logo Welcome, Guest. Please Login or Register.
December 13, 2019, 05:21AM
Home Help Search Login Register
News: Links to other chess computer sites at www.chesscomputers.org

  Chess Computer Community
  General Category
  Discussions about Chess Computers
(Moderators: Endspielgott, Overtom, bobosse)
  Senator module and repetition draws
previous next
Pages: [1] Reply Notify of replies Send the topic Print 
   Author  Topic: Senator module and repetition draws  (Read 1110 times)
Robert Weck
Hero Member
*****

Offline

Posts: 791



I love Boardnation!

View Profile
Re:Senator module and repetition draws
« on: January 18, 2006, 02:16AM »
Reply with quote


Quote from: PaulM on January 16, 2006, 07:06PM   

The program immediately showed Qg5 when I hit the [info] key.  It never wavered from that candidate move until it finally played it.

the same here with my Milano pro! On 30min/game he played Qg5 after <1 min.


Robert
Report to moderator Logged
Ernest Gundel
Guest

E-Mail
Re:Senator module and repetition draws
« Reply #1 on: January 16, 2006, 09:09PM »
Reply with quote

I'm guessing the problem may be the result of the Senator's relationship with it's big brother the Magellan, which has a slightly faster clock speed and (more importantly) a hash table.

I suspect the repetition recognition logic included the hash table. Every position in a game thus far would be inserted in the hash table with a value of zero, so if a search found that position, it would recognize it as leading to a draw. Maybe Saitek simply removed the hash table without putting in some other kind of logic to recognize draws.
Report to moderator Logged
exabyte
Jr. Member
**

Offline

Posts: 86



Chess is the Best

View Profile
Re:Senator module and repetition draws
« Reply #2 on: January 16, 2006, 07:06PM »
Reply with quote

Hi Jaco,

The program immediately showed Qg5 when I hit the [info] key.  It never wavered from that candidate move until it finally played it.

Regards
Paul
Report to moderator Logged
Jaco
Guest

E-Mail
Re:Senator module and repetition draws
« Reply #3 on: January 16, 2006, 11:38AM »
Reply with quote

Hi Paul,

Did your Mephisto Master immediately show Qg5 as the response (if you press "info" button while computer is thinking) or does it only switch to Qg5 later (you mention it played Qg5 after 6:30 only). I'd like to know whether it was considering another move before it chose Qg5.

In the case of my Mephisto Chess Challenger, the computer immediately chose Qg5 as it's favourite move, and never deviated from it - that shows that it immediately recognised the repetition draw from King moves.

Yes, you are correct in that the Master and the Senator use identical hardware and identical programs - that could mean that a mistake slipped in somewhere in the compilation of the senator's software.

I'm going to email Saitek on this issue (yeah right, as if I'm expecting to hear anything from them on this). But what the hell, let's at least see what they say :-)

Regards,
Jaco
Report to moderator Logged
exabyte
Jr. Member
**

Offline

Posts: 86



Chess is the Best

View Profile
Re:Senator module and repetition draws
« Reply #4 on: January 16, 2006, 09:00AM »
Reply with quote

In this position the Mephisto Master(Milano Pro which I thought was the same as the Senator module) after 6:30, the computer played Qg5.  Maybe it's not the exact same program.
Report to moderator Logged
drawmeister
Full Member
***

Offline

Posts: 162

How about a nice game of chess ?

View Profile E-Mail
Re:Senator module and repetition draws
« Reply #5 on: January 15, 2006, 12:51PM »
Reply with quote

Jaco,

No problems with the Magellan, as fast as i could see the info display it was showing Qg5 and giving itself a nice + 4.0 evaluation  .

It would be interesting to see if any other computers are reluctant to give up their Queen and are fooled by this position. I will try it on some others.

Very interesting.

Ian.
Report to moderator Logged
Jaco
Guest

E-Mail
Re:Senator module and repetition draws
« Reply #6 on: January 15, 2006, 07:05AM »
Reply with quote

Hi Ian,

Steve Blincoe confirmed to me earlier today that his senator module makes exactly the same mistake - I think we can take it for granted that there is a bug in the senator.

Yes please do test it on the Magellan. Given that the Magellan and the Senator are supposed to run identical programs only on different hardware, it will be interesting to see what happens.

Regards,
Jaco
Report to moderator Logged
drawmeister
Full Member
***

Offline

Posts: 162

How about a nice game of chess ?

View Profile E-Mail
Re:Senator module and repetition draws
« Reply #7 on: January 15, 2006, 05:51AM »
Reply with quote

That is very poor for the Senator not to play Qg5, especially as it is quite a highly rated machine.

I will try the same position out on it's big brother the Magellan......i will be amazed if it misses it too    !!.

This is a nice position to test on other computers.

Ian.
Report to moderator Logged
Jaco
Guest

E-Mail
Re:Senator module and repetition draws
« Reply #8 on: January 14, 2006, 12:19PM »
Reply with quote

Excellent position Ernest.

I tested it on my Mephisto Chess Challenger and on a Mephisto Senator Module (in an Exclusive Board).

The Chess challenger responded to 1. Rg2+ with Qg5 almost immediately.

The Senator on the other hand allowed the repetition draw, irrespective of the amount of thinking time allowed. It evaluated the position as about a 7 pawn advantage for black the entire time.

It would be interesting to see if other senator owners get exactly the same result. If so we might have uncovered a pretty obvious bug in the senator modules.

