Page 1 of 10

The Busted Dutch

Posted: Thu Sep 12, 2019 8:20 am
by Lyudmil Tsvetkov
I want to post one more topic since my last essential topic message some couple hundred years ago.
Forget about hundred.

Some say I have been original. Is that really so?

Now, Harm will really go crazy, and also Henk, Eelco (how are you doing, buddy?) and (unfortunately) Houdart.
Matter of fact is though, the Dutch (Defence) is busted.
Or at least, far far from optimal.
What do my year-long investigations show?



It proves that, on above tabiya position, representing the Dutch, white is in a possession of a very potent and incisive continuation, namely 2. d5!, after which, if black is ot indeed altogether busted, he is very very close to throwing in the handkerchief, towel, I mean.
How long for your engines to "discover" the abslutely best move, 2. d5! ?
If they are unable to do so after a reasonable time, what scores and what lines would they exhibit after 2. d5 has been forced?
I guess a score of +30-40 centipawns in white's favour is an appropriate assessment of the real situation, but what scores do your pet engines show?

PS. In case you are unable to make sense why 2. d5 is optimal, git a copy of my new book to know the truth, lo-ol. (Sorry, I couldn't resist.)

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 8:28 am
by Guenther
Lyudmil Tsvetkov wrote:
Thu Sep 12, 2019 8:20 am
I want to post one more topic since my last essential topic message some couple hundred years ago.
Forget about hundred.

Some say I have been original. Is that really so?

Now, Harm will really go crazy, and also Henk, Eelco (how are you doing, buddy?) and (unfortunately) Houdart.
Matter of fact is though, the Dutch (Defence) is busted.
Or at least, far far from optimal.
What do my year-long investigations show?



It proves that, on above tabiya position, representing the Dutch, white is in a possession of a very potent and incisive continuation, namely 2. d5!, after which, if black is ot indeed altogether busted, he is very very close to throwing in the handkerchief, towel, I mean.
How long for your engines to "discover" the abslutely best move, 2. d5! ?
If they are unable to do so after a reasonable time, what scores and what lines would they exhibit after 2. d5 has been forced?
I guess a score of +30-40 centipawns in white's favour is an appropriate assessment of the real situation, but what scores do your pet engines show?

PS. In case you are unable to make sense why 2. d5 is optimal, git a copy of my new book to know the truth, lo-ol. (Sorry, I couldn't resist.)
yawn... make a bold statement again and again (with nothing behind) - a perpetuum mobile, see the link below from two years ago.
viewtopic.php?f=2&t=64605&p=723293&hilit=dutch#p723225

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 9:26 am
by AdminX
Hi Lyudmill,

I just downloaded the kindle version of your book and just finished reading the forward. :D Anyway I let Leela (J13B2-136) sit on your Dutch opening position and it likes 2. g3. After giving it 2. d5 it feels black has a 49% chance of winning with Nf6.

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 9:26 am
by zullil
Lyudmil Tsvetkov wrote:
Thu Sep 12, 2019 8:20 am

Matter of fact is though, the Dutch (Defence) is busted.
Or at least, far far from optimal.
What do my year-long investigations show?


2.d5 is not even in the top twenty according to Lc0 (network 320x24.J13-swa-410000) with 100M nodes. Note: In the output that follows, moves are ranked from worst to best.

