Gothic Chess Opening Database??

[ColonelCrockett]

I clicked on the link in ChessCarpenter's old post and aparently the "Name your own Gothic Opening" webpage is no more? (or maybe the link was corrupted or has been moved? Could someone set me straight on this?

[GothicInventor]

Yes, that page was an "experimental" attempt to document some openings that were being played repetitively. It became futile to describe openings, as pawn moves could take you out of entire lines of play early on. This is something still being looked into.

There are some named Gothic Chess openings. I will have to publish another page of them.

[G0ldfish]

Privately I've been working on an opening database of Gothic Chess®, incorporating evaluations and known opening lines from numerous programs, such as Gothic Vortex®, Joker80 and SMIRF. So far however, this project is not open to the public since I'm unable to make a proper, easy to search and understand, web-based display of it that could be expanded as exploration goes. If one could write a separate program capable of outputting the knowledge to HTML, that would be neat; or alternatively, something like an applet that could handle such a database. This way, it (probably) is bound to stay in a Word doc format. Which is fine for me, perhaps not so for others... oh well.

Volunteers can mail me at DELETEp-kun80THIS@hotTHISmail.comTOO

GoldFish

[H.G.Muller]

Note that I changed the piece values in Joker80 last week to a set that was really measured from 10x8 games, rather than the usual values of 8x8 Chess. This seems to have a benificial effect on its rating of about 100 Elo points.

But I also expect these new piece values to have a profound effect on Joker80's opening play. In particular, it should be a lot more reluctant to trade Bishops for Knights, as the old version used to do quite often in order to reap some relatively small Pawn-strucure advantage (e.g. leaving the opponent with a doubled Pawn). Even gaining a full Pawn in a Bishop vs. Knight trade is now only neutral, if the Bishop was part of a pair (which, early in the opening, it usually is).

The opening lines played by the older version of Joker80 should therefore be considered suspect. I would not be surprised if the major part from the rating increase was due to suppressing losing opening trades of B vs. N. I saw many games where the old Joker had two Knights against the opponent's two Bishops...

[G0ldfish]

I hate to admit, dear Mr. Muller but for the time being (as the idea of compiling an opening book for GC ® came with SMIRF), I am not incorporating the evaluations of Joker80, only its suggested moves, that are subsequently evaluated by SMIRF for their numerical value.
Don't get me wrong, I'm not trying to say that the evaluation function or its numerical values are in any manner inferior in case of Joker80 -- it is just a habit that stays sorta hard to break, plus the convenience that I can cut&paste the lines in case of SMIRF (without needing any further conversion, like b1c3 to Nc3 or such) whereas in Winboard_F it is more problematic. Joker80 itself is a brutally strong one, hats down.

Don't let my current stance bother you, please; keep up the exquisite work.
GoldFish

[H.G.Muller]

Well, no problem. Just wanted to point it out. If the evaluation is faulty, the moves it prefers might also be bad moves. But if you only use it as a source of possible moves or likely errors, I guess it is OK. I don't know what procedure for building the book is.

I would be careful with Smirf's evaluations, though: they seem to be off by more than two Pawn regularly, even in nontactical positions. Most likely due to an unrealistically low evaluation of the Archbishop.

What exactly is the problem with WinBoard_F? (This to see if it can be fixed.) You do get the moves that are actually played in SAN there too, if you save the PGN, isn't it? Or is the problem that my engines do not support analysis mode?

[G0ldfish]

As of the problems... the "move now" menu option doesn't exactly do that. It stops thinking... in a while. Analysis can be done if the clock settings are set ridiculously high, and the first few moves, leading to the position in question, are done with "Edit game". But the engine window shows a different format, not like how it would appear in the "Move history" if the moves were made.. and since I'm interested in the lines, I do include them. Also, the functionality apparent in SMIRF (that shows the second best move as well, in parentheses) would be useful too.

If these are to be introduced, it would be invaluable
GoldFish

[H.G.Muller]

OK, these are not so much WinBoard_F problems (all this is supported there), but engine problems. My engines do not support analysis mode or the move-now command (and neither does Smirfoglot), and print their PV during thinking or pondering in long-algebraic, not as SAN. (WinBoard does consider this a comment field, and does not interpret it; it just passes it on.)

Logically, converting the PV to SAN, and giving the user the option how he wants the PV displayed, are GUI tasks: it would be silly to duplicate the effort in every engine. Unfortunately converting moves to SAN is a very cumbersome task, for which the GUI would really have to play out the PV (to know where pieces are standing, for disambiguation). And what is worse, as currently there are no restrictions on the format they use for printing the PV, needing a smart parser to extract the moves from between possible other chatter. I guess this is why no GUI I know implements it. Perhaps in the future, if I have a few spare days...

[G0ldfish]

Thank you in advance. In the meantime, I changed a few things in the opening collection format, and this way, it will incorporate evals as well as lines from Joker80/WinBoard_F. The SAN/LAN problem is being solved locally, although it's still a bother... oh well, if I have time to let a handful of engines calculate opening lines to depth 12-16 or so, I might just as well have the time to convert the lines manually

Anyhow, thank you.
GoldFish

[H.G.Muller]

Perhaps the following 'trick' might be of use:

Copy the position from which you do the analysis to the clipboard, and paste it into NotePad. This gives you a bare FEN string; type [FEN "..."] around it to make it look like a PGN tag.

Then copy/paste the PV you want to see from the engine-output window to behind the FEN tag in NotePad.

Start a second WinBoard_F in game-viewer mode, and copy the FEN-tag + PV from NotePad, and paste it as a game into this WinBoard_F. It starts to auto-play it, but clicking the >> button fast-forwards it to the end. Then you copy the game again to the clipboard from this second WinBoard_F, and paste it into NotePad. Now the moves will all be in SAN.

This is still a bit clumsy, mainly because you have to glue the FEN of the initial position to the PV. This because "paste game" will automatically setup a new game. It would be very nice if there was a command that allowed you to paste in some moves that were then done from the current position. Then you could skip the NotePad stuff, and directly paste PVs from one WB into the other, and copy PGN to the clipboard from there to recover the SAN. There is a menu item "step -> type in move ...", but it accepts only a single move. Perhaps I could make it such that multiple moves typed in here in edit-game mode would all be entered, so that you could paste the PV in there.