Testposition and HUGE fail high from Stockfish!!

[chrisw]

Poor Uri on an ignore list. It's an attempt at humor right? He was merely paraphrasing Chris. (Ignore lists on this forum are a bad idea I think. Soon nobody will be left..)

Oh, I thought it was Uri Blass. But it is not apparently? This is getting a bit confusing. Too many Uri's. Uri Blass, Uri Averny, Uri Geller.

paraphrasing me? you are joking, right?

1. I think Stockfish is probably the strongest engine there is.
2. I think it is stronger than Komodo and Leela-Zero.
3. I wouldn't assert it is stronger or weaker than AZ. Because. Nobody knows.
4. Free open sourced engines with good quality control will always be better than commercial ones. Because talent magnifier.

Difficult for you to be more wrong, actually. And WTF am I getting mentioned at all? Seems old programmers can't post on forums without getting somehow, what should I call it, do you think?

Did you take my post in this thread as "anti-Stockfish"? How bizarre. It was an intelligent and honest analysis of Stockfish's reasons for missing the move, based on actual study of actual Stockfish code by an actual person who can actually understand both at direct and meta level. I didn't notice anyone else making any smart analysis here, did you? It's seen as negative criticism? That is really bizarre. It's actually a piece of smart analysis of "What is to be done" to counter the NNs.

Obviously one should stick to just pouring out long computer mainlines as substitute for thinking. Then you can reply with some other long main line. Main line dueling. Really great communications. Not. Make sure you copy the previous ones into your post each time.

You are confusing me with the mass of critics who just offer negative shite. I offer positive smarts. Can't tell the difference? I think I go join Ed. Ridiculous.

[zullil]

CiChess on my phone finds mate in 12 after 4 minutes and ~1 billion nodes, gradually going down from +10 to -mate.
Eelco - truth. I never put anyone on ignore list, whatever they write, it's cheap

That's impressive. This search by Cfish-dev examined about 5 billion nodes:

info depth 61 seldepth 24 multipv 1 score mate 12 nodes 5059040296 nps 46834292 hashfull 573 tbhits 117653085 time 108020 pv d8c7 e8c8 g6f7 g8h8 c7e5 c8c5 e5b2 a8c7 b2a1 a5a4 a1b2 a4a3 b2a1 a3a2 a1b2 a2a1r b2a1 c7b5 f6e6 b5c3 a1c3 c5c3 f7f8q

No sign yet of any mate-in-11.

[zullil]

Did you take my post in this thread as "anti-Stockfish"? How bizarre. It was an intelligent and honest analysis of Stockfish's reasons for missing the move, based on actual study of actual Stockfish code by an actual person who can actually understand both at direct and meta level. I didn't notice anyone else making any smart analysis here, did you? It's seen as negative criticism? That is really bizarre. It's actually a piece of smart analysis of "What is to be done" to counter the NNs.

FWIW, I found your post to be as you describe it above---an intelligent and honest analysis. I simply replied with output from Stockfish-dev in an attempt to quantify just how difficult this position is for Stockfish.

[zullil]

Thanks for coming out of the woods after a while.
Now my ignore list finally seems complete.

Great reply. Made me laugh.

[chrisw]

Did you take my post in this thread as "anti-Stockfish"? How bizarre. It was an intelligent and honest analysis of Stockfish's reasons for missing the move, based on actual study of actual Stockfish code by an actual person who can actually understand both at direct and meta level. I didn't notice anyone else making any smart analysis here, did you? It's seen as negative criticism? That is really bizarre. It's actually a piece of smart analysis of "What is to be done" to counter the NNs.

FWIW, I found your post to be as you describe it above---an intelligent and honest analysis. I simply replied with output from Stockfish-dev in an attempt to quantify just how difficult this position is for Stockfish.

That's fine, no problem from my part there. I'm griping about the other guy who thinks it just ok to drag me into some other nonsense argument in which I have no part at all, and based on misreading and taking some kind of misplaced offence. It was a negative misinterpretation, false linking me with some idiotic ideas I obviously don't ascribe to in the first place, I don't tolerate it.

[Eelco de Groot]

My apologies for dragging your wellintended post through the mud. 'Paraphrasing' was not a good word and I knew that, but I could not find anything better in a hurry. Put me on ignore. File complaints. Please don't leave on my account. Too many people have already left this forum. It would be great to hear something from Cristophe Théron for instance. What happened to Bruce Moreland all these many years ago.