Code: Select all

0911 20:32:17.108149 140051065321216 ../../src/chess/uciloop.cc:218] << info string c1h6  (69  ) N:    1427 (+ 0) (P:  0.84%) (Q: -0.95274) (D:  0.015) (U: 1.17295) (Q+U:  0.22021) (V: -0.9789) 
0911 20:32:17.108163 140051065321216 ../../src/chess/uciloop.cc:218] << info string e1d2  (105 ) N:    3510 (+ 0) (P:  0.92%) (Q: -0.30344) (D:  0.322) (U: 0.52364) (Q+U:  0.22020) (V: -0.3060) 
0911 20:32:17.108173 140051065321216 ../../src/chess/uciloop.cc:218] << info string g2g4  (378 ) N:   12881 (+ 0) (P:  1.45%) (Q: -0.00499) (D:  0.331) (U: 0.22529) (Q+U:  0.22030) (V:  0.0266) 
0911 20:32:17.108181 140051065321216 ../../src/chess/uciloop.cc:218] << info string f2f4  (351 ) N:   15443 (+ 0) (P:  1.56%) (Q:  0.01810) (D:  0.430) (U: 0.20221) (Q+U:  0.22031) (V:  0.0471) 
0911 20:32:17.108190 140051065321216 ../../src/chess/uciloop.cc:218] << info string e2e4  (322 ) N:   17331 (+ 0) (P:  1.56%) (Q:  0.03943) (D:  0.337) (U: 0.18085) (Q+U:  0.22028) (V:  0.1204) 
0911 20:32:17.108199 140051065321216 ../../src/chess/uciloop.cc:218] << info string b1a3  (34  ) N:   18498 (+ 0) (P:  1.62%) (Q:  0.04491) (D:  0.412) (U: 0.17539) (Q+U:  0.22030) (V:  0.1318) 
0911 20:32:17.108207 140051065321216 ../../src/chess/uciloop.cc:218] << info string d4d5  (761 ) N:   20406 (+ 0) (P:  1.56%) (Q:  0.06677) (D:  0.396) (U: 0.15352) (Q+U:  0.22029) (V:  0.1287) 
0911 20:32:17.108216 140051065321216 ../../src/chess/uciloop.cc:218] << info string b2b4  (234 ) N:   21020 (+ 0) (P:  1.75%) (Q:  0.05360) (D:  0.392) (U: 0.16670) (Q+U:  0.22031) (V:  0.0962) 
0911 20:32:17.108224 140051065321216 ../../src/chess/uciloop.cc:218] << info string d1d2  (81  ) N:   24952 (+ 0) (P:  1.78%) (Q:  0.07717) (D:  0.403) (U: 0.14313) (Q+U:  0.22030) (V:  0.1211) 
0911 20:32:17.108233 140051065321216 ../../src/chess/uciloop.cc:218] << info string f2f3  (346 ) N:   25508 (+ 0) (P:  1.74%) (Q:  0.08339) (D:  0.363) (U: 0.13690) (Q+U:  0.22029) (V:  0.1770) 
0911 20:32:17.108241 140051065321216 ../../src/chess/uciloop.cc:218] << info string c1e3  (63  ) N:   44942 (+ 0) (P:  1.99%) (Q:  0.13183) (D:  0.400) (U: 0.08848) (Q+U:  0.22031) (V:  0.1627) 
0911 20:32:17.108262 140051065321216 ../../src/chess/uciloop.cc:218] << info string a2a4  (207 ) N:   48056 (+ 0) (P:  2.10%) (Q:  0.13286) (D:  0.410) (U: 0.08745) (Q+U:  0.22031) (V:  0.1554) 
0911 20:32:17.108271 140051065321216 ../../src/chess/uciloop.cc:218] << info string b1d2  (33  ) N:   50630 (+ 0) (P:  2.22%) (Q:  0.13251) (D:  0.406) (U: 0.08781) (Q+U:  0.22031) (V:  0.2007) 
0911 20:32:17.108280 140051065321216 ../../src/chess/uciloop.cc:218] << info string h2h4  (403 ) N:   62292 (+ 0) (P:  2.17%) (Q:  0.15049) (D:  0.395) (U: 0.06982) (Q+U:  0.22031) (V:  0.2065) 
