An opening blunder by Mephisto Chess Explorer (GK-2000)

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
Mentat
Member
Posts: 68
Joined: Sun Nov 02, 2014 1:11 am
Location: Nis, Serbia

An opening blunder by Mephisto Chess Explorer (GK-2000)

Post by Mentat »

The other day I played a blitz game (game in 10, sudden death vs Chess Explorer, reputedly a clone of GK-2000) and it made a horrible mistake on its move 7:

[Event "blitz sudden death at 10 min"]
[Site "?"]
[Date "8.11.2014"]
[Round "1"]
[White "Djordje"]
[Black "Mephisto Chess Explorer, H8, 10mhz"]
[Result "1-0"]

1. e4 e6 2. d4 d5 3. e5 c5 4. c3 Nc6 5. Nf3 Qb6 6. Bd3 cxd4 7. cxd4 Nxd4?? 8. Nxd4 Bc5 9. Be3 Qxb2 10. Nd2 Bxd4 11. Qa4+ Bd7 12. Qxd4 Qxd4 13. Bxd4... etc.

Now, Chess Explorer is supposedly a version of Fritz 1 which I do not have and I would love to see if someone has a slow 386 (16mhz or so) who might be able to check whether Fritz 1 is prone to making the same blunder. I also tested my Centurion (a clone of GK-2100 at the same lavel and it cooly played Bd7 avoiding the crass mistake). I do not own a Windows machine (only Macs and a Linux box) so I can't do it :(

Let us try to see which programs tend to overlook this simple threat (possibly by using the multimove feature leading them into this position in the French). Thanks in advance for the possible input.

Horizon regards,

Djordje
User avatar
Sargon1972
Member
Posts: 348
Joined: Sun Jul 14, 2013 9:30 am
Location: Dussen

Post by Sargon1972 »

Hello Djordje


Is this played out of book? i know even the Magallan have a opening bug
If it is a calculated move , it can be that the computer have experienced the fun level bug of many Morsch programs :lol:

Kr,Hans
Hello
I am from Netherlands and many years involved with computerchess
It,s a way of life i suppose :p

Kr,Hans
User avatar
Mentat
Member
Posts: 68
Joined: Sun Nov 02, 2014 1:11 am
Location: Nis, Serbia

Post by Mentat »

Sargon1972 wrote:Hello Djordje


Is this played out of book? i know even the Magallan have a opening bug
If it is a calculated move , it can be that the computer have experienced the fun level bug of many Morsch programs :lol:

Kr,Hans
Hi Hans. Thanks for replying. The Chess Explorer *was* out of book (the book was set to active, I think) and was calculating, I am sure. I am also sure that it wasn't the infamous "fun level bug" as I always take care to set the computer at the right level and press "clear" afterwards. Actually, in order to win the game I had to play pretty good moves as it was trying to wiggle out of the trouble and set some traps (after all it was only a game in 10 minutes).

I'll look up the Magellan book bug when I have enough time, thanks for pointing that out.

Good night.

Djordje
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Hi Djordje,
I have the Saitek Turbo Advanced Trainer, which is also a clone of the GK2000.
I tried the 10 minute blitz mode with your moves. It started right away with the French defense. After 5. .. Qb6 it was out of book. It took its time for the next moves, and played the decent 7. .. Qb7.

A tip to check how close a clone it is: recently I tried some positions from the BT-2450 test (see the wiki elo liste).
The four positions I tried, gave exactly the same results as for the GK2000:
BT- 8: 85 sec.
BT-15: 7 sec.
BT-21: 27 sec.
BT-29: 57 sec.

Regards, Paul.
User avatar
Mentat
Member
Posts: 68
Joined: Sun Nov 02, 2014 1:11 am
Location: Nis, Serbia

Post by Mentat »

paulwise3 wrote:Hi Djordje,
I have the Saitek Turbo Advanced Trainer, which is also a clone of the GK2000.
I tried the 10 minute blitz mode with your moves. It started right away with the French defense. After 5. .. Qb6 it was out of book. It took its time for the next moves, and played the decent 7. .. Qb7.

A tip to check how close a clone it is: recently I tried some positions from the BT-2450 test (see the wiki elo liste).
The four positions I tried, gave exactly the same results as for the GK2000:
BT- 8: 85 sec.
BT-15: 7 sec.
BT-21: 27 sec.
BT-29: 57 sec.

Regards, Paul.
Hi Paul,

well, that's interesting. You meant 7...Bd7, I guess. You started me thinking and there could be many reasons for the discrepancy in the moves of the two machines. The one that seems to me the most plausible is that the slightest of differences in the move execution time (by me and you) could have lead to 7... Bd7 . Another one could be the mode of operation of the machines --- the Chess Explorer runs on batteries only, while you may have had your machine plugged in, via adapter, etc.

