Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A. STEEN vs. FRUIT 2.2.1 {Posted at request of Graham Banks & M. Mon

Author: A. Steen

Date: 11:25:57 11/24/05

Go up one level in this thread


On November 24, 2005 at 11:31:29, stuart taylor wrote:

>So are you saying that the quality of Fruits play is lower than other programs
>today?


No.  It has in a few weeks become my preferred engine for analysis.

It is reliable and has few peculiarities.

But it is _peaceful_ compared to S9/F9, in my opinion.

This is hard to explain, because I do not understand it myself.  I can only
report that to roll out victories by me over Fruit is not difficult but to roll
out victories against F9/S9 is difficult.  They introduce Tal-esque
complications.  Maybe they only pretend to think those are good. :)  But when
they throw me five bad problems in a game, unless I solve 4 or 5 they win.

Fruit wins by superior strategic overview.  It does not falsely report winnings
as is the trademark of Fritz.   When I beat Fruit, post-mortem reveals it
generally knew it was going to lose.  This does not apply to the others,
especially Fritz.

That may prove some defect or quality in me, and nothing about Fruit. But I have
read others say much the same here before.



>I too thank you for all your input in this forum. I always felt (I'm approx 2200 elo) that


Thank you.

A moderator has (correctly IMO) today removed another analysis thread, because
the other party just could not stop simultaneously insulting me and making
analysis blunders.
If he made one but not the other, I would not expose his failings.
But when he does both, and persists with it, it becomes like a sacred duty to
expose. :)

Are you aware there seems a posse out there regularly emailing moderators asking
for me to be banned?  Most members seem to be people who I falsified (using the
word in its scientific sense). :)

I make it an absolute rule never to reply to off-board email, whatever the
source.  I will probably read it, but it will never be responded to.  That keeps
things above-board.


>computerchess never yet got properly off the ground, even though it
>was very succesful in several demonstrations against world champions.
>Really, if a World Champ would play in  the right way, the way he ought to be
>able to, I suppose he might win almost every game, at long time controls. But
>humans are humans, and don't do what ought to be able to be done.


I believe I can pass 80% against the current Fruit, even when it is properly
booked.  That will now have to wait.

As to Fritz9 and Shredder9, I need to put in greater effort.

My technique against all programs is the same - I must stay in control, and etc.
etc. you know the rest, no doubt it has been discussed 100 times here.  The
methods of "Father" have no interest for me as my aim is to improve my own chess
to the point when my brain is full or chess is empty.  As my brain is evidently
nearly empty, that leaves much room for progress, and chess is far from empty.
:)



>So do you predict that there will be a program in 2006 which will REALLY be
>interesting, and not merely a technically programmed number cruncher (i.e. made
>to overcome some programming faults of contemporary rival programs, and
>obviously, an almost certain winner against humans, like they ever were for
> many years already)?


Good question.  We already know that programs A, B, C can easily demonstrate the
property

 A > B > C > A

to a provable degree that will not happen with top humans, even allowing for the
greater variability of all things human.

The more there is automated tuning, the more the non-transitivity problem will
escalate.

At short time controls, I reported Fruit _almost_ having more problems with
Fritz 5.32 than with Fritz 9 - always a reason for keeping those old CDs and
engines!


>S.Taylor

Best,

A.S.



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.