0911 20:32:17.108288 140051065321216 ../../src/chess/uciloop.cc:218] << info string c1d2  (57  ) N:   64891 (+ 0) (P:  2.42%) (Q:  0.14566) (D:  0.407) (U: 0.07466) (Q+U:  0.22031) (V:  0.1802) 
0911 20:32:17.108296 140051065321216 ../../src/chess/uciloop.cc:218] << info string c2c3  (259 ) N:   68999 (+ 0) (P:  2.24%) (Q:  0.15515) (D:  0.418) (U: 0.06516) (Q+U:  0.22031) (V:  0.1830) 
0911 20:32:17.108304 140051065321216 ../../src/chess/uciloop.cc:218] << info string b2b3  (230 ) N:   77283 (+ 0) (P:  2.33%) (Q:  0.15994) (D:  0.415) (U: 0.06037) (Q+U:  0.22031) (V:  0.1804) 
0911 20:32:17.108314 140051065321216 ../../src/chess/uciloop.cc:218] << info string b1c3  (36  ) N:   82494 (+ 0) (P:  2.32%) (Q:  0.16390) (D:  0.400) (U: 0.05641) (Q+U:  0.22031) (V:  0.2372) 
0911 20:32:17.108323 140051065321216 ../../src/chess/uciloop.cc:218] << info string g1h3  (161 ) N:   85118 (+ 0) (P:  2.36%) (Q:  0.16487) (D:  0.396) (U: 0.05544) (Q+U:  0.22031) (V:  0.1824) 
0911 20:32:17.108332 140051065321216 ../../src/chess/uciloop.cc:218] << info string d1d3  (86  ) N:   89777 (+ 0) (P:  2.31%) (Q:  0.16876) (D:  0.369) (U: 0.05154) (Q+U:  0.22031) (V:  0.2438) 
0911 20:32:17.108341 140051065321216 ../../src/chess/uciloop.cc:218] << info string h2h3  (400 ) N:  122776 (+ 0) (P:  2.61%) (Q:  0.17766) (D:  0.405) (U: 0.04266) (Q+U:  0.22031) (V:  0.1957) 
0911 20:32:17.108350 140051065321216 ../../src/chess/uciloop.cc:218] << info string a2a3  (204 ) N:  124274 (+ 0) (P:  2.56%) (Q:  0.17911) (D:  0.411) (U: 0.04121) (Q+U:  0.22031) (V:  0.1847) 
0911 20:32:17.108366 140051065321216 ../../src/chess/uciloop.cc:218] << info string e2e3  (317 ) N:  167966 (+ 0) (P:  2.61%) (Q:  0.18916) (D:  0.408) (U: 0.03115) (Q+U:  0.22031) (V:  0.1820) 
0911 20:32:17.108374 140051065321216 ../../src/chess/uciloop.cc:218] << info string c1g5  (67  ) N:  307262 (+ 0) (P:  8.95%) (Q:  0.16194) (D:  0.379) (U: 0.05837) (Q+U:  0.22031) (V:  0.2063) 
0911 20:32:17.108382 140051065321216 ../../src/chess/uciloop.cc:218] << info string c1f4  (65  ) N: 1152437 (+ 0) (P:  8.51%) (Q:  0.20553) (D:  0.385) (U: 0.01478) (Q+U:  0.22031) (V:  0.2275) 
0911 20:32:17.108390 140051065321216 ../../src/chess/uciloop.cc:218] << info string g1f3  (159 ) N: 6811983 (+ 0) (P:  9.30%) (Q:  0.21644) (D:  0.435) (U: 0.00274) (Q+U:  0.21917) (V:  0.2218) 
0911 20:32:17.108398 140051065321216 ../../src/chess/uciloop.cc:218] << info string c2c4  (264 ) N: 13234059 (+ 0) (P: 11.33%) (Q:  0.21723) (D:  0.422) (U: 0.00172) (Q+U:  0.21894) (V:  0.2166) 
0911 20:32:17.108406 140051065321216 ../../src/chess/uciloop.cc:218] << info string g2g3  (374 ) N: 77012590 (+ 8) (P: 16.88%) (Q:  0.21988) (D:  0.414) (U: 0.00044) (Q+U:  0.22031) (V:  0.2371) 
0911 20:32:21.551563 140051073713920 ../../src/mcts/search.cc:384] Stopped search: Reached visits limit: 100000056>=100000000