Constructive criticism of Stockfish or for my part anything criticising Stockfish is fine, as long as there is an idea behind it that can be defended. It is after all a discussion forum and the impact of Stockfish, or for my part of computer chess, on chess itself is very large. I find I compare it with the impact AI may have on us and I really don't know if that is a good thing or an abomination. But discussing that trancends this post maybe in which I wanted to apologize to you. If you see or talk to Ed somewhere please pay him my regards.

Eelco

[Uri]

I'm not very smart or gifted but I believe that photographic memory plays a very very large role in being very very good at chess.

I believe that if someone doesn't have a very good memory he will never be a very good chess player. He/she may be good at other things in life but not at chess. Someone with Alzheimer or who has mental retardation will never be good at chess.

Many people in the world suffer from neurodegenerative diseases (neurodegenerative diseases are diseases of the brain) and I think this could at least partially explain why they find chess to be so difficult.

A good brain is very crucial for chess. I hope that future advances in cognitive neuroscience and cognitive psychology could explain why some people are so bad at chess.

Memory is very important in chess because you need to remember a very large number of positions and there are 10 to the power 120 of legal positions in chess. That's an astronomical number.

A Grandmaster probably remembers a very large number of tactical positions in addition to a lot of strategic information about the game.

Magnus Carlsen and Garry Kasparov (and many others btw) both have a photographic memory and that's why they are both so good at this game.

A supercomputer has tons of nonvolatile memory (a lot of nonvolatile memory can store a lot of chess knowledge and chess information almost forever) and this is why supercomputers are so strong at this game.

[Uri]

I think that memory and pattern recognition (pattern recognition requires memory but I may be mistaken) are more important in chess than intelligence.

For example Albert Einstein was very intelligent but not so good at chess so it means that chess is not just intelligence, it's more than that.

[Uri Blass]

I think that memory and pattern recognition (pattern recognition requires memory but I may be mistaken) are more important in chess than intelligence.

For example Albert Einstein was very intelligent but not so good at chess so it means that chess is not just intelligence, it's more than that.

1)With no memory there is no intelligence and intelligence is dependent on memory(you cannot answer a long question if you cannot memorize what is the question)


2)Identifying the patterns is clearly part of intelligence.
If you have only good memory you can memorize many positions but not identify that another position that you see has the same pattern.


Here is a simple example
The following positions are the same pattern (and I know without calculation of all possible moves of the king that they are mate) but without a minimal intelligence the human who memorize that the first is mate does not understand that the second is mate.
[d]4k3/4Q3/4K3/8/8/8/8/8 b - - 1 1
[d]8/8/8/8/8/kQK5/8/8 b - - 1 1

Note that there is also a danger of wrong identification of a pattern and 2 positions may be similiar but not the same.

3)You need to spend a significant time on chess to become a really strong chess player.
I do not think that Albert einstein spent a significant time on chess.
Maybe he could be a strong chess player in case of deciding to use most of his time for chess and we will never know.

4)being stronger in chess is not only about identifying patterns.
Chess players think during the game and it is important to know to think correctly to take advantage of your time better.

With the same pattern recognition knowledge
players who are able to follow correct thinking rules in chess are going to do better than players who forget to do it during the game.

For example one rule is always to consider checks and captures and I think that many weak players are going to perform better if before they do a move they use some time to check if there are captures or checks that they miss and they may even know about the rule but forget to try to follow it during the game.

[Uri Blass]

If the target is to compose positions when it happens then it is easy
Here is an example from -114 to mate in 8.