Regards,
Jaco
Report to moderator Logged
Overtom
Moderator
Co-admin
Hero Member
*****

Offline

Posts: 628



"Life is too short for Chess" (Lord Byron)

View Profile WWW E-Mail
Re:Senator module and repetition draws
« Reply #9 on: January 13, 2006, 04:58AM »
Reply with quote

I don't think it's only Senator that suffers from this ailment. I even wonder if it is a general problem with Mephisto Saitek computers.

Could the senator be made by the same programmer who had a hand in programming the Saitek Mephisto Maestro and the Saitek Mephisto Chess Challenger?

I found that the latter computers also had a fatal tendency to draw where they could easily have won.

I described a few games with this feature in some detail at:

  http://www.xs4all.nl/~tluif/weblog/13FEB05.html

  http://www.xs4all.nl/~tluif/weblog/13NOV04.html

Best regards,

tom
Report to moderator Logged
Robert Weck
Hero Member
*****

Offline

Posts: 791



I love Boardnation!

View Profile
Re:Senator module and repetition draws
« Reply #10 on: January 12, 2006, 11:20PM »
Reply with quote


Quote from: Ernest Gundel on January 12, 2006, 05:55PM   

I don't have any such games saved, but it's pretty easy to find a position where the Senator draws by repetition when it could easily win by making a small sacrifice. For example, black should easily win the following game:

CAPS ARE WHITE, lowercase are black


  {-----------------------
8| .  .  .  b  .  r  k  . |
7| .  R  .  .  .  .  .  . |
6| .  .  .  .  .  p  .  . |
5| q  .  .  .  .  .  .  . |
4| .  .  .  .  .  .  .  . |
3| .  .  .  .  .  .  .  . |
2| .  .  .  .  .  .  .  R |
1| .  .  .  .  .  .  K  . |
  ------------------------}
  A  B  C  D  E  F  G  H 

white to play


Thanks, i will test it!


Robert
Report to moderator Logged
Ernest Gundel
Guest

E-Mail
Re:Senator module and repetition draws
« Reply #11 on: January 12, 2006, 05:55PM »
Reply with quote

I don't have any such games saved, but it's pretty easy to find a position where the Senator draws by repetition when it could easily win by making a small sacrifice. For example, black should easily win the following game:

CAPS ARE WHITE, lowercase are black


  {-----------------------
8| .  .  .  b  .  r  k  . |
7| .  R  .  .  .  .  .  . |
6| .  .  .  .  .  p  .  . |
5| q  .  .  .  .  .  .  . |
4| .  .  .  .  .  .  .  . |
3| .  .  .  .  .  .  .  . |
2| .  .  .  .  .  .  .  R |
1| .  .  .  .  .  .  K  . |
  ------------------------}
  A  B  C  D  E  F  G  H 

white to play
Report to moderator Logged
Robert Weck
Hero Member
*****

Offline

Posts: 791



I love Boardnation!

View Profile
Re:Senator module and repetition draws
« Reply #12 on: January 10, 2006, 03:44AM »
Reply with quote


Quote from: Ernest Gundel on January  7, 2006, 09:54PM   

I noticed exactly the same thing with my Senator. It plays a very good game in general, but doesn't pay any attention to whether it is repeating moves until it has already drawn.

I find it difficult to understand how it could have such an obvious programming error. Even old machines I used 20 years ago, like the MM3 Rebel module, or the Novag Forte, wouldn't allow such bad draws.

Just fixing this one problem should give the Senator a 100+ bump in it's rating.

I have a Milano pro (the stand-alone version of the Senator Module), but have never seen this happening before!

Do you have the notation or the final moves and position of such a game?

Would be very interesting for me, if it is possible to repeat this with a Milano pro...


thank you,
Robert Weck
Report to moderator Logged
Ernest Gundel
Guest

E-Mail
Re:Senator module and repetition draws
« Reply #13 on: January 7, 2006, 09:54PM »
Reply with quote

I noticed exactly the same thing with my Senator. It plays a very good game in general, but doesn't pay any attention to whether it is repeating moves until it has already drawn.

I find it difficult to understand how it could have such an obvious programming error. Even old machines I used 20 years ago, like the MM3 Rebel module, or the Novag Forte, wouldn't allow such bad draws.

Just fixing this one problem should give the Senator a 100+ bump in it's rating.
Report to moderator Logged
Jaco
Guest

E-Mail
Senator module and repetition draws
« Reply #14 on: January 7, 2006, 07:24AM »
Reply with quote

Hi all,

I've been playing my senator module against various other computers to get a feel for the playing style of the engine, and I noticed something funny - the senator would allow a draw by threefold repetition in positions where it is clearly ahead (own evaluation of position at least 2 pawns better). Yet it would allow the other engine the draw.

The first time it happened was against my Mephisto Chess Challenger. I made sure after that to press the ACL button before any comp vs comp games to reset any funny settings. But yesterday the exact same thing happened when I played it against CSTAl 2 on reduced strength (setting on 15).  The position was in the endgame with both sides having a rook, but the senator being up a knight for a pawn. The position was fairly open (senator was definitely in control of the position - so no chance of an incorrect assessment of a drawn endgame position)

Anyone know whether this is a known bug in the senator module?

Regards,
Jaco
Report to moderator Logged
Pages: [1] Reply Notify of replies Send the topic Print 
previous next
Jump to:  

Login with username, password and session length

Chess Computer Community
Get your Ad Free Message Board at Boardnation!   © 2013 Boardnation  Web hosting by Nubis