Idiot 1 - Computer 0 ... Latvian Counter Gambit

Originally posted by SwissGambit
Turning off the opening book is just lobotomizing the thing in a different way. The machine should be able to use any and all information the programmer can give. This includes algorithms, tablebases, and databases.

if you designed the perfect prog it should play the opening with no need for instructions? you think? - I will test this for you I have fritz 5 and I recon I can beat it in the Halloween attack with its book off .

Originally posted by e4chris
if you designed the perfect prog it should play the opening with no need for instructions? you think? - I will test this for you I have fritz 5 and I recon I can beat it in the Halloween attack with its book off .

Everything a computer program does is based on instructions given to it by a programmer.

I could not care less about any results you may get against your lobotomized Fritz 5.

Originally posted by SwissGambit
Everything a computer program does is based on instructions given to it by a programmer.

I could not care less about any results you may get against your lobotomized Fritz 5.

ok thanks script kiddie

you know I used 2b one too and am retired now, you know they use the Halloween attack to test chess progs , guess not ho hum

Actually, I noticed that if you let Rybka think for a while from the first moves of the game, it will play something the Ruy Lopez and the Berlin Wall. Pretty solid openings moves(of course, in blitz it would play something weird like 1. Nc3)

Originally posted by hamworld
Actually, I noticed that if you let Rybka think for a while from the first moves of the game, it will play something the Ruy Lopez and the Berlin Wall. Pretty solid openings moves(of course, in blitz it would play something weird like 1. Nc3)

I think a lot of the e4 openings are intuitive, I worked out that if you programmed a computer with simple rules I could get them to play the 4 nights game and they do....

Here's another game Computer 1 - petrov defence, have won before with the same sac - the computer plays quite elegantly;

I see my miserable name has popped up a few times.
Don't use me as benchmark v a computer it's a fail.

Well e4chris you are clearly enjoying yourself.
And who knows you might just discover the glich of gliches.
The one bizarre opening that defeats all computers.

Millions of people have them but they are looking at what the GM's are playing.
You are pioneering. Digging into the guts of the thing and exposing it's dark secrets.

You never know. Chess, and I believe totally in the game.
May have protected itself against this electronic onslaught and there
is a series of opening moves that just screws it up.
A certain order of moves that no human would play that turns the machine
against itself. Throws it a wobbly.

Get it out of it's book early on and it will play moves no human would consider.

Switch off it's Book and it plays variations which theory considers are tame.
1.e4 e5 2.Nf3 Nf6 3.Nxe4 Qe7. (RJ Does Rybka do this with it's book off.)

So the knack will be get it out it's book very early.

Try playing the 3...Qe7 Latvian reversed. See what happens.

1. e4 e5 2. a3 Nf6 3. f4 Nxe4 4. Qe2


There are glitches in Friz6 (may even up to Fritz9 - no idea never had it)
Fritz6 at one time was being praised as much as Rybka etc is today.
And yet...

Eduard Nemeth - Fritz 6 10 minute game



And I recall seeing this position from years ago. White to play.


Fritz6 (and apparently Fritz9...don't know about FRitz9) analyse this
with White to play and it consider the best move for White is 1. Nxe4.

It (they) actually miss it's a forced mate in two.
Fritz6/Fritz9 owners try it.

Apparently the under promotion and the fact the Rook taking a piece
with check throws the computer into an electronic sulk.
They are after all only machines. Nothing more.



This link will interest you:

http://www.becomeawordgameexpert.com/computers.htm

The above game was taken from there.

Originally posted by greenpawn34
I see my miserable name has popped up a few times.
Don't use me as benchmark v a computer it's a fail.

Well e4chris you are clearly enjoying yourself.
And who knows you might just discover the glich of gliches.
The one bizarre opening that defeats all computers.

Millions of people have them but they are looking at what the GM's are playing.
You /www.becomeawordgameexpert.com/computers.htm

The above game was taken from there.