[d]2k5/ppp5/8/8/NNBK4/QQQR3q/QQQB2q1/QQQR4 b - - 0 1

Stockfish_18082021_x64_modern:
1/1 00:00 819 410k -114.55 1. ... Qhg4+ 2.Kc5
2/3 00:00 2k 771k -114.58 1. ... Qf2+ 2.Be3
3/6 00:00 3k 1,513k -114.85 1. ... Qh8+ 2.Ke3 Qxc3 3.Nxc3
4/8 00:00 5k 1,826k -114.37 1. ... Qh8+ 2.Ke3 Qhh3+ 3.Kf4 Qhg4+ 4.Ke5 Q4e4+ 5.Kf6
5/10 00:00 12k 2,477k -116.14 1. ... Qhg4+ 2.Bf4 Qxc2 3.Be6+ Kb8 4.Qxc7+ Qxc7 5.Qxc7+ Ka8 6.Bxg4
6/12 00:00 17k 2,877k -104.57 1. ... Qh8+ 2.Kc5 b6+ 3.Kb5 Qe8+ 4.Rd7 Qxd7+ 5.Nc6 Qgxc6+ 6.Kb4 Kb7
7/12 00:00 18k 2,998k -104.32 1. ... Qh8+ 2.Kc5 b6+ 3.Kb5 Qe8+ 4.Rd7 Qxd7+ 5.Nc6 Qgxc6+ 6.Kb4 Kb7 7.Qh8
8/17 00:00 26k 3,206k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
9/13 00:00 33k 3,258k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
10/15 00:00 35k 3,146k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
11/13 00:00 48k 3,422k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
12/19 00:00 58k 3,604k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
13/15 00:00 74k 3,346k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
14/12 00:00 88k 3,518k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
15/19 00:00 107k 3,552k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
16/15 00:00 145k 3,444k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
17/14 00:00 187k 3,594k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
18/17 00:00 258k 3,588k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
19/18 00:00 323k 3,632k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
20/19 00:00 412k 3,642k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
21/19 00:00 554k 3,876k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
22/25 00:00 689k 4,007k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
23/24 00:00 902k 4,197k 0.00 1. ... Qh8+ 2.Ke3 Qh6+ 3.Kd4
24/23 00:00 1,360k 4,431k +M8 1. ... Qh8+ 2.Ke3 Qg5+ 3.Ke4 Qh7+ 4.Kd4 Qhg7+ 5.Ke4 Q7g6+ 6.Kd4 Q6f6+ 7.Ke4 Qff5+ 8.Kd4 Qe5+

Well not really this is the target, since engines have to keep for some time the "losing" score before they go to opposite evaluation.
In my position it keeps it for 1.6 billion nodes in yours it finds in a fraction of a second and 26000 nodes.

