From: leung kaiwan gmail com> Date: 4 mar 2007 Subject: Re: A Comparison of Shogi, Chess, Xiangqi ------=_Part_27791_19971398.1173001883893 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline Hi, Nice to see the ideas of other shogi friends. My two cents are as below. Regards, Kai Wan Leung In terms of pieces valuations, I would like to add another feature about shogi. In both xiangqi and chess, pieces are always on the board until being removed. The pieces which control more squares are more valuable. And those in the centre or close to the kings are more valuable than the ones in outlying areas. In shogi, however, pieces may be on board or in hand. As most shogi pieces move in short distance, they move slowly and control nearby squares mainly. Pieces in hand, however, may drop to any vacant square (subject to a few restrictions like nifu), but do not control any square immediately. A piece in hand may drop onto a decisive position and controls many critical squares in the future. But in immediate term, it does not affect the board. So the values of pieces in hand are far more volatile than on board. And their aggregation for evaluating the overall strength of a player's position tends to be much more vulnerable. When both players exchange the same pieces, e.g. rook, each side converts the piece from being on board to being in hand, which may arrive on the board again anytime anywhere. This creates a wormhole against the conventional space time (position & number of moves to be taken) wisdom in chess position valuation. And this leads to a trade-off between movement efficiency and control of positions as a piece in hand may go anywhere in 1 move but immediately controls 0 square. Only in shogi, your pieces may belong to your opponent. Player A's pieces on board may become player B's pieces in hand, then B's pieces on board, and then A's pieces in hand, and eventually A's pieces on board again. So you have to think which pieces will remain in use and which will be exchanged or sacrificed. For example, if you promote a pawn to tokin to attack, probably you have prepared to exchange it for another piece. But perhaps you won't be as willing and confident to exchange a rook for another piece. Hence, there can be two values for a shogi piece, namely use value and exchange value. When we exchange pieces in chess and xiangqi, most of the time we simplify positions. But in shogi, we add complications. Applying my earlier idea that the outcome of any move may be likely to win, likely to lose, or uncertain, moving pieces from board to hand adds uncertainty and moving pieces from hand to board reduces uncertainty. Exchanging pieces not only alter the total value of your pieces and your opponent's. That also affects your possibility and certainty to win. I am not a computer programmer and don't know how the above may or may not affect making shogi AI. But there seems to be some additional considerations to quantitatively evaluate shogi positions. There is a Japanese term "sabaki" which is about efficiency of pieces movement but I don't know how to say about it. May anyone have some idea please feel free to share. Concerning zugzwang, first of all I 'd like to thank Doug for introducing this vocabulary (afraid to say that I don't know this word before). If both you and your opponent make a move, and your move is more useful than your opponent's, then you get a better position. Even when your move does not add value to your own position, if that effectively eliminate any good move from your opponent, the great minus for your opponent is indeed a big plus for you - in this zero sum game. I think my game with Mr. Takahashi in last year's WOSC is a good illustration. This is a xiangqi-like shogi game and is available at http://www.kaiwan.hk/shogi/wosc/06wosc6.htm . In move 39 (Japanese counting) I played B*a8 to protect c6 and Mr. Takahashi's previous bishop drop suddenly became a burden as it might become mine. The P-9f in move 73 is a waiting move that helps me survive. The P*6h in move 115 also effectively forestalled any future attack. In this game I had been trying to avoid unnecessary exchange in order to limit Mr. Takahashi's pieces in hand. Sometimes it is an amazing strategy against Japanese players. Without exchange, capture, and drop, a shogi game is proceeded with moves only, like chess and xiangqi. (If chess players only move pieces well but don't drop pieces as skillfully as native shogi players, what would happen if a shogi game is played without drop? This is hypothetical and totally out of topic here, but may be very interesting.) The essence of this technique is not to avoid fighting, but to control the time to fight, by manipulating the level of complications in terms of how many pieces are in hand. Many shogi and chess players take complications for granted, but we often limit variations