Author: Albert Silver
Date: 07:40:41 09/12/04
Go up one level in this thread
On September 12, 2004 at 05:46:21, Ed Schröder wrote:
>On September 11, 2004 at 23:12:46, Albert Silver wrote:
>
>>One of the common seen comments on Pro Deo is that it is a poor blitz player.
>>I've seen this comment in several Computer Chess forums. However, one can't help
>>but notice how unusually quickly it plays compared to the time it has alloted,
>>no matter what the time control. At the Pro Deo parameter page
>>(http://members.home.nl/matador/engine.htm) there is a switch that allows one to
>>have Pro Deo play faster or slower on average:
>>
>>[Time Control ? 125] * 125
>>[Time Control Status ? PB ONLY] * ALWAYS|PB ONLY
>>
>>(to activate, one must replace the '?' with '=')
>>
>>I ran 2 blitz matches using the Nunn 2 set with Pro Deo playing against Ruffian
>>1.05 at 5 mins + 3 sec increment, in one set using the default settings
>>(September 1st) and in the other using the time control parameter. The only
>>other change I made was to turn off all learning (position, engine, book) so as
>>to be sure the conditions were identical. It's not conclusive but it is very
>>interesting, as 80 games were played in all.
>>
>>Conditions: Athlon 2400XP+, 32 MB hash, Arena 1.08 GUI, pondering off.
>>
>>Without Time Control setting: Pro Deo lost 16 - 24 to Ruffian 1.05
>>
>>With the Time Control setting (at 125): Pro Deo *won* 20.5 - 19.5
>>
>> Albert
>
>
>Let me clarify.
>
>[Time Control = 125] * 125
>[Time Control Status = PB ONLY] * ALWAYS|PB ONLY
>
>First of all "PB ONLY" stands for "Permanent Brain Only". Meaning that the 25%
>increase of time management is only active when using the Permanent Brain (PB).
>If you want the option to be active in (say) regular eng-eng matches where the
>PB is off by default then set the switch to ALWAYS.
>
>I don't know how you have tested but if the 2 matches were without PB you have
>played with indentical settings.
>
>The switch is there because I think that increasing the time control works when
>the PB is active and doesn't work very well when it is not.
>
>As you see testing is hell.
I *know* that testing is hell... Though thankfully it wasn't because I had
screwed up the conditions. :-)
I made this fancy personality and tested it against Ruffian and Junior. I liked
it because it was very aggressive and very dynamic. It scored identically to
them compared to the default settings. Identical. And no, I didn't make a
mistake when loading them. Different games, different wins and losses.
I was disappointed and after doing some more testing noticed that all the
switches I had made such as increasing the weak pawn values, stronger square
values, and above all the Pawn Pressure (at 160-170 for the middlegame it plays
a very interesting game), my (or Pro Deo's rather) tactics went to hell. It was
missing shots and when testing seemed to take longer on most tactics compared to
the default set. I don't count a right move right if it gets it right for some
microscopic positional reason. It has to see it was right. So while it might
have been playing better positionally (presumably), it was failing in the
tactics department. Sigh!
Albert
>
>My best,
>
>Ed
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.