Numerical Valuations

Discuss anything you like about chess related matters in this forum.
User avatar
MJMcCready
Posts: 3191
Joined: Mon Jun 24, 2013 2:30 pm

Numerical Valuations

Post by MJMcCready » Sun Jan 10, 2021 12:01 am

When did we start assigning numerical values to pieces and pawns? I ask because I struggle to understand how or why a pawn is given a value of 1. How is this achieved when pawns have different positions on the board which in turn alters their evaluation? The a&h pawns obviously have fewer options than all other pawns yet are attributed the same value? Who came up with that idea? Similarly the closer to the centre a pawn is, generally the greater the value it is given. Is the value of one supposed to reflect the mean or the median? It doesn't seem refined or that it's been well thought through at all, moreover that pawns are given the value of one, simply for the sake of convenience. So when did we start doing this? Who came up with the idea in the first place?

Roger de Coverly
Posts: 21312
Joined: Tue Apr 15, 2008 2:51 pm

Re: Numerical Valuations

Post by Roger de Coverly » Sun Jan 10, 2021 12:48 am

MJMcCready wrote:
Sun Jan 10, 2021 12:01 am
Who came up with the idea in the first place?
It's likely lost in the mists of time back to the earliest writers on chess. Think of it as a crude first approximation to an evaluation function as used by chess engines and as taught to beginners as an initial guide to relative piece values.

User avatar
MJMcCready
Posts: 3191
Joined: Mon Jun 24, 2013 2:30 pm

Re: Numerical Valuations

Post by MJMcCready » Sun Jan 10, 2021 12:53 am

But its one that has been factored into computerized valuations. It seems to me that one of that many strengths shown by Alphazero is its distinct lack of fixed materialization. Given that there are 8 pawns all with different starting positions how can they all be given the same value?

John McKenna

Re: Numerical Valuations

Post by John McKenna » Sun Jan 10, 2021 1:50 am

Don't think I contributed to the older thread.

Glad to have got the chance to add -

https://en.chessbase.com/post/the-joys- ... the-pieces
Last edited by John McKenna on Sun Jan 10, 2021 9:24 pm, edited 2 times in total.

User avatar
JustinHorton
Posts: 10364
Joined: Mon Aug 04, 2008 10:06 am
Location: Somewhere you're not

Re: Numerical Valuations

Post by JustinHorton » Sun Jan 10, 2021 8:09 am

Didn't we do this
"Do you play chess?"
"Yes, but I prefer a game with a better chance of cheating."

lostontime.blogspot.com

Roger de Coverly
Posts: 21312
Joined: Tue Apr 15, 2008 2:51 pm

Re: Numerical Valuations

Post by Roger de Coverly » Sun Jan 10, 2021 11:34 am

MJMcCready wrote:
Sun Jan 10, 2021 12:53 am
But its one that has been factored into computerized valuations.
With ever increasing sophistication until the Deep Mind researchers came up with a different approach.

in the starting position it doesn't matter whether or not all pawns have the same value, because both sides have the same structure. Advantages and disadvantages of different pawn structure have been a part of chess for very many years. Authors seem to have been able to discuss these without worrying about what numeric values to assign. It's quite possible that some of the many people who have written or attempted to write chess engine software have tried this.

John McKenna

Re: Numerical Valuations

Post by John McKenna » Sun Jan 10, 2021 12:04 pm

Yes, Roger, but for us humans is it not helpful to have a finer and more nuanced guide (han the crude: P=1, B =N =3, R=5, Q=9, K=10 to infinity?) to the varying values of the pieces and pawns in mind as the game progresses from start to finish?

Using such a guide productively in practice is the real difficulty due to the changing nature of the position and the varying utility of the pawns and pieces. It can useful to try to keep it in mind and try to apply it when and where possible if time allows.

Some kind of constantly varying "table of values" may exist and become more refined in the mind the stronger a player gets. Who knows?

Roger de Coverly
Posts: 21312
Joined: Tue Apr 15, 2008 2:51 pm

Re: Numerical Valuations

Post by Roger de Coverly » Sun Jan 10, 2021 12:31 pm

John McKenna wrote:
Sun Jan 10, 2021 12:04 pm
Yes, Roger, but for us humans is it not helpful to have a finer and more nuanced guide (han the crude: P=1, B =N =3, R=5, Q=9, K=10 to infinity?) to the varying values of the pieces and pawns in mind as the game progresses from start to finish?
If material is exactly equal in terms of the pieces and pawns on the board, it remains equal regardless of what values are assigned to them. If there's an imbalance, other factors come into consideration, space, initiative and coordination for example before a verdict can be attempted. .

User avatar
MJMcCready
Posts: 3191
Joined: Mon Jun 24, 2013 2:30 pm

Re: Numerical Valuations

Post by MJMcCready » Sun Jan 10, 2021 3:01 pm

JustinHorton wrote:
Sun Jan 10, 2021 8:09 am
Didn't we do this
You sound as happy as ever, as per usual.

User avatar
Matt Mackenzie
Posts: 5232
Joined: Tue Mar 31, 2009 11:51 pm
Location: Millom, Cumbria

Re: Numerical Valuations

Post by Matt Mackenzie » Sun Jan 10, 2021 3:09 pm

MJMcCready wrote:
Sun Jan 10, 2021 3:01 pm
JustinHorton wrote:
Sun Jan 10, 2021 8:09 am
Didn't we do this
You sound as happy as ever, as per usual.
Just asking, had you forgotten you started a previous thread on this topic less than three months ago?
"Set up your attacks so that when the fire is out, it isn't out!" (H N Pillsbury)

User avatar
MJMcCready
Posts: 3191
Joined: Mon Jun 24, 2013 2:30 pm

Re: Numerical Valuations

Post by MJMcCready » Sun Jan 10, 2021 4:55 pm

Except he's not just asking is he?

User avatar
JustinHorton
Posts: 10364
Joined: Mon Aug 04, 2008 10:06 am
Location: Somewhere you're not

Re: Numerical Valuations

Post by JustinHorton » Sun Jan 10, 2021 5:30 pm

<shrug> I am really unsure why you started another thread as if the previous one didnt exist - why not just return to the old one if you have anything further to add?
"Do you play chess?"
"Yes, but I prefer a game with a better chance of cheating."

lostontime.blogspot.com

John McKenna

Re: Numerical Valuations

Post by John McKenna » Sun Jan 10, 2021 9:23 pm

Don't think I contributed to the older thread.

Glad to have got the chance to add -
Which, despite Roger's modern machine-like approach, goes all the way back to a rukh is worth one dirham (and the rest is history).