book builder file

[Dann Corbit]

I added a bunch of data to the default bookbuilder file.
I only added position and ce because that is all that I know how to do.
Anyway, if you use bookbuilder, here is the file (careful, I used the same name as the original):


All the added positions have been played at least five times. I did not add endgame positions because they don't belong in an opening book.

[Dann Corbit]

I tried to make a Rebel rbm file from it, but at about 1 GB of output, the program crashed.

[jefk]

ok, interesting.
as for making an rbm file, i've done it in the past, but maybe
the file you made became too big and there's a bug...

[jefk]

PS which engine did you use to add the extra (end) nodes ?
my plan here is to analyze them with the latest Obsidian, but i'll have a look
what you made (also for some correspondence games i've recently started).
I thought the default book was book.pbs (maybe bb.pbs was something else).
And discovered a mistake in the line e4 e5 Nf3 Nc6 and then Nfg1?? again.
So i removed the latter move, done a full minimax,
then 1.e4 gets 0.16, 1.Nf3 0.12 1.e3 or d4 0.10 and 1.c4 0.06
But there are some odd lines as e4 e5 Nf3 Nc6 Na3!? with highest eval
followed only by .. h5. Which again shows that this method of only
minimax (not combined with game statistics) has it's drawbacks

[jefk]

so, somehow you didn't use the latest book.pbs i think which was part of the
bbsetup,exe end September. As i'm becoming lazy my intent is not to produce
a new setup every month or so, but post updates in the files/updates/ directory.

Made some corrections to your work, and merged with my latest book , endnodes
analyzed with Obsidian finished for about 75 pct, result called merged.pbs.

https://sourceforge.net/projects/bookbu ... s/download

What's the use of all this, some may ask, as we already have the Chinese database,
and indeed if i would continue with book building, it would become similar, with lots
of 0.0 evals (depending on engine, with Obsidian the nr of zero's is smaller);
and for quick and simple corrections to the bb book, i sometimes simply look
at this Chinese database https://www.chessdb.cn/queryc_en/

Answer (whats the use of it): well for me it's learning by repetition, when i'm fiddling around
with new opening books, somehow some variations are kept in my mind, which is useful
for human otb or online chess. But of course many players have/use different
methods for learning opening lines in their repertoire(s), whereby the latter can be
different for blitz (eg. with gambits) or slow, more serious games (with 'best' lines).
At higher levels of play people learn more opening theory from playing actual games
(in their repertoire) than by continuous computerized 'book building'.

Anyway, in a week or so another update of this merged book, with the rest of my book.pbs endnodes
analyzed with Obsidian, because these evals make better distinction i think which sort of
endgames are easier (or more difficult) to play . Whereby the latest book.pbs will be included in
a new setup file (whether the merged.pbs would become more suitable as default book
remains to be seen for me, to be decided after further inspection (and/or correction). Overall i'm
not going to spend much more time on such exercises. Besides the mentioned the Chinese database
there also exists the -comprehensive- opening theory from Chessbase, whereby imo
understanding (of opening plans etc) is more important than remembering -and/or
continuously updating- computer moves and their evals.

[Dann Corbit]

The analysis is with Stockfish.
In general, the depth of analysis is between 70 and 36 plies for most of the data.
Some rarely played positions are only to 30 plies
Some of the rarely played leaf nodes are only to 23 plies.
Rarely played means 5 games or less. I have a large database, so there are still a lot of games with 5 position repeats.