of moves and complications in xiangqi. Similar strategy was applied in my game with Mr. von Oosten. In this game I suffered from material loss but maintained a stronger castle. So I tried to dig into a small loophole in the edge before Mr. von Oosten started a large-scale fighting. URL: http://www.kaiwan.hk/shogi/wosc/06wosc8.htm Though not intentionally, in my game with Mr. Cheymol, the early rook-bishop exchange transformed my idle rook on board to the later powerful bishop drop. This exchange and the difference in kings' positions decided the game. URL: http://www.kaiwan.hk/shogi/wosc/06wosc9.htm The above are my own examples of playing unaggressive shogi. These may not be examples of zugzwang but the question "when not to fight" can give you an interesting idea of how to play a game. If zugzwang is defined as forcing your opponent make a move with negative value, my idea is that you may minimize the value of your opponent's move, maybe from positive to nearly zero, sometimes negative. Meanwhile, the most well known related example is the bishop exchange with 1-move loss opening. By taking an extra move, P-2e is avoided to allow subsequent N-2e. Instead of forcing your opponent to make a bad move, you avoid a bad move yourself by spending it anyway. In Japanese, a game is also known as teai. The term literally means hand-interpose, relates to the concept that the hands or the moves of the two players intersect or intervene each other. The effect of each move may be constructive or destructive, to yourself or to your opponent. In Chinese, games like xiangqi and go (weiqi) are also known as hand-talk. This means that people do not speak, but communicate with hand, by making moves to show their thinking and intention to the others. On 02/03/07, Larry Kaufman comcast net> wrote: > Piece values in shogi make for an interesting topic. Of course there > is some fluctuation during the game, but the most fundamental rule is this: > a piece in hand is always assumed to have full value (except for many pawns, > due to the double pawn rule), but a piece on the board is devalued if it is > not able to attack the enemy king or defend its own king. Thus, if a general > is on your far left while both kings are on your far right, its value is > well below the value of generals in the castle. For rooks and bishops, this > is not much of an issue as they can quickly swing from one part of the board > to another. > For pieces in hand or in a castle, the table of values is pretty > reliable. That's why Bonanza is so darn strong! > > Larry Kaufman > > ----- Original Message ----- > *From:* sean humby gmail com> > *To:* shogi topica com > *Sent:* Wednesday, February 28, 2007 8:51 AM > *Subject:* Re: A Comparison of Shogi, Chess, Xiangqi > > > > *It is a testament to how dynamic shogi is that there is no "point scale" > for the pieces.* > > This is not an accurate statement. There are several scales given in > various books by famous pros. For lance, knight, silver, gold, bishop, and > rook (unpromoted), with pawn as 1, Tanigawa gives 5-6-8-9-13-15, while other > pros give 3-4-5-6-8-9 (or 10), with other values for promoted pieces. I > personally teach 4-5-7-8-11-13, which is a compromise between the above that > I think works quite well. > What is accurate is that there is much less agreement on the proper > scale than there is in chess, and also that it is far more useful in the > early stages than later on, when king safety becomes critical. > > This is an interesting piece of wisdom. I had thought general shogi wisdom > was to encourage more fluid thought than assigning static values to pieces. > Teaching that a pieces intrinsic value is dependant on a myriad of factors > including game stage, king safety, attack potential, and the opponent's > possible use for the piece. The values you've listed from various pros seem > like they can and should fluctuate. For example I've learned that in the > opening a kaku can have a higher value than a hisha for it's ability to > weave through the pawn lines. In practice I rarely make that trade but I can > certainly see the logic behind dropping a bishop in a sore spot in the > enemies defense. > > Not trying to argue... rather to get some clarity. Is there any more > detail regarding these value-systems and how to effectively apply them to > any situation? > > --^^--------------------------------------------------------------- > This email was sent to: lkaiwan gmail com > > EASY UNSUBSCRIBE click here: http://topica.com/u/?a2i6Ys.bsEusY.bGthaXdh > Or send an email to: shogi-unsubscribe topica com > > For Topica's complete suite of email marketing solutions visit:http://www.topica.com/?p=TEXFOOTER > --^^--------------------------------------------------------------- > > ------=_Part_27791_19971398.1173001883893 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline

Hi,

Nice to see the ideas of other shogi friends.  My two cents ar= e as below.

Regards,

Kai Wan Leung

 

In terms of pieces valuations, I would like to add another feature about= shogi.  In both xiangqi and chess, pieces are always on the board unt= il being removed.  The pieces which control more squares are more valu= able.  And those in the centre or close to the kings are more valuable= than the ones in outlying areas.  In shogi, however, pieces may be on= board or in hand.  As most shogi pieces move in short distance, they = move slowly and control nearby squares mainly.  Pieces in hand, howeve= r, may drop to any vacant square (subject to a few restrictions like nifu),= but do not control any square immediately.  A piece in hand may drop = onto a decisive position and controls many critical squares in the future.&= nbsp; But in immediate term, it does not affect the board.  So the val= ues of pieces in hand are far more volatile than on board.  And their = aggregation for evaluating the overall strength of a player's position = tends to be much more vulnerable.=20

When both players exchange the same pieces, e.g. rook, each side convert= s the piece from being on board to being in hand, which may arrive on the b= oard again anytime anywhere.  This creates a wormhole against the conv= entional space time (position & number of moves to be taken) wisdom in = chess position valuation.  And this leads to a trade-off between movem= ent efficiency and control of positions as a piece in hand may go anywhere = in 1 move but immediately controls 0 square.=20

Only in shogi, your pieces may belong to your opponent.  Player A&#= 39;s pieces on board may become player B's pieces in hand, then B's= pieces on board, and then A's pieces in hand, and eventually A's p= ieces on board again.  So you have to think which pieces will remain i= n use and which will be exchanged or sacrificed.  For example, if you = promote a pawn to tokin to attack, probably you have prepared to exchange i= t for another piece.  But perhaps you won't be as willing and conf= ident to exchange a rook for another piece.  Hence, there can be two v= alues for a shogi piece, namely use value and exchange value.=20

When we exchange pieces in chess and xiangqi, most of the time we simpli= fy positions.  But in shogi, we add complications.  Applying my e= arlier idea that the outcome of any move may be likely to win, likely to lo= se, or uncertain, moving pieces from board to hand adds uncertainty and mov= ing pieces from hand to board reduces uncertainty.  Exchanging pieces = not only alter the total value of your pieces and your opponent's. = ; That also affects your possibility and certainty to win.=20

I am not a computer programmer and don't know how the above may or m= ay not affect making shogi AI.  But there seems to be some additional = considerations to quantitatively evaluate shogi positions.  There is a= Japanese term "sabaki" which is about efficiency of pieces movem= ent but I don't know how to say about it.  May anyone have some id= ea please feel free to share.=20

Concerning zugzwang, first of all I 'd like to thank Doug for introd= ucing this vocabulary (afraid to say that I don't know this word before= ).  If both you and your opponent make a move, and your move is more u= seful than your opponent's, then you get a better position.  Even = when your move does not add value to your own position, if that effectively= eliminate any good move from your opponent, the great minus for your oppon= ent is indeed a big plus for you - in this zero sum game.=20

I think my game with Mr. Takahashi in last year's WOSC is a good ill= ustration.  This is a xiangqi-like shogi game and is available at http://www.kaiwan.hk/sh= ogi/wosc/06wosc6.htm .
In move 39 (Japanese counting) I played B*a8 to protect c6 and Mr= . Takahashi's previous bishop drop suddenly became a burden as it might= become mine.  The P-9f in move 73 is a waiting move that helps me sur= vive.  The P*6h in move 115 also effectively forestalled any future at= tack.  In this game I had been trying to avoid unnecessary exchange in= order to limit Mr. Takahashi's pieces in hand.  Sometimes it is a= n amazing strategy against Japanese players.  Without exchange, captur= e, and drop, a shogi game is proceeded with moves only, like chess and xian= gqi.  (If chess players only move pieces well but don't drop piece= s as skillfully as native shogi players, what would happen if a shogi game = is played without drop?  This is hypothetical and totally out of topic= here, but may be very interesting.)  The essence of this technique is= not to avoid fighting, but to control the time to fight, by manipulating t= he level of complications in terms of how many pieces are in hand.  Ma= ny shogi and chess players take complications for granted, but we often lim= it variations of moves and complications in xiangqi.