Thanks for the games, not sure about beating rybka... I think there might be a win in the Latvian... I've seen the Nemeth games, would be good if he posted a few losses, computers do counter attack! like the game above. But there search algorithm is driven by material gain, its always the first moved considered, and forms the search tree. The Saiteks very good at this, why I sac to try and win. In the Halloween attack game I put up it plays Nxg2 - a bad move that wins a pawn but pins the knight. I've noticed both computers have a weakness for pins.

Thanks, I like to play the Latvian...but always... lost 🙁.

But I like your Idea of 3 ... Qe7 !!

Hi GP.

I think most computer programs might fail in the last position. I read somewhere that the developers simply consider it a bad idea to make the program look at underpromotions. The program's main strength is still brute force and it's got to be efficient (meaning its purpose is to maximize its rating). Underpromotions make sense so rarely and introduce so many additional nodes to the game tree that if the program looked at them it would actually play worse chess on average.

And yes, I don't have a life, and I spend my time reading what people say on the internet. I'm halfway through I think.

Edit: Okay, I was wrong. The thing I read is this: http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=1042

They say Rybka actually does consider knight underpromotions. It's bishop underpromotions it doesn't consider. I did seem strange that knight underpromotions wouldn't be considered since they're not that rare actually.

I can't check it anyway.

Originally posted by WanderingKing
Hi GP.

I think most computer programs might fail in the last position. I read somewhere that the developers simply consider it a bad idea to make the program look at underpromotions. The program's main strength is still brute force and it's got to be efficient (meaning its purpose is to maximize its rating). Underpromotions make sense so rarely and ouldn't be considered since they're not that rare actually.

I can't check it anyway.

It makes sense that a program would ignore Bishop promotions because there are not many positions where promotion to B is required to win or draw. (Some people have composed them, although I don't have an example readily available).

Hi SG.

There are examples in the linked thread.

A reason for taking a Bishop would be something like this.

White to play.


1.c8=Q+ Kg5 2.Qxh3 is stalemate.
1.c8=B+ K?? 2.Bxh3 is a simple win.

Programming underpromotions seems inefficient indeed. And not much is lost:
- in tablebase endgames they are considered anyway (up to 6 pieces left)
- in other cases:
* bishop and rook moves are included in the queen. In rare cases, stalemate is avoided by promoting to one of them.
* only the knight attacks squares missed by the queen. So except for the queen, it's the most valuable promotion.

Originally posted by tvochess
Programming underpromotions seems inefficient indeed. And not much is lost:
- in tablebase endgames they are considered anyway (up to 6 pieces left)
- in other cases:
* bishop and rook moves are included in the queen. In rare cases, stalemate is avoided by promoting to one of them.
* only the knight attacks squares missed by the queen. So except for the queen, it's the most valuable promotion.

I've read a few computers have bugs with them, but under promotions are very rare, I doubt it effects there strength much.

Here is the latest win against the Saitek Master. It does not fall for this line given 20 sec per move. But I've been practicing on blitz, having lost 5 games I caught it out in a 10 min game - shows the weakness of pawn / piece grabbing.

I don't want to sound like E Nemeth - the computer mostly destroys me in 10-15 moves but this line it is a real sucker for and is dead in 20 moves. I think I'll be able to win at higher levels.

DEAD BIRD

Last 3 moves are there to explain the position; the computer actually played 21 QxNe3 - Loosing the game, below shows why it has to sac the queen.



I guess that's 'not the bird opening' I looked it up, but I knew the computer would take g5, h6 and probably accept a centre gambit after that, developing all my pieces incl a rook whilst its just opened up its king. On higher levels it initially declines h6.

Any interesting prog games ?

Just to show the computers not stupid here its playing at 30 sec per move (it thought for 1 min+ before playing 3.e4)

Suffice to say I didn't win! gone in 20 moves - its a very sharp position - whites dark squared bishop can't be captured despite being en prise for a few moves.