Which is the maddest, craziest engine?

[jefk]

castle opposite and pawn storm whenever possible.

ideally it then should play with an opening book which facilitates such plans

NB this is not so easy, as it usually depends on what the opponent plays/has played
whether castling long is a good idea. Example after 1.e4 most opening lines result in short
castling (both sides) Although in the Sicilian, eg Najdorf (and Richter Rauzer)
and Pirc there are some/quite a few lines where castling long is a good idea.
An easier plan is e.g a line as 1.d4 Nf6 2.g4!? (...Nxg4? 3.Rg1 with castling long later on.

For Black maybe similar, but even more 'difficult' (just a matter of working it out, not
for getting crazy games btw but more for getting a special kind of unbalanced games

PS forum is rather slow again at this moment

[OliverBr]

I would like to see an engine with a UCI setting to castle opposite and pawn storm whenever possible.
That would be a blast.

I can do it with OliThink. I would create a branch in GitHub and a nice LiChess Account, similar to OliThinkFlash, OliThinkNomaterial, OliThinkUnderpromotion.
It wouldn't be a UCI setting first, but it could do:

1) wait until opponent castles (question is waiting how long? also: what if he doesn't castle at all?)
2) then castles opposite, if possible.
3) storms each of the 4 other pawns. Maybe a fifth? (No, let's keep it cilivized... for now):

There is a catch: It has to disregard book openings, unless someone is prepared to prepare a special set of openings..

[OliverBr]

I would like to see an engine with a UCI setting to castle opposite and pawn storm whenever possible.
That would be a blast.

1) wait until opponent castles (question is waiting how long? also: what if he doesn't castle at all?)
2) then castles opposite, if possible.
3) storms each of the 4 other pawns. Maybe a fifth? (No, let's keep it cilivized... for now):

More precisely:

1) Opponent castles => Castling to the opposite side (with pawn-storming oponnent's castling side)
2) Opponent loses castling rights to one side => Castling to that side (and pawn-storming the other side)
3) Oponnent loses all castling rights => Castling to the side, where the king-is-not (and pawn-storming the other side)

There is a catch: It has to disregard book openings, unless someone is prepared to prepare a special set of openings..

This is still unsolved, because the UCI protocol give all power to the GUI. So the engine cannot decide, when to leave the book. Any suggestions here?

[OliverBr]

...

There is a catch: It has to disregard book openings, unless someone is prepared to prepare a special set of openings..

This is still unsolved, because the UCI protocol give all power to the GUI. So the engine cannot decide, when to leave the book. Any suggestions here?

I have abandoned the project without coding one single line. The stateless concept of UCI (and so of all modern engines) does not support it, with reason: A position does not change because of what happened in the history. (except three-fold repetition, a completely other topic).