A new book for Fruit-derived engines - 600+ first games

[Marc Lacrosse]

As Toga is presently the only free actively developed opensource engine in the Fruit family and probably the best free opensource engine ever produced it could be nice to provide a good tailored book for it.

I am busy trying to produce something significantly better than the books I made for Fruit 2.2. (this is quite an interesting personal challenge for me as I think these books were good).

The book is tuned for Toga and mainly tested with it but it is my strong conviction that it should also do very well for Fruit 2.1 or even Fruit 2.2/2.2.1 and GambitFruit 4bx.

I have a few preliminary results with an intermediate version.

All these are blitz games with ponder off on a Athlon 2900+ with 128MB hash, tablebases off.
Hiarc10, Shredder9, Fritz9 and Fruit 2.2.1 playing with default parameters and a neutral book :

Code: Select all

Toga II 1.2 B2a - Hiarcs 10          32.0 - 18.0   +24/-10/=16 64.00%
Toga II 1.2 B2a - Shredder 9.1 UCI   35.0 - 15.0   +28/-8/=14  70.00%
Toga II 1.2 B2a - Fruit 05/10/15     34.5 - 15.5   +23/-4/=23  69.00%
Toga II 1.2 B2a - Fritz 9            35.0 - 15.0   +28/-8/=14  70.00%

I also got the same unfinished book tested against Rybka 1.01 Beta 13d 32-bit playing with its Jeroen Noomen book :

Code: Select all

Toga II 1.2 B2a - Rybka 1.01 Beta 13d 32-bit   162.0 - 238.0   +89/-165/=146 40.50%

Rybka is really on another planet!.
However, as far as I know this is the first record of an engine reaching a 40%+ result against Rybka.

I am busy testing Toga + my new unfinished book at blitz against Fruit 2.2.1 + performance.bin
So far the results are +6/-2/=12 (60.0%) for Toga.

So the new Toga (with this book) could well be the best presently available engine behind superstar Rybka at least in blitz conditions ....

Marc

P.S.

The book should be available at mid-april.
It will be offered for free with the following restrictive conditions:
- its use (as is or modified by any mean) will have to be mentioned together with my authorship in any published tournament results and games.
- its use (as is or modified by any mean) will have to be mentioned together with my authorship whenever an engine makes use of it in an internet-relayed game and/or tournament.

[Roger Brown]

(snipping all the way)


The book should be available at mid-april.
It will be offered for free with the following restrictive conditions:
- its use (as is or modified by any mean) will have to be mentioned together with my authorship in any published tournament results and games.
- its use (as is or modified by any mean) will have to be mentioned together with my authorship whenever an engine makes use of it in an internet-relayed game and/or tournament.

Hello Marc,

I am interested in precisely how you will enforce this. I assume that you are counting on moral persuasion.



I urge you to copyright it in some way. I do not know how opening moves can be copyrighted but I would think that an opening book would qualify as intellectual property.

Later.

[Marc Lacrosse]

The book should be available at mid-april.
It will be offered for free with the following restrictive conditions:
- its use (as is or modified by any mean) will have to be mentioned together with my authorship in any published tournament results and games.
- its use (as is or modified by any mean) will have to be mentioned together with my authorship whenever an engine makes use of it in an internet-relayed game and/or tournament.

Hello Marc,

I am interested in precisely how you will enforce this. I assume that you are counting on moral persuasion.



I urge you to copyright it in some way. I do not know how opening moves can be copyrighted but I would think that an opening book would qualify as intellectual property.

Later.

There is no way to enforce anything like this.
As it will be given for free, I just hope moral persuasion will be enough.
As for copyrighting the book I am sure I will not do it.
Isn't intellectual ownership de facto protected as long as there is enough evidence that you are the real owner or producer of something?
Sure I would not be glad if someone decides to sell or even give it for free under his name as if it had been made by himself .
But yet if the book is a little bit modified I cannot imagine how I could prove anything.
And in the simpler case where someone would have an engine playing a few games online or in a tournament with a book of mine without telling it, how could I even myself detect it ?

By the way, are other engine books copyright-protected ?

Marc

[Roger Brown]

There is no way to enforce anything like this.
As it will be given for free, I just hope moral persuasion will be enough.
As for copyrighting the book I am sure I will not do it.
Isn't intellectual ownership de facto protected as long as there is enough evidence that you are the real owner or producer of something?
Sure I would not be glad if someone decides to sell or even give it for free under his name as if it had been made by himself .
But yet if the book is a little bit modified I cannot imagine how I could prove anything.
And in the simpler case where someone would have an engine playing a few games online or in a tournament with a book of mine without telling it, how could I even myself detect it ?

By the way, are other engine books copyright-protected ?

Marc

Hello Marc,

I am unclear on this point but I would think that if the book is sold as part of a software package - such as the Lokasoft and Chessbase products are - which is itself copyright protected then everything included in the package (on the CD etc.) is covered unless expressly not so covered.

I am not a lawyer however so that may not be so.



All the best.

Later.

[Marc Lacrosse]

With the preliminary version of my new book, Toga II 1.2b2a has just beaten Fruit 2.2.1 playing with its own book.

Code: Select all

1 Toga II 1.2 Beta2a - new book +42/-14/=44 64.0/100
2 Fruit 2005/10/15 - own book   +14/-42/=44 36.0/100

One hundred games is a small number but according to Steve Maughan's "WhoIsBetter" utility, Toga's superiority was so clear in the test that there is less than 1% chance that Toga is not really stronger than Fruit in these testing conditions (blitz, ponder off, no TB)...

Marc