Reply
 
LinkBack Thread Tools Display Modes
  #1   Report Post  
Old January 26th 11, 01:39 AM posted to rec.games.chess.misc,rec.games.chess.analysis
external usenet poster
 
First recorded activity by ChessBanter: Dec 2009
Posts: 1,146
Default Taylor Kingston Saves Rybka from Fritz, Fixes 'Several Errors'

On Jan 2, 9:38 am, Taylor Kingston
wrote:

Do you or do you not have any examples [plural] where, as you
boldly claimed, you 'corrected' Rybka's 'several' [i.e. beyond
plural] mistakes in this game? Your countless adoring fans await
your answer.


Another was he

6k1/6p1/p7/5pn1/2R5/7P/6PK/8 w - - 0 19

Rybka wanted to play 19.g4. I preferred to observe the general
principle that pawn moves should be made sparingly in the endgame,
especially when it's an unmoved pawn. Keeping reserve tempi, i.e. the
option of a one- or two-square move as required, can make the
difference between wining and losing or drawing.



In point of fact, even older versions of Rybka displayed a certain
reluctance to [recklessly] advance pawns-- except when there was
a tactical justification for doing so. I noticed that many of
Rybka's
tournament games exploited the fact that many other engines were
less inclined to restrain their pawns than was Rybka-x.
Anyway, it's good to see that TK is on top of one basic endgame
strategem.


Thus I played 19.Rf4, which was tied with 8 other moves for 5th-13th
place in Rybka's evaluation, after 19.g4, Kg3, Ra4 and Rc6. This not
only preserved my reserve tempi, it forced Black to use one of his
with 19...g6.



TK appears to have overlooked a simple tactic here. Forget about
the
a-pawn for a moment-- it is doomed in any case. Let's try to focus
on
what really matters he the kingside pawns.

After 19. Rc5 there is a very simple tactical problem-- how to
avoid
losing the f-pawn without hanging the knight.
Now, tries like ...Ne6 and ...Ne4 drop a fundamentally important
pawn
to Rxf5, so that's a wrap-- white will eventually win the a-pawn and
black
will be down the exchange AND down a pawn. Even Phil Innes could
win this. (Well, maybe not Phil Innes.)

So that leaves the try ...g6, saving the f-pawn-- but at what
price? Ah,
there's the rub.
After 19. Rc5 g6 20. g4, the f-pawn cannot capture due to a
lateral
pin (i.e. RxN), so we get: 20. ...Nf3+ 21. Kg3 and now the knight
must
be saved, and this turns the reins back over to white. Or does it?
Not
so fast, Freddie. Black can jump his horse to d2, from which the
beast
threatens a killing fork: ...Ne4+!

But white has a few tactical tricks up his sleeve as well. After
21. Kg3
Nd2, one try is 22. Rc6, a fork involving one of the key pawns
discussed
earlier. Note that up 'till now not even a single kingside pawn has
been
exchanged-- white's winning advantage has not been compromised in
that respect. Now, if black tries to maintain material equality in
terms
of kingside pawns, he hangs the a-pawn AND compromises his pawn
structure by allowing gxf5-- creating a weak isolani as well as
allowing
white an outside passed pawn at h3. Hmm. But if black chooses to
temporarily rescue his outside passer by replying ...a5, he hangs the
pawn on g6 AND the pawn on f5, which is even more disasterous! But
this is an inferior try... for WHITE!
The acid test is what happens after 21. Kg3 Nd2 22.Rc2! Suffice
it
to say that white is not having *any* difficulties here with the issue
of
running out of pawns to promote, leaving him only up the exchange.
To the contrary, the main difficulty for white is deciding which
pawns
to snatch and precisely when to gobble them up-- provided he focuses
on tactics, which here seem to be black's Achilles' heel because the
rook is so effective on an open board.


