About gambit play, a topic which occasionally became popular
in relation to the Patricia(3.1) engine, i had a look to a (pdf) book
which claims to list all gambits (and their names):
https://scacchiegambetti.blogspot.com/2 ... bolic.html
It is a very long E- ‘book’ (a large pdf), ten thousand pages, but most is just opening tables printed with Scid.
As some of you know i’ve uploaded a ‘free’ (opensource) program Bookbuilder where it’s possible to add (sub) variation names to opening variations
https://sourceforge.net/projects/bookbuilder/
Although i already had included quite some gambit names, i now went through the gambit names in the above ‘encyclopedia (and variations). First checked if the gambit moves were already in my Bbuider book and then added the name if it wasn't there already. My list of gambit names wasn't certainly complete
so I added much more. The result (new gambit names) you can see in the file ‘userns’ which you can find in the sourceforge entry for Bookbuilder (listed as a separate file but also included in the big bb4setup.exe).
The main use of this (these names) for human players is that with knowing the (gambit) names, it’s easier to remember specific gambits instead of simply algebraic number sequences (unless you have an eidetic memory). NB there are no unique definition for gambit names; not even for opening names but that’s a broader topic: there for example also is a ‘Caxton list’ of opening names:
https://sourceforge.net/projects/bookbu ... f/download
But i haven’t checked there the gambit names (sometimes it’s also not so clear if something is a gambit or not).
Then there also is quite a long list of openings (and names in the appendix of the ‘ OXFORD Companion’ to chess
www.amazon.com/Oxford-Companion-Chess-S ... 0198661649
As for opening names in general, apparently the Fide years ago tried to come to a standardization of opening names, but didn’t never was fully accepted. There are names attached to the various ECO codes, but this doesn’t give enough detail, thus New in Chess years ago invented another more detailed code system; also never fully accepted (and also not always names attached to the different codes). Whether the list of names in the Oxford companion is the ‘best’ list of names i dont’know; programs as Fritz and Chessbase are not so good in displaying names of opening variations; chess.com is a lot better. Anyway, back to gambits, i’ve updated my gambit polyglot book which you can find here:
https://sourceforge.net/projects/chess- ... n/download
The more solid gambits have higher playing frequencies, the wilder ones
lower frequency (set sometimes a bit arbitrarily i admit) and I omitted
quite some crazy/erroneous gambits whether they have a name or not.
Finally i’ve let Patricia 3.1 play quite some ( blitz) games against other similar strength engines with conventional book, and filtered out with the Pohl tool the most interesting wins (sacrifices etc),
result some 511 games (360 with 3/1 timing, and the rest 7/3):
https://www.dropbox.com/scl/fi/rb7ey058 ... aqx1l&dl=1
You will see wild games, with interesting combinations, (more/mostly gambits, and more solid –and much less incorrect (*)- sacrifices and/ or combinations). Mostly interesting for human players i suspect (and P3 fans of course). Who can simply load them in eg. the free Chessbase Reader, and then have a look at the games (which are adjudicated automatically once the scores were about +6 or so for P3). For some players (more intermediate than beginners) this can have educational value, of course using such sharp tactics is not the way to beat computers but against humans it can work sometimes (actually when having a much stronger opponent, it’s often advised to complicate the game, in order to have at least a little chance, otherwise you will be slowly demolished). Also in blitz, using gambits (if you know them) can be fun of course..
(*) against SF17 on a fast comp some of these combinations might not work, ie. be entirely correct but
then i’m waiting for a stronger Patricia, and later i will try similar games with rapid control.
https://sourceforge.net/u/jcek/profile
gambits (and their names)
Moderators: hgm, chrisw, Rebel
-
jefk
- Posts: 833
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
-
Graham Banks
- Posts: 42938
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
-
jefk
- Posts: 833
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
Re: gambits (and their names)
Related to gambit play is ofcourse the question how to play ‘best’ chess in the opening phase. And this again is related to the issue if a win for White is possible (or not, the later begin true.) As posted some time ago in the general topics (a discontinued thread with topic ‘proof’) , I explored my reasoning method for games as chess and "Connect Four," and now found it even more applicable than my earlier analysis for chess. As written up more in detail in the subtopic ‘kindergarten’ because that is where an earlier thread was moved to by the moderator (after some provocative scorn and criticism.
With a branching factor of 7 (compared to about 20 in chess), the first player can win in ten ways after four moves, while the second player can prevent this in 28 ways. After five moves, these numbers are 27 and 40, respectively, and continue to grow: 45/57 after six moves, 62/73 after seven, and 84/92 after eight. The increase in winning possibilities outpaces the prevention options, leading to the conclusion that Connect Four can theoretically favor the first player. This phenomenon is analogous to the Achilles and the Tortoise problem; initially, it seems the second player’s prevention options exceed the first player’s winning chances, but eventually, the first player can secure a win. This has been supported by earlier research, including Allen/Allis (1998) and subsequent number crunching.
My approach demonstrates a theoretical win for Connect Four, a step beyond earlier solutions by Allis and Allen, who used computational methods. I propose an "ultraweak solution" (UWS) to illustrate the first player’s potential for victory. Although my reasoning may seem simple, it’s grounded in established heuristics in mathematics.
