It gives away the queen for free, or refuses to make a "mate in 1", etc.
It's not playing chess. Nothing in the analysis can be trusted!

This is a late middlegame/early endgame involving queens and pawns, plus one
piece. There is a lot of tablebase access. Try it for yourself, if you wish.
Here's the fen. I have analyzed it to a level of 24/58 taking 12+ hours, and
the result is still nonsense!

8/1B3pk1/2Q1p3/2q1p1pp/8/6P1/5PKP/8 b - - 0 1

Analysis by Shredder704.eng (3/4/2003): [344kN/s]

47...Qd4 48.Qf3 Kg6 49.Bc6 g4 50.Qb3 f5 51.Qxe6+ Kg5 52.h3 Qa1 53.hxg4 hxg4
54.Be8 e4 55.Qe7+ Kh6 56.Kh2 Qb2
+- (1.74) Depth: 11/28 00:00:08 3427kN, tb=41
+- (1.77) Depth: 24/58 12:15:52 15245243kN, tb=13125095

Shredder 7 GUI, Athlon XP 2400+ 2GHz, 1GB RAM, hash 512MB, tablebase cache
64MB, all 3-4-5 piece tablebases

--
Euc1id

Euc1id wrote:

It gives away the queen for free, or refuses to make a "mate in 1", etc.
It's not playing chess. Nothing in the analysis can be trusted!

This is a late middlegame/early endgame involving queens and pawns, plus one
piece. There is a lot of tablebase access. Try it for yourself, if you wish.
Here's the fen. I have analyzed it to a level of 24/58 taking 12+ hours, and
the result is still nonsense!

8/1B3pk1/2Q1p3/2q1p1pp/8/6P1/5PKP/8 b - - 0 1

Analysis by Shredder704.eng (3/4/2003): [344kN/s]

47...Qd4 48.Qf3 Kg6 49.Bc6 g4 50.Qb3 f5 51.Qxe6+ Kg5 52.h3 Qa1 53.hxg4 hxg4
54.Be8 e4 55.Qe7+ Kh6 56.Kh2 Qb2
+- (1.74) Depth: 11/28 00:00:08 3427kN, tb=41
+- (1.77) Depth: 24/58 12:15:52 15245243kN, tb=13125095

Shredder 7 GUI, Athlon XP 2400+ 2GHz, 1GB RAM, hash 512MB, tablebase cache
64MB, all 3-4-5 piece tablebases

--
Euc1id

Well, I can't see a mate in 1 either. The given analysis seems to be the
best move order for black.

--
Gero H. Marten

Computers are like air conditioners.
As soon you open windows, they stop working.
- Remove the "x" for e-mail -

"Euc1id" wrote in message
news
It gives away the queen for free, or refuses to make a "mate in 1", etc.
It's not playing chess. Nothing in the analysis can be trusted!

This is a late middlegame/early endgame involving queens and pawns, plus
one
piece. There is a lot of tablebase access. Try it for yourself, if you
wish.
Here's the fen. I have analyzed it to a level of 24/58 taking 12+ hours,
and
the result is still nonsense!

8/1B3pk1/2Q1p3/2q1p1pp/8/6P1/5PKP/8 b - - 0 1

Analysis by Shredder704.eng (3/4/2003): [344kN/s]

47...Qd4 48.Qf3 Kg6 49.Bc6 g4 50.Qb3 f5 51.Qxe6+ Kg5 52.h3 Qa1 53.hxg4
hxg4
54.Be8 e4 55.Qe7+ Kh6 56.Kh2 Qb2
+- (1.74) Depth: 11/28 00:00:08 3427kN, tb=41
+- (1.77) Depth: 24/58 12:15:52 15245243kN, tb=13125095

I don't get your problem - Qd4 is an ok move to make here.

