The Official Fishbook Opening Books Released

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Dann Corbit
Posts: 10476
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: The Official Fishbook Opening Books Released

Post by Dann Corbit » Fri Oct 21, 2016 9:46 am

There are five and a half million distinct positions in the book.
All the arguments for and against are purely hand-waiving.

The value (or lack thereof) of the book can only be proven by direct tests.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

drj4759
Posts: 86
Joined: Mon Nov 17, 2014 9:05 am

Re: The Official Fishbook Opening Books Released

Post by drj4759 » Sat Oct 22, 2016 10:33 am

Fishbook was not designed to defeat Brainfish but to neutralize it. I Repeat... TO NEUTRALIZE.

Brainfish is being updated everday, so is Fishbook. It would be like both are dogs chasing their own tails in vain. For as long as Brainfish is beating Fishbook, it will use the lines that beat it to apply it as dose of Brainfish's own medicine. There is no end to this until the respective authors gets bored or the opening lines are exhausted or there is no room for Arena which has a 2GB file limit.

Brainfish is an opening specialist, not a generalist. It specialize in the e4-e5 opening sequence. When I discovered it in the middle of my development, I just assimilated most of its e4-e5 opening repertoire and then bingo!, it was neutralized.

So therefore, do not conclude that Fishbook is not useful for general purpose opening use. There is no restriction that it will behave only in that way. It can be flexible when the need arises as when it found difficult to beat Fauzi 2.2 (the strongest before Brainfish) at first. But with training and assimilation, it finally beat it.

In the end the chess opening will reach near perfection that most games will end in draws when both chess engines are approximately of the same strength.

JJJ
Posts: 1295
Joined: Sat Apr 19, 2014 11:47 am

Re: The Official Fishbook Opening Books Released

Post by JJJ » Sat Oct 22, 2016 11:15 am

fantasmadel50 wrote:The cerebellum a book engine, .bin, therefore any gui you have to enter the path, depending on where your pc has istalado in the book. 1) Position the engine in the gui, 2) go into its configuration (engine parameters) 3) I am writing a book path 4) AHI I write Route 5) EJ C: \ BRIANFISH \ Cerebellum_Light.bin
Thanks, but when I do that, Arena is saying it couldn't open the path of the book when I start the engine.

leslies
Posts: 21
Joined: Fri Jan 11, 2013 4:04 pm

Re: The Official Fishbook Opening Books Released

Post by leslies » Sat Oct 22, 2016 1:51 pm

the first thing you have to do is copy and paste cerebellum book folder onto the brainfish engine folder,that way the gui can find it and open it.I hope this helps.

jefk
Posts: 395
Joined: Sun Jul 25, 2010 8:07 pm
Location: the Netherlands
Contact:

Re: The Official Fishbook Opening Books Released

Post by jefk » Sat Oct 22, 2016 2:00 pm

drj4759 wrote:Try it before posting sarcastic comments and shut up if you find out.
well i see that the time control of the games played was 1/1,
so we could wonder how this book would perform
under slower time controls, eg standard.
yes, it's a matter of trying, but it costs time you know
jefk

drj4759
Posts: 86
Joined: Mon Nov 17, 2014 9:05 am

Re: The Official Fishbook Opening Books Released

Post by drj4759 » Sat Oct 22, 2016 2:29 pm

Time control probably does not matter with an Arena book. Internally, Arena does not record the strength of a move by way of evaluation score. Instead, it classifies moves as either win, loss or draw. In essence, Arena is collecting statistics to control the choice of moves which is refined by filtering minimum number of moves, minimum wins, white win%, black win%, priority, number of games and win percentage.

I used 1/1 time control to collect games in the first half stage of Fishbook development, then increased it to 2/2 in the next quarter and 3/3 in last stage. The data is not uniform but is the result of reacting to the results of each PGN collection run with various random parameters.

I have yet to hear somebody's tests that has Brainfish/Cerebellum trashed Fishbook with Goby. It would be nice if someone with spare time and resources could test Fishbook at longer time controls and post the same for everybody's benefit.

Post Reply