Any More Strategical Ideas for a Test Suite?

[fern]

Perhaps it lacks a device to koiw when to fight for a draw. It would involve a change of heuristics. BTW, not to foster pieces changes...
Fern

[Nimzovik]

Sweet Rosy O'Grady! I do not believe my ears (eyes.) Finally we are talking a relative paradigm shift. A program that can play like Nimzovich. A truly Hypermodern program. (unlike the Hypermodern setting in the Hiarcs program which is somewhat misleading) I have been begging for this for years. Pursuant to this of course ....I wonder if you will find such positions in the old masters games that you are looking for. Not to mention that you might look for games that play on 'one color of squares' like Nimzo did. Tho I do not believe the implemetation of such is going to be easy as this appears to be a Humans forte. I have noted historically that Kittingers W-chess program I believe was aimed toward this end (Hypermodern type chess) .... One might examine his program.

[swami]

Some more Ideas:

• Passive Defense
  The only defense possible albeit passive but enough to slow down or put end to opponent's attack.
  
  Undermining the Centre
  Placing pieces in such a way that they co-ordinate and undermine the opponents pieces which temporarily or completely control the centre.
  
  Taking control of the Centre
  Place pieces to control the Centre part of the board from where they influence many squares on the board.
  
  Threatening Moves
  Advancing pieces and marching towards opponents camp either with the hidden threat or without any immediate purpose but in the hope that they co-ordinate and wait for him to change the position of his pieces so the hidden nature of the threat flourishes itself.
  
  Advancement of Castled King side pawn Cover:
  When to advance the king side pawn for an attack. Tests King safety.
  
  Prophylaxis
  Defence from obvious weakness before your opponent has even begun to attack.

[swami]

I've listed about 25 themes here:
http://sites.google.com/site/strategict ... al-ideas-1

There's still 24 more to go.

[swami]

Some more Ideas:

• Pawn Race: [Thanks, Jesper! for this]
  Advancing the pawns in late middle game transitions to endgame.
  
  King to the Centre:
  Bringing the King to the centre in the late middle game/early endgame.
  
  Castling:
  Late Castling either on the queen side or on king side as the form of defense in the Late openings/Early middlegame.
  
  Repositioning of the Pieces:
  When you find yourself in a situation where your attack halts and that your opponent defended well, your only resort is to reposition your pieces and attack again in a different way.

[swami]

Was going through some of the games, and these ideas occurred to me. Therefore, I wanted to update the list.

• Pawn Break
  Advancing the pawn in attempt to break open the file. Especially in fortress structures the pawn break helps in activating the play.
  
  Offer of Simplification
  This was released. See STS 7.0. Placing Pieces in direction of trade.
  
  Offer of Simplification with pawns
  Advancing the pawn in direction of the trade.

[Howard E]

Thanks for these test suites.
This might be a challenge but having positions that test the plus/minus factor in an unbalanced position. No best move required but does for example a program see that a well placed knight on f5
with king attack chances compensate for the exchange.

[swami]

Thanks for these test suites.
This might be a challenge but having positions that test the plus/minus factor in an unbalanced position. No best move required but does for example a program see that a well placed knight on f5
with king attack chances compensate for the exchange.

Thanks for your contribution. I'd like to know if there are any procedures to design the test suite that specifically tests the + or - scores in evaluation?

Your example sounds like something that's somewhat connected to the elements of "Massing near the King for the attack", "Knight Outposts", "Threatening Moves", "Bishop vs Knight" and "Positional Exchange Sacrifice"

[Howard E]

That's true, best if it is specific in nature. The plus/minus eval thought is too general.

[Howard E]

Is it possible to create a suite that takes the 20 hardest positions from each of the already released ones? One would need a databse of which positions were not solved by some of the better programs. Do you have such records?