--
GCP

Perhaps 47...Qd4 is an OK move, but the analysis given in the engine pane is
usually absurd. Try playing over those lines on a regular chessboard. I only
gave the last one that appeared at level 24/58. If I recall correctly, the
prior one at level 23/xx gave away the queen for free. The last one at 24/58
is also ridiculous, as you'll see if you play it over on a chessboard. Black
starts playing give-away-the-pawns chess with 50...f5
--
Euc1id

"Gian-Carlo Pascutto" wrote in message
...
| "Euc1id" wrote in message
| news | It gives away the queen for free, or refuses to make a "mate in 1", etc.
| It's not playing chess. Nothing in the analysis can be trusted!
|
| This is a late middlegame/early endgame involving queens and pawns, plus
| one
| piece. There is a lot of tablebase access. Try it for yourself, if you
| wish.
| Here's the fen. I have analyzed it to a level of 24/58 taking 12+ hours,
| and
| the result is still nonsense!
|
| 8/1B3pk1/2Q1p3/2q1p1pp/8/6P1/5PKP/8 b - - 0 1
|
| Analysis by Shredder704.eng (3/4/2003): [344kN/s]
|
| 47...Qd4 48.Qf3 Kg6 49.Bc6 g4 50.Qb3 f5 51.Qxe6+ Kg5 52.h3 Qa1 53.hxg4
| hxg4
| 54.Be8 e4 55.Qe7+ Kh6 56.Kh2 Qb2
| +- (1.74) Depth: 11/28 00:00:08 3427kN, tb=41
| +- (1.77) Depth: 24/58 12:15:52 15245243kN, tb=13125095
|
| I don't get your problem - Qd4 is an ok move to make here.
|
| --
| GCP
|
|

You didn't look very hard. Set up that position and generate the lines of
analysis. The free queen give-away happened at level 23/xx if I recall
correctly, perhaps before. The last line at 24/58 is also absurd. Play it
over on a regular chessboard. If you're at least an average chess player,
you'll see that it's nonsense.
--
Euc1id

"NN" wrote in message
...
| Euc1id wrote:
|
| It gives away the queen for free, or refuses to make a "mate in 1", etc.
| It's not playing chess. Nothing in the analysis can be trusted!
|
| This is a late middlegame/early endgame involving queens and pawns, plus
one
| piece. There is a lot of tablebase access. Try it for yourself, if you
wish.
| Here's the fen. I have analyzed it to a level of 24/58 taking 12+ hours,
and
| the result is still nonsense!
|
| 8/1B3pk1/2Q1p3/2q1p1pp/8/6P1/5PKP/8 b - - 0 1
|
| Analysis by Shredder704.eng (3/4/2003): [344kN/s]
|
| 47...Qd4 48.Qf3 Kg6 49.Bc6 g4 50.Qb3 f5 51.Qxe6+ Kg5 52.h3 Qa1 53.hxg4
hxg4
| 54.Be8 e4 55.Qe7+ Kh6 56.Kh2 Qb2
| +- (1.74) Depth: 11/28 00:00:08 3427kN, tb=41
| +- (1.77) Depth: 24/58 12:15:52 15245243kN, tb=13125095
|
| Shredder 7 GUI, Athlon XP 2400+ 2GHz, 1GB RAM, hash 512MB, tablebase
cache
| 64MB, all 3-4-5 piece tablebases
|
| --
| Euc1id
|
| Well, I can't see a mate in 1 either. The given analysis seems to be the
| best move order for black.
|
| --
| Gero H. Marten
|
| Computers are like air conditioners.
| As soon you open windows, they stop working.
| - Remove the "x" for e-mail -

"Euc1id" wrote in message
ink.net...
Perhaps 47...Qd4 is an OK move, but the analysis given in the engine pane
is
usually absurd. Try playing over those lines on a regular chessboard. I
only
gave the last one that appeared at level 24/58. If I recall correctly, the
prior one at level 23/xx gave away the queen for free. The last one at
24/58
is also ridiculous, as you'll see if you play it over on a chessboard.
Black
starts playing give-away-the-pawns chess with 50...f5

If you're referring to the moves and the end of the varation,
this is quite normal. The analysis becomes less deep as you
go deeper into the variation. At the very end of the line,
the moves become very unreliable and sometimes nonscenscial.
Every chessprogram has this, not just Shredder.