Unlike the aformentioned 23.Kh4 Ke5 24.Rxe6+? line, I won't say that
Rybka was definitely wrong here. Maybe 19.g4 still wins.
But it seemed so contrary to the endgame principles I've learned that I
chose not to play it.



This is *not* a quiet position. There are lots of tactics here.
And
by ignoring this fact, TK risked mucking up what would have been a
routine 'mop up' for Rybka-4. Lucky for him, his opponent was a
relative patzer-- a man who only 'talks' a good game.


I have now presented three instances where I preferred human
judgement to Rybka's in this analysis.



Radical rephrasing noted. Previously, we were told that it had
been
'necessary' to correct Rybka's 'several' errors in order to salvage
the win.
Such blatant lies may pass muster in certain circles hereabout, but
then,
even an inveterate liar can't fool... 'all of the people, all of the
time.'
  #2   Report Post  
Old January 26th 11, 01:56 PM posted to rec.games.chess.misc,rec.games.chess.analysis
external usenet poster
 
First recorded activity by ChessBanter: Jun 2009
Posts: 3,256
Default Taylor Kingston Saves Rybka from Fritz, Fixes 'Several Errors'

On Jan 25, 8:39*pm, The Master wrote:
On Jan 2, 9:38 am, Taylor Kingston
wrote:


6k1/6p1/p7/5pn1/2R5/7P/6PK/8 w - - 0 19


* Rybka wanted to play 19.g4. I preferred to observe the general
principle that pawn moves should be made sparingly in the endgame,
especially when it's an unmoved pawn. Keeping reserve tempi, i.e. the
option of a one- or two-square move as required, can make the
difference between wining and losing or drawing.


* Thus I played 19.Rf4, which was tied with 8 other moves for 5th-13th
place in Rybka's evaluation, after 19.g4, Kg3, Ra4 and Rc6. This not
only preserved my reserve tempi, it forced Black to use one of his
with 19...g6.


* TK appears to have overlooked a simple tactic here. Forget about
the
a-pawn for a moment-- it is doomed in any case. * Let's try to focus
on
what really matters he the kingside pawns.

* *After 19. Rc5 there is a very simple tactical problem


Yes, Greg. That has already been discussed. If you'll back up a few
posts you'll see that I have already noted how 19.Rc5 would have won
faster. I also explained why it was not played.
My version of Rybka has no built-in user interface; it must be run
as an alternate engine under other software. When playing Lopez, I ran
Rybka under Fritz. In analyzing later, I ran Rybka under ChessBase.
Turns out Rybka runs much better under CB. Even Fritz runs better
under CB than under Fritz!
Rybka under Fritz did not rate 19.Rc5 very highly, but under CB it
quickly had 19.Rc5 as #1.
  #3   Report Post  
Old January 28th 11, 08:49 AM posted to rec.games.chess.misc,rec.games.chess.analysis
external usenet poster
 
First recorded activity by ChessBanter: Dec 2009
Posts: 1,146
Default Taylor Kingston Saves Rybka from Fritz, Fixes 'Several Errors'

On Jan 26, 8:56*am, Taylor Kingston
wrote:

* My version of Rybka has no built-in user interface; it must be run
as an alternate engine under other software.



Apart from the most recent versions of Rybka, a few of which ship
with a user interface specifically designed for Rybka-4, *all*
earlier
versions of the program were chess engines designed to run under
a separate GUI. Put another way, every time someone posted here
in rgc about 'Rybka,' they were running the engine under one GUI or
another. This is nothing new or special, and *you* do have control
of the settings, provided you can locate them.


When playing Lopez, I ran Rybka under Fritz.



My old version of 'Fritz' was a ChessBase product, combining the
'Fritz' engine with their most popular GUI, sold together as a
package.


In analyzing later, I ran Rybka under ChessBase.
Turns out Rybka runs much better under CB. Even Fritz runs better
under CB than under Fritz!



