A question about the value of books in closing gap between egnines

[Uri Blass]

SSDF list has for example
1 Lc0 0.29.0 Cuda-808544 3060Ti 3586 30 -28 600 63% 3493
2 Stockfish 16 x64 1800X 3.6 GHz 3582 50 -48 200 58% 3523
132 Fruit 2.2.1 Athlon 1.2 GHz 2830 18 -18 1465 59% 2767

Suppose you give fruit a very good book to fit the style of fruit(not only good moves but moves that do not lead to positions that fruit blunders).

1)How many rating points can fruit earn in the SSDF?

[Jouni]

My guess: 10-20 points = practically nothing.

[Peter Berger]

SSDF list has for example
1 Lc0 0.29.0 Cuda-808544 3060Ti 3586 30 -28 600 63% 3493
2 Stockfish 16 x64 1800X 3.6 GHz 3582 50 -48 200 58% 3523
132 Fruit 2.2.1 Athlon 1.2 GHz 2830 18 -18 1465 59% 2767

Suppose you give fruit a very good book to fit the style of fruit(not only good moves but moves that do not lead to positions that fruit blunders).

1)How many rating points can fruit earn in the SSDF?

About 100 would be my guess.

It is not rocket science either. You give it the ability to draw more games with black. Style doesn't matter too much.

Easiest would probably be to just let it play the Berlin against 1. e4.

[jdart]

My experience is that books matter less for the very top engines such as Stockfish. These engines have high draw rates when matched against other similarly-ranked engines, and can regularly draw from even an inferior position. (That is why TCEC uses very unbalanced opening posiitions). But for lower-ranked engines, it does make a difference. Books tend to be constructed from high-level games, including correspondence games and engine games. So the engine is basically playing Stockfish or Stockfish-level moves in the opening. That can be an advantage, if it is lower-rated than Stockfish.