"Don't trust anything but the first move"

If you want to see further analysis, play along the suggested
moves and let the program analyze, dont just play out the
mainvariation at once.

--
GCP

I disagree. The Fritz engines had the same kind of problem, but they
eventually fixed it. It is indefensible for such nonsense to appear in the
engines pane as legitimate analysis. It must be a bug in the engine and/or
GUI. There's no other explanation, and it is not acceptable. Even at the
very end of a long analysis line, it should not be nonsensical such as
giving away the queen, failing to "mate in 1", etc. That only means there is
a bug, nothing more.

owever these problems with Shredder704.eng are NOT appearing at the very end
of a long analysis line. In some cases they appear only a few ply deep, near
the beginning of the line, perhaps only 2-3 ply deep. Again, this must be a
bug...IMHO. And that casts a great deal of doubt about the reliability of
Shredder 7, both the GUI and the engines.
--
Euc1id

"Gian-Carlo Pascutto" wrote in message
...
| "Euc1id" wrote in message
| ink.net...
| Perhaps 47...Qd4 is an OK move, but the analysis given in the engine
pane
| is
| usually absurd. Try playing over those lines on a regular chessboard. I
| only
| gave the last one that appeared at level 24/58. If I recall correctly,
the
| prior one at level 23/xx gave away the queen for free. The last one at
| 24/58
| is also ridiculous, as you'll see if you play it over on a chessboard.
| Black
| starts playing give-away-the-pawns chess with 50...f5
|
| If you're referring to the moves and the end of the varation,
| this is quite normal. The analysis becomes less deep as you
| go deeper into the variation. At the very end of the line,
| the moves become very unreliable and sometimes nonscenscial.
| Every chessprogram has this, not just Shredder.
|
| "Don't trust anything but the first move"
|
| If you want to see further analysis, play along the suggested
| moves and let the program analyze, dont just play out the
| mainvariation at once.
|
| --
| GCP
|
|

And that casts a great deal of doubt about the reliability of
Shredder 7, both the GUI and the engines.

My point still stands.

And you might wonder why some other engines have the same 'bug'.

And you might wonder why despite having this 'bug' Shredder 7.04
is leading the SSDF.

--
GCP

"Gian-Carlo Pascutto" wrote in message
...
| And that casts a great deal of doubt about the reliability of
| Shredder 7, both the GUI and the engines.
|
| My point still stands.

No, it does not. It is nonsense, like the analysis of Shredder704.eng.

| And you might wonder why some other engines have the same 'bug'.

That's easy, because they also have a bug.

| And you might wonder why despite having this 'bug' Shredder 7.04
| is leading the SSDF.

Because of one of the following, probably:
(1) The version of Shredder 7/704 released on the CD and via the March 2003
update is not the same as used by SSDF for testing. Or...
(2) Or SSDF is a bought & paid for subsidiary of Chessbase which pays them a
lot of money under the table to post their latest software release at the
top of the list. Or...
(3) Or, both of the above.

I know they have paid-for shills frequenting this NG and the popular chess
forums. So everything and the source must be evaluated independently,
trusting nobody and no source, until it is scientifically proven as fact.
--
Euc1id

They must cut off the analysis at the point where what remains is
reliable.
Otherwise it is a bug.

'Reliable' is not a true or false condition.

The moves become less reliable as you go deeper.
When do you decide to cut off?

The primary function of a chess engine is not to play chess, contrary to
popular belief in certain circles, but to produce reliable analysis --
without which it is useless, as far as I'm concerned! If you want to use
it
to play chess, fine, but I don't. I only use chess software for analysis.
That is my only application for it -- and for a lot of others too, I
imagine. When it falls short in that area, then it becomes worthless to
me.

As I already said, trust the first move, not any of the others,
and play along the moves instead of skipping to the end of the
variation at once. This advice applies to _all_ chess engines.

--
GCP