While skeptics may dismiss my findings, similar doubts arose when Allen first published his solution, which was later corroborated by Allis. Though some criticized the computational feasibility at the time, Allis's strategies ultimately confirmed that the first player can always win. Conversely, in chess, the number of prevention options grows much faster than possible winning moves, suggesting a theoretical draw. While this may not satisfy all critics, it aligns with established mathematical principles. I am currently refining my approach for chess, similar to my analysis of Connect Four. My reasoning implies that searching for a winning opening for White may be futile, as the game trends towards a draw with optimal play.
In conclusion, my method has shown promise for both Connect Four and potentially chess. Debates over definitions of "ultraweak solutions" are less relevant given modern perspectives on mathematics. Ultimately, I believe exploring opening strategies, especially in human play, offers more value than simply seeking the strongest engine.
This is the reason I find the Patricia engine interesting (as discussed in the topic ‘gambit’s) and you also can see it sometimes in some modern chess books as eg:
https://www.newinchess.com/beyond-material
________________________________________
With a branching factor of 7 (compared to about 20 in chess), the first player can win in ten ways after four moves, while the second player can prevent this in 28 ways. After five moves, these numbers are 27 and 40, respectively, and continue to grow: 45/57 after six moves, 62/73 after seven, and 84/92 after eight. The increase in winning possibilities outpaces the prevention options, leading to the conclusion that Connect Four can theoretically favor the first player. This phenomenon is analogous to the Achilles and the Tortoise problem; initially, it seems the second player’s prevention options exceed the first player’s winning chances, but eventually, the first player can secure a win. This has been supported by earlier research, including Allen/Allis (1998) and subsequent number crunching.
My approach demonstrates a theoretical win for Connect Four, a step beyond earlier solutions by Allis and Allen, who used computational methods. I propose an "ultraweak solution" (UWS) to illustrate the first player’s potential for victory. Although my reasoning may seem simple, it’s grounded in established heuristics in mathematics.
While skeptics may dismiss my findings, similar doubts arose when Allen first published his solution, which was later corroborated by Allis. Though some criticized the computational feasibility at the time, Allis's strategies ultimately confirmed that the first player can always win. Conversely, in chess, the number of prevention options grows much faster than possible winning moves, suggesting a theoretical draw. While this may not satisfy all critics, it aligns with established mathematical principles. I am currently refining my approach for chess, similar to my analysis of Connect Four. My reasoning implies that searching for a winning opening for White may be futile, as the game trends towards a draw with optimal play.
In conclusion, my method has shown promise for both Connect Four and potentially chess. Debates over definitions of "ultraweak solutions" are less relevant given modern perspectives on mathematics. Ultimately, I believe exploring opening strategies, especially in human play, offers more value than simply seeking the strongest engine.
This is the reason I find the Patricia engine interesting (as discussed in the topic ‘gambit’s) and you also can see it sometimes in some modern chess books as eg:
https://www.newinchess.com/beyond-material
________________________________________
-
Dann Corbit
- Posts: 12701
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: gambits (and their names)
Here is your gambits.bin file as PGN and EPD:
The EPD file contains the terminating EPD position for each of the PGNs.
The EPD file contains the terminating EPD position for each of the PGNs.
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.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
-
Dann Corbit
- Posts: 12701
- Joined: Wed Mar 08, 2006 8:57 pm
- Location: Redmond, WA USA
Re: gambits (and their names)
Either I made a mistake in filtering or there are many duplicates in the Gambits EPD file because the gambits file shrinks enormously when I filter for uniqueness (less than 5000 unique positions):
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.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
-
jefk
- Posts: 833
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
Re: gambits (and their names)
Dann, it's maybe because of many transpositions that there occurs the shrinking you talked about.
And thus may occur when you do the same thing (making epd's and a pgn) out of a polyglot book.
As for the pgn, in principle you could use it to make a book in other formats, (eg abk or ctg) but
the pgn you made has lost the properties which were in the polyglot format, namely, that
the book tries to play gambits as often as reasonably possible (using the frequencies).
When i made the first version, i mixed a pgn with gambits for White and gambits for Black,
in such a way that the mixed result will (almost) always choose the gambit move (using frequencies
of 100 pct or 0 pct in polyglot), if you see what i mean (there was quite some subjective fuddling
in later versions with the frequencies); in practice the cannot always play a gambit, as the
opponent can often avoid (or decline) the gambit.
The recent 'jja' book conversion tool would probably be more suitable to convert 'my'
gambit book to other formats btw (tried it once to get a ctg format but i wasn't
happy with the result, so i kept working only on the .bin version(s)
regards
PS and i'll now have a look to your posting about declined gambits
And thus may occur when you do the same thing (making epd's and a pgn) out of a polyglot book.
As for the pgn, in principle you could use it to make a book in other formats, (eg abk or ctg) but
the pgn you made has lost the properties which were in the polyglot format, namely, that
the book tries to play gambits as often as reasonably possible (using the frequencies).
When i made the first version, i mixed a pgn with gambits for White and gambits for Black,
in such a way that the mixed result will (almost) always choose the gambit move (using frequencies
of 100 pct or 0 pct in polyglot), if you see what i mean (there was quite some subjective fuddling
in later versions with the frequencies); in practice the cannot always play a gambit, as the
opponent can often avoid (or decline) the gambit.
The recent 'jja' book conversion tool would probably be more suitable to convert 'my'
gambit book to other formats btw (tried it once to get a ctg format but i wasn't
happy with the result, so i kept working only on the .bin version(s)
regards
PS and i'll now have a look to your posting about declined gambits