MCG PRO Ending Weakness?

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
Post Reply
User avatar
fourthirty
Full Member
Posts: 763
Joined: Fri Dec 06, 2013 8:46 pm
Location: San Francisco

MCG PRO Ending Weakness?

Post by fourthirty »

To those who are much smarter than I...

I just finished Game 4 of my challenge between the MCG and MCG PRO (game posted in Tournament Results & Games forum). The MCG PRO finally played a strong and interesting game. However, I do have a question regarding the endgame.

Towards the end, the PRO (white) advances the "a" file pawn down for the promotion, which results in this position after move 57.a8=Q Kd7:
[fen]Q7/3k4/8/8/1Q4p1/1P4P1/6KP/8 w - - 1 58[/fen]

The PRO then plays 58.Qd4+, which results in a mate two moves later.
[fen]Q7/3k4/8/8/3Q2p1/1P4P1/6KP/8 w - - 2 58[/fen]

However, if the PRO had played 58.Qe4 instead of 58.Qd4+:
[fen]Q7/3k4/8/8/4Q1p1/1P4P1/6KP/8 w - - 2 58[/fen]

It would have resulted in a mate the next move via:
58...Kc7 59.Qec6#, or
58...Kd6 59.Qac6#

Instead, the game ended a move later with:

58.Qd4+ Ke6 59.Qg8+ Ke7 60.Qdd8#
[fen]3Q2Q1/4k3/8/8/6p1/1P4P1/6KP/8 w - - 6 60[/fen]

My question is - would the MCG PRO be expected to have played 58.Qe4, thus ending the game a move earlier? Is this considered a weakness in the endgame?

Or, was the MCG PRO already on a "Mate In" path, which it followed blindly, as long as the MCG played the expected moves (and since they both have the same engine, the moves were expected)?

Schach-computer states the MCG has 32 KB Hash Tables, but Hash is not listed for the MCG PRO. Was there simply too much material left on the board to take advantage of this?

Need help understanding...

Greg
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

fourthirty wrote:


Schach-computer states the MCG has 32 KB Hash Tables, but Hash is not listed for the MCG PRO.

Greg
For The Record..
MCGP has 160 kb Ram...which is about 5x more then the MCG

i tried the position on the MCGP's "mate finder" level and it plays 1.Qe4 instantly..the mate finder level finds the shortest possible mate
no idea why Lang's engine chooses a longer mate(by 1 ply) at 30 sec per move..perhaps his engine ..when in a normal game.. only searches for a mate ..any mate..and plays it..would i call that a "weakness"??
to me..mate is mate ..so not really..perhaps "nuance" is a better word
if it were a bug in the engine it would miss the move even on "mate finder"

i also entered the position into the MCGE(one is always looking for positions to enter into a piece recognition board.. to justify ones purchase)
it plays exactly the same as the MCGP
at 30 sec...1.Qd4 on mate finder..1.Qe4

Bugged By Mate Regards
Steve
donkeylane
Full Member
Posts: 679
Joined: Mon Aug 29, 2016 8:31 pm
Location: Cheshunt, Hertfordshire, UK

Post by donkeylane »

Know the feeling,those damn piece recognition systems eh?
User avatar
fourthirty
Full Member
Posts: 763
Joined: Fri Dec 06, 2013 8:46 pm
Location: San Francisco

Post by fourthirty »

Steve B wrote: For The Record..
MCGP has 160 kb Ram...which is about 5x more then the MCG
Ahhh, I see that now. Thanks.
Steve B wrote:perhaps his engine ..when in a normal game.. only searches for a mate ..any mate..and plays it..would i call that a "weakness"??
to me..mate is mate ..so not really..perhaps "nuance" is a better word
if it were a bug in the engine it would miss the move even on "mate finder"
Well, that is true.
Image
Steve B wrote:i also entered the position into the MCGE(one is always looking for positions to enter into a piece recognition board.. to justify ones purchase)
it plays exactly the same as the MCGP
at 30 sec...1.Qd4 on mate finder..1.Qe4

Bugged By Mate Regards
Steve
Interesting. Thanks for trying the position on the MCGE (don't have one of those yet - still saving). Glad I could help you justify!

Robert Frost Two Roads Regards,
Greg
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Steve B wrote: i tried the position on the MCGP's "mate finder" level and it plays 1.Qe4 instantly..the mate finder level finds the shortest possible mate
no idea why Lang's engine chooses a longer mate(by 1 ply) at 30 sec per move..perhaps his engine ..when in a normal game.. only searches for a mate ..any mate..and plays it..would i call that a "weakness"??
to me..mate is mate ..so not really..perhaps "nuance" is a better word
if it were a bug in the engine it would miss the move even on "mate finder"
Hi Steve,

I think it is not a bug. Lately I tried a mate in 3 search with a computer at analysis level and it comes with a mate also one ply longer. If I set it on "find mate in 3" then it finds the mate correctly. I do not see that as a bug. It is just that in the first case the machine stops searching because it is satisfied to find a winning move. It is very likely that until then it did not even try the move that leads to the shorter mate sequence.

Saving energy regards,
Paul
2024 Special thread: viewtopic.php?f=3&t=12741
2024 Special results and standings: https://schaakcomputers.nl/paul_w/Tourn ... 25_06.html
If I am mistaken, it must be caused by a horizon effect...
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

paulwise3 wrote:
Steve B wrote: i tried the position on the MCGP's "mate finder" level and it plays 1.Qe4 instantly..the mate finder level finds the shortest possible mate
no idea why Lang's engine chooses a longer mate(by 1 ply) at 30 sec per move..perhaps his engine ..when in a normal game.. only searches for a mate ..any mate..and plays it..would i call that a "weakness"??
to me..mate is mate ..so not really..perhaps "nuance" is a better word
if it were a bug in the engine it would miss the move even on "mate finder"
Hi Steve,

I think it is not a bug. Lately I tried a mate in 3 search with a computer at analysis level and it comes with a mate also one ply longer. If I set it on "find mate in 3" then it finds the mate correctly. I do not see that as a bug. It is just that in the first case the machine stops searching because it is satisfied to find a winning move. It is very likely that until then it did not even try the move that leads to the shorter mate sequence.

Saving energy regards,
Paul
Hi Paul
thats basically what i said
not a bug and probably the computer just makes the first mate move it sees
Agreed Regards
Steve
Post Reply