendly angel to tap me on the shoulder and say “White to play and win,” I was able to figure it out for myself.

22. Nb5! …

I enjoy this sort of move. White tells Black, “Go ahead and have your kicks. Play a few checks if you want, but at the end of it you have to deal with the mating threats, Qxa7 mate and Nxc7 mate.” The only way for Black to stop both threats is to sacrifice his queen.

22. … Qxb2+ 23. Kd1 Qxb5 24. Qxb5 and White won.

The rest is mop-up.

Print Friendly, PDF & Email

{ 1 comment… read it below or add one }

brabo January 27, 2020 at 12:21 am

Did you try the free Lc0 which is very similar to Fat Fritz? See my article http://chess-brabo.blogspot.com/2019/09/leela-lc0.html for a guide about how to install it. Some games can be found on http://chess-brabo.blogspot.com/2019/09/testing-chess-engines.html

I am using a portable of about 1000 euro which includes the bit slower Nvidia GTX 1600i. Anyway Lc0 defeated on this portable Stockfish 10 with 54,5 – 46,6 so this cheaper pc is more than ok to analyze. I am sure if you would build your own PC that you can save 50%.

Reply

Leave a Comment

Previous post:

Next post: