Page 2 of 2

Re: Lomonosov - what is the mate-distance...

Posted: Fri Sep 27, 2019 3:20 pm
by Paloma
It was once reported here that syzygy-DTM is under development.

How far is this project? Or has it been given up?

Does anyone know more? Ronald de Man?

Re: Lomonosov - what is the mate-distance...

Posted: Fri Sep 27, 2019 3:34 pm
by Nordlandia
If i recall correctly, Ronald hinted that Syzygy DTM was under testing during christmas time in 2017. The project was postponed for unknown reasons. Cfish did support DTM locally on his machine afaik.

Re: Lomonosov - what is the mate-distance...

Posted: Fri Sep 27, 2019 8:07 pm
by Ovyron
Paloma wrote:
Fri Sep 27, 2019 10:08 am
Online Syzygy is a mess. It's crap.
Not the shortest way to mate.

In your pgn 11.Nb1? is -8# (11...g3 8#), best is 11.Kb2 -12# (6 men Nalimov),
your 12.Nd2 (-6#) lost another move, best: 12.Kb2 -7# (12...g2 7#)
13.Nb1? best is 13.Kb1
After 13.Nb1 is g1Q mate in 4, not your 13...h4 (mate in 7)

In your pgn after 15.Nb1 is h2 wrong, best g1Q+
After your h2 is 16.Nc3 best, not your Nd2
18.Nb1 wrong, best is 18.Kb2 Qc1+ 19.Ka2 Qxd2#

Therfore i asket to Lomonosov. :)
o_O

Are the people at https://syzygy-tables.info aware of this? Are they planning into fixing it? They should at least put a disclaimer in their main page that their results might not be reliable. Anybody visiting their site would wrongly believe they're the ultimate resource on this.

Re: Lomonosov - what is the mate-distance...

Posted: Fri Sep 27, 2019 10:00 pm
by MonteCarlo
It's not that the syzygy tables are "wrong" simpliciter or unreliable.

They're DTZ, not DTM, and DTZ-optimal play will very often not be DTM-optimal.

Since the OP wanted to know shortest forced mate, offering a DTZ-optimal solution from syzygy didn't help him much :)

Re: Lomonosov - what is the mate-distance...

Posted: Sat Sep 28, 2019 12:35 am
by Ovyron
Oh right, I've already made my opinion clear...
Ovyron wrote:
Mon Jul 29, 2019 9:47 pm
Vinvin wrote:
Mon Jul 29, 2019 6:54 pm
Click on the book icon and you will see the distance to conversion
Really? You call that better??

Check this position out:



What does the 7men Lomonosov app say?

Image

Oh, okay, so Bf4!! and Kf5!! are best, because they mate in 18 and other moves waste time.

What does lichess say?

Image

"Derp! Be3+ has best distance to conversion! After it capturing the Bishop is white's worst move! Let's give the Bishop up for no reason at all!"

:shock:

Why would someone recommend a site that suggests moves as bad as Leela's on the endgame, that doesn't know how many moves to mate. It's beyond me.

Distance to conversion sucks.
We still need a site that allows you to check distance to mate :(

Re: Lomonosov - what is the mate-distance...

Posted: Sat Sep 28, 2019 12:00 pm
by syzygy
Ovyron wrote:
Fri Sep 27, 2019 8:07 pm
Paloma wrote:
Fri Sep 27, 2019 10:08 am
Online Syzygy is a mess. It's crap.
Not the shortest way to mate.

In your pgn 11.Nb1? is -8# (11...g3 8#), best is 11.Kb2 -12# (6 men Nalimov),
your 12.Nd2 (-6#) lost another move, best: 12.Kb2 -7# (12...g2 7#)
13.Nb1? best is 13.Kb1
After 13.Nb1 is g1Q mate in 4, not your 13...h4 (mate in 7)

In your pgn after 15.Nb1 is h2 wrong, best g1Q+
After your h2 is 16.Nc3 best, not your Nd2
18.Nb1 wrong, best is 18.Kb2 Qc1+ 19.Ka2 Qxd2#

Therfore i asket to Lomonosov. :)
o_O

Are the people at https://syzygy-tables.info aware of this? Are they planning into fixing it? They should at least put a disclaimer in their main page that their results might not be reliable. Anybody visiting their site would wrongly believe they're the ultimate resource on this.
The information there is totally correct. There was never a claim that DTZ gives the shortest mate, which obviously it does not. What it does do is win every winnable position and draw every drawable position, even taking into account the 50-move rule (which Lomonosov does not).

Re: Lomonosov - what is the mate-distance...

Posted: Sat Sep 28, 2019 11:21 pm
by Ovyron
I see, so there's different schools of thought (if a mate in 51 by Lomonosov might actually be a 50-move draw, perhaps you have to use DTZ instead until you can be sure DTM is correct.) This is kind of like Chess philosophy, not even when the game is solved people can agree on what's the best course of action.