And thanks for the BT tip. I will test the Explorer as soon as I can on the positions you listed.

rgds.

Djordje
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Hi Djordje,
I have my machine runnning on batteries too, and they are in it running for more then 8 hours at least :-). With the permanent brain running, the move execution time could play a role, but I doubt it.
Anyway, I will try it again shortly.

Paul.
User avatar
Mentat
Member
Posts: 68
Joined: Sun Nov 02, 2014 1:11 am
Location: Nis, Serbia

Post by Mentat »

paulwise3 wrote:Hi Djordje,
I have my machine runnning on batteries too, and they are in it running for more then 8 hours at least :-). With the permanent brain running, the move execution time could play a role, but I doubt it.
Anyway, I will try it again shortly.

Paul.
I just found enough time to run the BT 2450 set of four positions:

BT 2450 no.8: 71 sec +0.3
BT 2450 no.15: 7 sec +0.7
BT 2450 no.21: 21 sec +0.0
BT 2450 no.29 : 31 sec +0.9

It seems that the Chess Explorer is a GK-2000 clone as the results are pretty much similar, but note that they are *not* identical... The issue of the Chess Explorer blundering away that piece on d4 in the closed French remains moot, I guess.
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Hi Djordje,

That looks good, but I am a little surprised by your timing of BT-29. With that position, for my machine I noticed that after a while it came to the right move, but a little later it switched again to another move. Only after 57 seconds it came back to the right move and stayed there. Are you sure you did not overlook that?

As for the blunder, to be sure I will test it one more time tomorrow.
User avatar
Mentat
Member
Posts: 68
Joined: Sun Nov 02, 2014 1:11 am
Location: Nis, Serbia

Post by Mentat »

paulwise3 wrote:Hi Djordje,

That looks good, but I am a little surprised by your timing of BT-29. With that position, for my machine I noticed that after a while it came to the right move, but a little later it switched again to another move. Only after 57 seconds it came back to the right move and stayed there. Are you sure you did not overlook that?

As for the blunder, to be sure I will test it one more time tomorrow.

Paul,

to tell you the truth, I may have been a little rash on this last position. The machine did change its mind a couple of times, finding the right move, then going elsewhere and the first time it settled on Ke1 and had a plus score I jotted down the time. When/if I have the time to go through the same position once again I will report. One more thing I did was to try to reproduce the original blunder on my Chess Explorer, but this time at game in 15. As was only to be expected, the CE avoided the losing move and played instead Bb4+, followed by Bd7 next. However, what I found interesting was that the CE didn't seem to have taken more time for the good move by playing Bb4+ than at the faster level (G in 10)?

Anyway, lots of fun with these little plastic toys that still do not fail to keep me interested.

Dj.
User avatar
Mentat
Member
Posts: 68
Joined: Sun Nov 02, 2014 1:11 am
Location: Nis, Serbia

Post by Mentat »

paulwise3 wrote:Hi Djordje,

That looks good, but I am a little surprised by your timing of BT-29. With that position, for my machine I noticed that after a while it came to the right move, but a little later it switched again to another move. Only after 57 seconds it came back to the right move and stayed there. Are you sure you did not overlook that?

As for the blunder, to be sure I will test it one more time tomorrow.
Paul,

just ran BT2450 #29 again: 43 sec, Ke1 +0.9. This time I am positive that the time is correct. Thanks for helping me get it right.

Dj.
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Djordje,

Thanx for your testing, very interesting! It seems that your machine runs faster then the "original" GK2000 clones. And comparing the specs that makes sense, because (looking at the specs in the wiki elo liste) your machine has a 16-bit processor, and the GK2000 and my machine have an 8-bit processor. So your machine should be a little stronger :-)

Paul.
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Djordje,

I just retried your game at the 10 minutes level, and I'm ashamed to say that my machine now makes the same blunder...
I retried again, waiting one or two minutes before playing 7. cxd4, but still it played Nxd4?. Maybe this time I waited too long ;-).
Interesting though is, that further on it did NOT play 10. .. Bxd4, but 10. .. f6 instead.

To make it up to you, I did some extrapolation on your BT results. Conservatively estimating an average of 10% to 15% faster solution time of all 16 solved positions (and who knows solving 1 or 2 extra positions!?), your machine has a BT_2450 rating between 1892 and 1899, where the GK2000 scores 1876.
(note that in the formula BT = 2450 - 2 x GZ, GZ is expressed in minutes)

And you're right, it is fun playing around with these toys ;-). I've become an adept only a few months ago!
Post Reply