Perhaps the settings are different. In my current setup, I have
some
-- but certainly not all -- of the endgame tablebases, but rather
than
risk slowing the search down significantly by selecting a frequent
look
for a match in the endgame, I instructed the GUI/engine to only check
for such a match occasionally. This is because the vast majority of
my work involves openings and middlegame positions, and because I
don't have 'Shredder' with its much more compact tablebases. My
computer's hard drive lacks the capacity to effectively utilize 7-man
tablebases or even 6-man without say, first compacting them to the
size of Phil Innes' brain-- in which sorry state they could not easily
be
read or understood.


* Rybka under Fritz did not rate 19.Rc5 very highly, but under CB it
quickly had 19.Rc5 as #1.



On my machine, with my current setup, even Rybka-4 commits the
Reinfeldian blunder of willingly saddling itself with a rookpawn after
the
inferior move g4. Of course it would not actually play that move
since
it prefers Rc5! due to the fact that it wins material rather
quickly. In
fact, following its 'best play' forward just a couple of plies, Rybka
soon
spots a forced mate for white in the line I discussed earlier, ending
with
Rc2.

In competition with other chess engines, Rybka would be running on
vastly superior hardware with a killer openings book (which I do not
have) and I expect, armed to the teeth with at least the six-man
tablebases (to cover up such glaring weakness{es}) in the endgame,
in case any of its competitors should be so fortunate as to survive
that
long. Even so, I am disappointed that the programmers would 'leave
in' such grotesque a weakness as this one, even if it is always
successfully covered up in competitive play by the use of tablebases,
the channeling of Smyslov's ghost, or some other secondary means.

All that aside, I have yet to see an example of Rybka-4 goofing up
in the middlegame-- which is what I imagined TTK had in mind when
venturing that it was 'necessary' for him to intervene 'several' times
in
order to rescue Rybka from Fritz-Innes-Lopez. It seems to me that
there is but one weakness here which requires any intervention-- the
well-known rookpawn problem. Every sane mathematician I know
comes up with a total of one 'necessary' intervention, although the
minority report (compiled by a Dr. Innes) lists this as equivalent to
TTK's 'several' talley, arguing that one is equal to many in the
sense
that all things are equal-- 'sum' are just more equal than others....



  #4   Report Post  
Old January 28th 11, 08:26 PM posted to rec.games.chess.misc,rec.games.chess.analysis
external usenet poster
 
First recorded activity by ChessBanter: Oct 2007
Posts: 3,073
Default Taylor Kingston Saves Rybka from Fritz, Fixes 'Several Errors'

On Jan 28, 3:49*am, The Master wrote:

for a match in the endgame, I instructed the GUI/engine to only check
for such a match occasionally. * This is because the vast majority of
my work involves openings and middlegame positions, and because I
don't have 'Shredder' with its much more compact tablebases. * My
computer's hard drive lacks the capacity to effectively utilize 7-man
tablebases or even 6-man without say, first compacting them to the
size of Phil Innes' brain-- in which sorry state they could not easily
be read or understood.


R
O
T
F
L
M
A
O




Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is Off
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
GetClub Chess made Twice Better. Sanny rec.games.chess.analysis (Chess Analysis) 25 June 19th 08 10:05 AM
Fritz 10 problem, running out of time and weak Fritz 10 KPR rec.games.chess.computer (Computer Chess) 3 October 11th 07 10:14 PM
Fritz 10 vs Rybka 2.2 Paul rec.games.chess.computer (Computer Chess) 18 February 8th 07 06:45 PM
Fritz 10 vs Rybka 2.2 Rob Mitchell rec.games.chess.misc (Chess General) 2 December 21st 06 10:35 PM
Strongest program C.C.LONGTHORP rec.games.chess.computer (Computer Chess) 37 December 19th 05 07:21 PM


All times are GMT +1. The time now is 05:52 AM.

Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
Copyright 2004-2019 ChessBanter.
The comments are property of their posters.
 

About Us

"It's about Chess"

 

Copyright © 2017