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 10:13 am
by reflectionofpower
AdminX wrote:
Thu Sep 12, 2019 9:26 am
Hi Lyudmill,

I just downloaded the kindle version of your book and just finished reading the forward. :D Anyway I let Leela (J13B2-136) sit on your Dutch opening position and it likes 2. g3. After giving it 2. d5 it feels black has a 49% chance of winning with Nf6.
I agree because even though it looks like d5 will cramp black up white over extends himself and black will only pick apart the d5 pawn and other subsequent fortifying moves.

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 10:22 am
by OneTrickPony
How is 30-40 centipawns "busted"?
Black is struggling in both the mainline and Bf4 setups. It's very hard to get anything below +1 for black after 2.Bf4 and the resulting positions are very painful and passive. Even if 2.d5 is good you have to beat the other options for it to be worth it and that's a very high bar in case of the opening as bad as Dutch.

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 11:29 am
by zullil
AdminX wrote:
Thu Sep 12, 2019 9:26 am
Hi Lyudmill,

I just downloaded the kindle version of your book and just finished reading the forward. :D Anyway I let Leela (J13B2-136) sit on your Dutch opening position and it likes 2. g3. After giving it 2. d5 it feels black has a 49% chance of winning with Nf6.
Yes, after 2.d5, Lc0 suggests that 2...Nf6 immediately equalizes:

Nodes: 100M, N/s: 21.5k, Time: 1h 17m (limit met)

49.6% Nf6 Nc3 e6 a3 Qe7 Bg5 c6 dxe6 Qxe6 Nf3 d5 e3 Nbd7 Ne2 Ne4 Ned4 Qf7 Be2 a6 O-O g6 Bf4 Bg7 Nd2 O-O c4 Nxd2 Qxd2 dxc4 Rac1 b5 b3 c5 Nf3 c3 Rxc3 Bxc3 Qxc3 Bb7 Rd1 Qe7 Bh6 Rfe8 Bg5 Qe6 a4 Bc6 h4 Rac8 Bf1 Bxf3 gxf3 c4 bxc4 bxa4 Kg2 Nc5 Rd5 Qf7 Bf4 Rc6 Qa3 Qe7 Bg5 Qf8 Bf4 Rec8 Be5 Qe7 Bd4 (N: 98.9M, P: 32.7%)

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 12:48 pm
by AdminX
Following the analysis of your 4th variation from your book. Leela prefers 3. Nc3 here over 3.c4

Image

And really digs 4. ... a5 over 4. ... d6

Image

See what you made me do. :wink: Now back to game one. :lol: For the record I play humans so all ideas are welcome here.

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 3:29 pm
by Peter Berger
Lyudmil Tsvetkov wrote:
Thu Sep 12, 2019 8:20 am
I want to post one more topic since my last essential topic message some couple hundred years ago.
Forget about hundred.

Some say I have been original. Is that really so?

Now, Harm will really go crazy, and also Henk, Eelco (how are you doing, buddy?) and (unfortunately) Houdart.
Matter of fact is though, the Dutch (Defence) is busted.
Or at least, far far from optimal.
What do my year-long investigations show?



It proves that, on above tabiya position, representing the Dutch, white is in a possession of a very potent and incisive continuation, namely 2. d5!, after which, if black is ot indeed altogether busted, he is very very close to throwing in the handkerchief, towel, I mean.
How long for your engines to "discover" the abslutely best move, 2. d5! ?
If they are unable to do so after a reasonable time, what scores and what lines would they exhibit after 2. d5 has been forced?
I guess a score of +30-40 centipawns in white's favour is an appropriate assessment of the real situation, but what scores do your pet engines show?

PS. In case you are unable to make sense why 2. d5 is optimal, git a copy of my new book to know the truth, lo-ol. (Sorry, I couldn't resist.)
Your claim simply makes no sense for simple and logical reasons.

There are three main systems for Black in the Dutch: the Stonewall, Iljin-Genewski and the Leningrad.

It could be argued that 2.d5 puts a stop to the Stonewall, for most obvious reasons ;). But then, the Stonewall isn't even the most popular system in the Dutch, and general consensus is that white has several ways to reach a tiny but long-lasting advantage.

But if we assume that black wants to reach one of the other setups, 2.d5 is certainly very helpful from black's point of view. And even if you want to really make this move, you could always do it at a later point in time, black won't try to stop it anyway.

Can you give some general idea on why this reasoning is supposed to be wrong?

Peter

Re: The Busted Dutch

Posted: Thu Sep 12, 2019 4:02 pm
by Lyudmil Tsvetkov
AdminX wrote:
Thu Sep 12, 2019 9:26 am
Hi Lyudmill,

I just downloaded the kindle version of your book and just finished reading the forward. :D Anyway I let Leela (J13B2-136) sit on your Dutch opening position and it likes 2. g3. After giving it 2. d5 it feels black has a 49% chance of winning with Nf6.
Hello dear.
2. g3 is a good move (to fianchetto the bishop), but unfortunately it will allow 2...d5, after which white has hardship getting an edge.
On 2...Nf6, 3. c4 (to anchor d5) is very strong.
I don't expect of engines of this generation to grok all the subtleties, as their eval is subpar, but anyway.

Thanks for the purchase!
I'll eat some vegetables today.
If another purchase comes, I'll buy meat, lo-ool.