Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Is knowing more good or bad?
- - By Kapaun (****) Date 2009-10-15 12:28
Well - knowing more is generally good, that's no question. But in chess, could it be that there is a point, where knowing more becomes bad? Let me give an example. Let's assume you are looking 7 plies ahead. And there is a variation which seems promising, so you are playing it. Now let's assume you are looking 10 plies ahead and find that this particular variation isn't quite as good any more. So you play another one which now seems the best - although it's little more than mediocre. Would looking more plies ahead finally kill all promising variations? If chess really is a draw this could be true. Or not? Vas says that each plie ahead increases ELO - but the amount of added ELO is sinking from plie to plie. Maybe it will reach zero at some point? Or even a negative value? Is chess a game where the person wins who can best ride over the barely frozen lake?
Parent - By Felix Kling (Gold) Date 2009-10-15 12:36
the theories that you become worse when being "better" are only theoretical constructions. Reality is much easier :)
Parent - - By Highendman (****) Date 2009-10-15 12:46
No. Knowing more is better in any domain in life, imho :)

You also asked if looking ahead further in chess is better. The answer for that is 100% yes.

Basically - you want the best possible evaluation function of a position (equals knowing more) and to look as far ahead at as many relevant positions (with emphasise on relevant - knowing what to prune out from the search and what not to) as possible.
Parent - By George Tsavdaris (****) Date 2009-10-15 13:07

>You also asked if looking ahead further in chess is better. The answer for that is 100% yes.


I'm not sure at all the answer is yes.
I guess looking further more plies can never be bad(generally), but perhaps there maybe a point(diminishing returns) where looking further gives no gain at all, so better is not happening.
Parent - - By Kapaun (****) Date 2009-10-15 13:07
Asking again: Would looking more plies ahead finally kill all promising variations? And if no - why not?
Parent - By SR (****) Date 2009-10-15 21:58
I think what you suggest have been demonstrated in checkers as well as othello. In an article  available at
web.cs.ualberta.ca/~jonathan/Grad/Papers/dim.ps
it is argued that the diminishing return is also applies to chess. In other words additional plies of search is translated
into decreasing benefits. Here is a quote from the paper:

"The results in other games raise a troubling question: Why doesn't chess show diminishing
returns? This paper attempts to address this problem."
Parent - By Bouddha (****) Date 2009-10-15 13:13

> Knowing more is better in any domain in life, imho <IMG class="sic sic_sml_pos" alt=:-) src="/mwf/epx.png">
>


I am really not sure about that.
Sometimes its better not to know !

regards  ;-)
Parent - - By Bouddha (****) Date 2009-10-15 13:17
Chess history gives good examples.

Lasker was known to play moves which he knew was not the best but playing psychological move to maximize results.
So knowing more is maybe not worse if you can then set your mind to play moves that gives you the most chances to maximize the result OTB knowing that no one is perfect.

I am not an adept of this playing style.
Mine is more like I play the move which is think is the best one.
Parent - - By Lukas Cimiotti (Bronze) Date 2009-10-15 13:20

>Mine is more like I play the move which is think is the best one.


Anything else could seriously backfire - especially when you play the cluster ;)
Parent - By Bouddha (****) Date 2009-10-15 13:40
:-)

Well there is no psychology playing the cluster.
You just have to try and play better than it which is coming more and more difficult.

I do still think that its better for humans to have closed positions.
French & KID is the best for black to have a chance and win against the cluster.

With white, there are many options.

But I currently failed to demonstrate how to win against the cluster. I was close a few times but its all.

Lukas, I wish you and Rybka team all the best for the coming tournament. I will follow the games closely.
Do you know about a place/way we can follow the games ?

regards
Parent - By onursurme (***) Date 2009-10-15 21:14
While knowing more about the position and looking more plies ahaed, you must also know more about your opponent.
If you know enough about your opponent, you can also guess its replies for all possible moves, and then you can choose a move which you think your
opponent will reply with a wrong move.
for example, in this multi-pv demonstration :
1. Nd5 0.0
2. Qe8 0.0
3. c3 0.0
.....
Imagine that only 3 moves will draw the game for you, and all other moves will make you lose the game.
If you know that your opponent will reply with right moves to your 1st and 3rd move, and with a great possibility it will reply with a losing move to your 2nd move,
you will chose the 2nd move, even though in theory it will cause a draw.

Searching your second move 25 plies ahaed shows that you lose, but searching it 40 moves ahaed shows that it is a draw.
You can search 40 plies but your opponent can search 25 plies. Then you can know that when you make the 2nd move, your opponent will
misevaluate the position and make a wrong move. For your 1st and 3rd moves, searching for 25 plies is enough to find the right reply,
so you don't want to make your 1st or 3rd move because you know that your opponent will be able to find the right answers with its ability
to search 25 plies ahaed in this position.
Up Topic Rybka Support & Discussion / Rybka Discussion / Is knowing more good or bad?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill