I don't understand the difference between the evaluation of the position and the evaluation of the different variants that are shown in the analysis.
When a certain line is chosen, the evaluation of that new resulting position should be the same as that shown in the evaluation of the line shown before, but it is not. Why is it not the same?
For example if we play 1. d4 Nf6 2.c4 e6 the position evaluation is +0.3.
And here we have some lines options:
a) 3. Nf3... Line Evaluation is +0.3
b) 3. g3 ... Line Evaluation is +0.2
c) 3. Nc3 ...Line Evalutation is 0.0
d) 3. a3 ...Line Evaluation is 0.0
e) 3. e3 ... Line Evaluation is 0.0
But if I choose c) option, 3.g3, after new position evaluación we can see 0.0 instead of +0.2
?
What do you mean?
For starters, an engine evaluation near the beginning of the game needs to run for a long time. Just a few moves in, anything within a third of a point might be equivalent.
Look at something 10 feet away. Take a step. It now appears slightly closer. Same with the engine. After completing a step, the goal is a little closer.
Your unstated assumption that the position will have One True Value is not correct. The engine is making its best guesses. As it gets more information, it makes better guesses.
Just disregard engine evaluations in the early opening. 32 men positions are too complex with too many possibilities to come up with any meaningful engine evaluation within any calculation horizon.
I have not asked the question correctly. My english is not very good.
On the lichess website / tools / analysis board. We can use both database and computer analysis enable.
After navigate thow the database for example with 1. d4 Nf6 2.c4 e6 the position evaluation is +0.3. This is shown at the top.
Also we get best continuations with their corresponding evaluations. Correct, right?
Here we have the lines options sugested by Stockfish 12+WASMX Depth 43CLOUD:
a) 3. Nf3... Line Evaluation is +0.3
b) 3. g3 ... Line Evaluation is +0.2
c) 3. Nc3 ...Line Evalutation is 0.0
d) 3. a3 ...Line Evaluation is 0.0
e) 3. e3 ... Line Evaluation is 0.0
In this moment the position evalution is +0.3, and after click 3.g3 for example the new position evaluation must be +0.2 how we can see in the options listed above. But no, the new evaluation of 1. d4 Nf6 2.c4 e6 3. g3 now is 0.0, instead of +0.2 as expected.
What is the difference then between the evaluation of the position and the evaluation of the lines?
#5
None of these evaluations has any real meaning.
You get a more or less correct evaluation around move 13, not move 3.
#5 "In this moment the position evalution is +0.3, and after click 3.g3 for example the new position evaluation must be +0.2 "
No. Once you click 3.g3, the engine looks farther and gets a different eval. This is what ericmsd in #3 was pointing out.
@tpr "You get a more or less correct evaluation around move 13, not move 3."
Is the same in all cases, move 13 than move 45 or move 78. The position evaluation no match with line evaluation. I have check this a lot of times. I want to know why. If the expected eval before move is x when you play that move, the new position eval should match with the predicted x.
In all fairness they are similar, but they are not the same. Why?
As you chose you half-move candidate, the engine keeps pushing its horizon boundaries by the same amount of chess time unit.
hence it has the possibility of "seeing" new positions where to call its evaluation function, and hence new information sent back the path you went into.
Ok, I think I get it. Thank you all.
Let's say that the evaluation of the position corresponds to, for example, move N and the evaluation of the line with move N-1.