In this case try this trivial mate in 7 that I composed

[d]kb1b1b1b/b1b1b1b1/B1N4b/8/8/8/8/4K3 w - - 0 1

Stockfish_18090410_x64_modern:
1/1 00:00 412 206k -28.99 1.Kf1
2/2 00:00 534 267k -28.87 1.Kf1 Bec5
3/4 00:00 862 431k -28.89 1.Nxd8 Bb4+ 2.Kf1 Bxd8
4/5 00:00 2k 1,001k -29.10 1.Kf1 Bad4 2.Nxd8 Bcxd8
5/6 00:00 4k 1,905k -29.29 1.Kf1 Bad4 2.Bd3 Bed6 3.Nxd8
6/6 00:00 8k 2,514k -29.33 1.Kf1 Bad4 2.Bd3 Bhg5 3.Nxd8 Bcxd8
7/10 00:00 14k 2,781k -29.30 1.Bd3 Kb7 2.Nxd8+ Bcxd8 3.Be4+ Kc7 4.Kf1 Bad4 5.Kg2
8/10 00:00 25k 3,568k -29.44 1.Bd3 Bh4+ 2.Ke2 Bdg5 3.Kf1 Baf2 4.Kg2 B8c5 5.Be4
9/11 00:00 36k 3,555k -29.40 1.Bd3 Bh4+ 2.Ke2 Bdg5 3.Be4 Bac5 4.Ne7+ Ka7 5.Nf5 Bb4 6.Nxh4
10/11 00:00 56k 3,308k -29.43 1.Bd3 Bh4+ 2.Kf1 Bdg5 3.Be4 Baf2 4.Ne7+ Ka7 5.Nf5 Bfd6
11/14 00:00 142k 3,639k -29.63 1.Kf1 Bad4 2.Kg2 Bb4 3.Kf3 Bd8f6 4.Kg2 Bbd6 5.Nxd4 Bxd4 6.Bc4 Bfe7
12/16 00:00 388k 4,038k -29.77 1.Kd1 Bgd4 2.Kc2 Bhe5 3.Bf1 Kb7 4.Bg2 Kb6 5.Nxd4 Bxd4 6.Bf1 Bce5 7.Be2
13/19 00:00 619k 4,453k -30.00 1.Kd1 Bgd4 2.Ke2 Bh8g7 3.Bd3 Kb7 4.Be4 Kb6 5.Kd3 Bhe3 6.Ke2 Bge5 7.Kf1 Beh6 8.Nxd4 Bxd4
14/20 00:00 864k 4,697k -30.02 1.Kd1 Bgd4 2.Ke2 Bhe5 3.Kf3 Bhe3 4.Kg2 Bab6 5.Kf1 Be7d6 6.Kg2 Bd8e7 7.Nxd4 Be3xd4 8.Bf1
15/21 00:00 1,165k 4,793k -30.03 1.Kd1 Bgd4 2.Ke2 Bhe5 3.Bd3 Kb7 4.Be4 Kb6 5.Kd1 Be7d6 6.Kc2 Bd8e7 7.Bg2 Bhe3 8.Nxd4 B3xd4 9.Kb3
16/20 00:00 1,794k 5,142k -30.12 1.Kd1 Bgd4 2.Ke2 Bhe3 3.Bd3 Kb7 4.Be4 Kb6 5.Kd1 Bhg7 6.Kc2 Bge5 7.Kd1 Bfg7 8.Kc2 Be7d6 9.Kb3
17/20 00:00 2,562k 5,371k -30.13 1.Kd1 Bgd4 2.Ke2 Bhe3 3.Kf3 Bhe5 4.Kg2 Bb4 5.Bc8 Bfd6 6.Kf1 Bab6 7.Kg2 B8e7 8.Nxb4 Bxb4 9.Bh3
18/21 00:00 3,126k 5,436k -30.11 1.Ke2 Bgd4 2.Kd3 Bhe3 3.Bc8 Bhe5 4.Kc2 Bfg7 5.Ba6 Bcd6 6.Kb3 Bd8c7 7.Bc8 Bab6 8.Ba6 B3g5 9.Nxd4 Bbxd4
19/24+ 00:03 20,760k 5,886k -29.37 1.Ke2
19/24+ 00:04 28,021k 5,948k -29.03 1.Ke2
19/24+ 00:05 32,160k 5,984k -28.58 1.Ke2
19/24+ 00:05 33,335k 5,993k -28.00 1.Ke2
19/24+ 00:05 34,923k 6,011k -27.26 1.Ke2
19/27+ 00:08 49,308k 6,128k -26.30 1.Ke2
19/27 00:14 93,851k 6,262k -25.98 1.Ke2 Bge5 2.Kd3 Bad4 3.Kc4 Bb4 4.Nxb4 Ka7 5.Bc8 Bxb4 6.Kxb4 Be7+ 7.Kc4 Bhe3 8.Bf5 Bhg7 9.Be4 Bcd6 10.Kb3 Kb6
20/16+ 00:15 99,140k 6,279k -25.89 1.Ke2
20/16+ 00:16 103,350k 6,290k -25.80 1.Ke2
20/25+ 00:20 130,440k 6,331k -25.67 1.Ke2
20/25+ 00:26 166,155k 6,329k -25.49 1.Ke2
20/25+ 00:32 208,042k 6,312k -25.24 1.Ke2
20/25+ 00:37 237,467k 6,325k -24.90 1.Ke2
20/25+ 00:41 262,348k 6,344k -24.45 1.Ke2
20/25+ 00:42 267,901k 6,348k -23.86 1.Ke2
20/25+ 00:44 283,516k 6,366k -23.11 1.Ke2
20/28 00:56 359,481k 6,405k -22.33 1.Ke2 Bh2 2.Kd3 Bc3 3.Kc4 Ba3 4.Kb5 Bh8e5 5.Nxd8 Bbc7 6.Nc6 Bfg7 7.Bc8 Bg5 8.Be6 Bab8 9.Bd5 Beg3 10.Ka6 Ba7
21/20+ 01:00 388,156k 6,419k -22.24 1.Ke2
21/20- 01:01 393,694k 6,422k -22.77 1.Ke2 Bgf6
21/20+ 01:01 394,153k 6,422k -22.28 1.Ke2
21/20+ 01:03 411,259k 6,429k -22.10 1.Ke2
21/28- 01:07 432,932k 6,436k -23.24 1.Ke2 Bce5
21/28+ 01:07 433,395k 6,435k -22.37 1.Ke2
21/28- 01:16 488,080k 6,396k -23.68 1.Ke2 Beg5
