Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Bringer Evaluation - It Really Is Strong!

Author: Graham Laight

Date: 15:53:23 01/09/00

Go up one level in this thread


On January 09, 2000 at 17:02:53, Gerrit Reubold wrote:

>Hi Graham,
>
>
>On January 08, 2000 at 11:32:53, Graham Laight wrote:
>
>>I have tried Bringer - and I like it! Thanks for telling me about it.
>>
>>It took less than 3 minutes to download, and, after unzipping it, I could fire
>>it up and play it!
>>
>>I played it myself and was comprehensively beaten.
>>
>>The following games were all played on an HP Onmibook 4100, 266 Mhz, NT4, and I
>>have not downloaded Bringer's optional "extra large" opening book, because I am
>>personally not very interested in openings (which is often a grave weakness in
>>my game!).
>>
>>Against Kasp2100, it went into the late middlegame a pawn ahead, and 2100
>>cleverly created a passed pawn in the centre. Before it was able to do anything,
>>though, Bringer swapped off the pieces, then created it's own passed pawn on the
>>a-file - but this passed pawn was out of reach of 2100's king. I resigned on
>>behalf of 2100.
>>
>>I then tried it against 2 DOS programs on my 200 Mhz PC (this is probably fair,
>>because I assume that NT steals a lot of processor time):
>>
>>Against CS-Tal '96, Bringer kept on raising it's evaluation move by move, then
>>dropping back down to almost equal again! The game was very exciting, and the
>>suspense was almost unbearable. Suddenly, Bringer's evaluation went high, and
>>eventually it demonstrated why it was so confident by comprehensively winning.
>>
>>Finally, I tried it against Rebel Decade 2, assuming that this strong program
>>would beat it easily. Wrong again! Bringer's evaluation rose high long before
>>Decade's moved away from close to equal. Bringer won the game brilliantly. Even
>>more astonishingly - Decade 2 consistently used much more time to select its
>>moves.
>>
>>Against all 3 strong programs, it always seemed to evaluate the position better
>>than they did, and saw the evaluation changes sooner than they did. It seemed to
>>have a better understanding of the positions.
>
>Good results for the Bringer, however, don't overestimate its strength after a
>few games. Playing it against the top engines (commercial and freeware) will
>give you different results.

OK. I have not tried to give an actual estimate of strength. But I do know this:
of the programs in my collection, it is the first one where I truly have no idea
why it's choosing the moves it is. When I got comprehensively beaten by it, I
assumed I had just played badly. After playing the other games, and watching how
(from the eval) it clearly saw things MUCH earlier than Decade 2.0 (which
absolutely crushes me), I had to say it felt strong.

Playing it reminded me of the only time I have played a human IM. I kept
thinking, "this is OK - I'm doing very well against such a strong player". Then,
all of a sudden, my position just totally collapsed, I was check mated, and I
didn't know why.

Watching Bringer play reminds me of this experience - I just don't understand
why it chooses the moves it does - victory just "emerges".

But, as a (self estimated) 1700 player, I cannot truly judge it's strength. But
it sure as hell seems strong to me.

>>
>>There are a few changes I would personally like to see in this program:
>>
>>* Should recognise draw by repetition. When tinkering with it myself, we moved
>>back and forth through the same position several times, and it did not flag the
>>draw
>
>Someone else (Peter Berger) mentioned this some days ago, unfortunately, I am
>unable to reproduce it. Bringer claims a draw, both when playing it in its own
>GUI and as a winboard engine. How do I reproduce the bug? BTW, was Bringers
>score 0.00 while repeating the moves?

Sorry - I didn't take note of that - I'll try again tomorrow.

What I did was play until I was losing, then swapped sides, and tried to win.
Then we got to a position where both players were moving a piece back and forth
between 2 squares. After 3 reps, something should have happened - but it kept on
repeating, at 5 seconds per move.

btw - I'm using the "current" version - not the new "beta" version. I'm using
the Bringer GUI (which is good, btw).

>>
>>* An "autoplay" option would be nice
>
>Do you mean a possibility for Bringer playing a position agains Bringer? I had

Yes - the program going ahead and automatically making the moves for both sides.

>this once, broke it a year ago and didn't care about repairing it: the
>"Computer-Computer" button is disabled. I don't think this feature is important
>with the possibility to interface Bringer to other engines per Autoplayer or per
>Winboard. Is it important?

It may not be important to everyone, but I personally use it quite a lot.
Sometimes, I feel that the only way to get closer to the truth in a position is
to go away and let the computer "play it out" until the situation becomes more
clear. So, offering it as a menu option would make this "customer" very happy!

>>
>>* I would like to see the squares flash when the computer moves. I don't like
>>having to look at the move list to see what the opponent has done. I wish more
>>programs would flash the squares being moved to/from like Hiarcs can
>
>I thought about small "recent-squares-indicators", instead of flashing the
>squares. Maybe a thin frame or a dot. If I have time, I will implement this for
>the next version.

Excellent! It doesn't have to be flashing squares. CM2100 (yes - I admit to
being a happy owner of it!) animates its piece moves. Anything like this would
be good. Bringer moves the piece instantaneously - unless you're focused on the
right part of the board, you'd miss it.

>>
>>Overall though, I wholeheartedly approve of this nice program!
>>
>>-g
>
>
>Greetings,
>Gerrit

Good luck in the ongoing blitz tourny currently in progress!

Thanks for making the effort to make the process of getting and running this
nice program so easy.

-g



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.