21/28+ 01:16 488,512k 6,396k -22.75 1.Ke2
21/28+ 01:16 489,659k 6,395k -22.00 1.Ke2
21/30 01:35 607,576k 6,357k -21.77 1.Ke2 Bd2 2.Kxd2 Bce5 3.Kd3 Bgf6 4.Kc4 Bb4 5.Nxb4 Bxb4 6.Kxb4 Bbd6+ 7.Kc4 Bad4 8.Bc8 Bhg7 9.Be6 Ka7 10.Bd5 B8b6 11.Kb5 Kb8 12.Kc4
22/28- 01:36 612,733k 6,355k -21.86 1.Ke2 Bd2
22/28+ 01:36 613,042k 6,356k -21.77 1.Ke2
22/28- 01:39 631,409k 6,351k -24.33 1.Ke2 Bce5
22/28+ 01:39 631,866k 6,351k -21.79 1.Ke2
22/29+ 01:55 726,740k 6,317k -21.54 1.Ke2
22/29 01:59 752,902k 6,318k -21.65 1.Ke2 Bh4 2.Kd3 Bb4 3.Nxb4 Bc5 4.Nc6 Bb4 5.Nxb4 Ka7 6.Bc8 Bde7 7.Nc6+ Kb6 8.Bd7 Bcd6 9.Ke2 Bd4 10.Kf1 Kc7 11.Nxb8 Kxb8 12.Be6 Bhe5 13.Kg2
23/31+ 01:59 755,916k 6,318k -21.56 1.Ke2
23/31+ 02:09 815,949k 6,303k -21.47 1.Ke2
23/31+ 02:19 881,203k 6,300k -21.34 1.Ke2
23/31+ 02:33 967,993k 6,295k -21.16 1.Ke2
23/31+ 02:49 1,069,305k 6,312k -20.91 1.Ke2
23/31+ 03:07 1,184,940k 6,319k -20.57 1.Ke2
23/31+ 03:13 1,225,500k 6,326k -20.13 1.Ke2
23/31+ 03:14 1,228,230k 6,326k -19.54 1.Ke2
23/31+ 03:30 1,329,993k 6,314k -18.79 1.Ke2
23/31+ 04:19 1,634,973k 6,300k -17.84 1.Ke2
23/33+ 05:36 2,100,650k 6,242k -16.62 1.Ke2
23/33+ 06:08 2,301,451k 6,243k -15.07 1.Ke2
23/34 06:51 2,570,369k 6,251k -13.68 1.Ke2 Bd2 2.Kxd2 Bb4+ 3.Kd3 Bgd4 4.Kc4 Bce5 5.Nxd8 Bab6 6.Nc6 Bd8 7.Nxd8 Bbc7 8.Nc6 Bba5 9.Kb5 Bhf6 10.Nxa5 Bxa5 11.Kxa5 Bfd6 12.Bc8 Ka7 13.Kb5
24/24- 06:51 2,571,379k 6,251k -14.92 1.Ke2 Bd2
24/24+ 06:53 2,584,376k 6,252k -13.68 1.Ke2
24/26- 07:02 2,640,287k 6,247k -16.85 1.Ke2 Bb4
24/26+ 07:02 2,641,360k 6,247k -14.28 1.Ke2
24/26+ 07:02 2,641,681k 6,247k -14.02 1.Ke2
24/26+ 07:03 2,644,675k 6,247k -13.69 1.Ke2
24/34+ 07:18 2,737,198k 6,241k -13.24 1.Ke2
24/34+ 07:22 2,762,259k 6,245k -12.66 1.Ke2
24/34+ 07:23 2,771,366k 6,242k -11.89 1.Ke2
24/34+ 07:31 2,824,007k 6,249k -10.93 1.Ke2
24/34+ 07:38 2,869,429k 6,253k -9.71 1.Ke2
24/34+ 07:47 2,923,419k 6,259k -8.17 1.Ke2
24/34+ 07:47 2,927,486k 6,259k -5.50 1.Ke2
24/34+ 07:51 2,955,565k 6,263k -3.03 1.Ke2
24/34+ 08:45 3,462,747k 6,595k +0.09 1.Ke2
24/34+ 08:45 3,464,577k 6,596k +4.00 1.Ke2
24/34+ 08:45 3,464,981k 6,596k +8.97 1.Ke2
24/34+ 08:45 3,469,188k 6,596k +16.05 1.Ke2
24/34+ 08:48 3,489,894k 6,599k +24.07 1.Ke2
24/34+ 08:50 3,503,654k 6,600k +36.72 1.Ke2
24/34+ 08:51 3,504,420k 6,599k +M21 1.Ke2
24/34 09:33 3,886,390k 6,777k +M9 1.Ke2 Bd2 2.Kxd2 Bh4 3.Kd3 Bb4 4.Kc4 Baf2 5.Kxb4 Bc3+ 6.Kb5 Bce1 7.Bc8 Bc5 8.Ka6 Bba7 9.Bb7+
25/16 09:35 3,905,622k 6,783k +M7 1.Ke2 Bd2 2.Kd3 Bda5 3.Kc4 Bgc3 4.Kb5 Bhd4 5.Bc8 Bfg7 6.Ka6 Bg3 7.Bb7+
26/14 09:41 3,941,668k 6,782k +M7 1.Ke2 Bd2 2.Kd3 Bda5 3.Kc4 Bgc3 4.Kb5 Bhd4 5.Bc8 Bfg7 6.Ka6 Bg3 7.Bb7+

Edit:Note that I made the problem relatively easy.
If the white king is at h1 it is mate in 8 and I could also put the black bishops to have more options in black squares(in my position bishops at b8 d8 f8 h8 cannot move and it make it easier for chess programs to solve).

Edit 2:I could even make it mate in 9 by havin Bf1-a6 as first move.