Similar strategy was applied in my game with Mr. von Oosten.  In th= is game I suffered from material loss but maintained a stronger castle.&nbs= p; So I tried to dig into a small loophole in the edge before Mr. von Ooste= n started a large-scale fighting. =20
URL: http://www= .kaiwan.hk/shogi/wosc/06wosc8.htm

Though not intentionally, in my game with Mr. Cheymol, the early rook-bi= shop exchange transformed my idle rook on board to the later powerful bisho= p drop.  This exchange and the difference in kings' positions deci= ded the game.=20
URL: http://www= .kaiwan.hk/shogi/wosc/06wosc9.htm

The above are my own examples of playing unaggressive shogi.  These= may not be examples of zugzwang but the question "when not to fight&q= uot; can give you an interesting idea of how to play a game.  If zugzw= ang is defined as forcing your opponent make a move with negative value, my= idea is that you may minimize the value of your opponent's move, maybe= from positive to nearly zero, sometimes negative.

Meanwhile, the most well known related example is the bishop exchange wi= th 1-move loss opening.  By taking an extra move, P-2e is avoided to a= llow subsequent N-2e.  Instead of forcing your opponent to make a bad = move, you avoid a bad move yourself by spending it anyway.  In Japanes= e, a game is also known as teai.  The term literally means hand-interp= ose, relates to the concept that the hands or the moves of the two players = intersect or intervene each other.  The effect of each move may be con= structive or destructive, to yourself or to your opponent.  In Chinese= , games like xiangqi and go (weiqi) are also known as hand-talk.  This= means that people do not speak, but communicate with hand, by making moves= to show their thinking and intention to the others.=20

 

On 02/03/07, L= arry Kaufman < comcast net" target=3D"_blank">lkaufman@comca= st.net > wrote:

     Piece values = in shogi make for an interesting topic. Of course there is some fluctuation= during the game, but the most fundamental rule is this: a piece in hand is= always assumed to have full value (except for many pawns, due to the doubl= e pawn rule), but a piece on the board is devalued if it is not able to att= ack the enemy king or defend its own king. Thus, if a general is on your fa= r left while both kings are on your far right, its value is well below the = value of generals in the castle. For rooks and bishops, this is not mu= ch of an issue as they can quickly swing from one part of the board to= another.=20
     For pieces in= hand or in a castle, the table of values is pretty reliable. That's wh= y Bonanza is so darn strong!
 
     Larry Kaufman=  
----- Original Message -----
From: gmail com" onclick=3D"return top.js.OpenExtLink(window,event,thi= s)" href=3D"mailto:shumby gmail com" target=3D"_blank">sean humby
Sent: Wednesday, February 28, 2007 8= :51 AM
Subject: Re: A Comparison of Shogi, = Chess, Xiangqi

 

It is a = testament to how dynamic shogi is that there is no "point scale" = for the pieces.

     This is not an accurate statement. T= here are several scales given in various books by famous pros. For lance, k= night, silver, gold, bishop, and rook (unpromoted), with pawn as 1, Ta= nigawa gives 5-6-8-9-13-15, while other pros give 3-4-5-6-8-9 (or 10), with= other values for promoted pieces. I personally teach 4-5-7-8-11-13, which = is a compromise between the above that I think works quite well.=20

     What is accurate is that the= re is much less agreement on the proper scale than there is in chess, and a= lso that it is far more useful in the early stages than later on, when king= safety becomes critical.=20
 
This is an interesting piece of wisdom. I had thought= general shogi wisdom was to encourage more fluid thought than assigning st= atic values to pieces. Teaching that a pieces intrinsic value is dependant = on a myriad of factors including game stage, king safety, attack potential,= and the opponent's possible use for the piece. The values you've l= isted from various pros seem like they can and should fluctuate. For exampl= e I've learned that in the opening a kaku can have a higher value than = a hisha for it's ability to weave through the pawn lines. In practice I= rarely make that trade but I can certainly see the logic behind dropping a= bishop in a sore spot in the enemies defense.=20
 
Not trying to argue... rather to get some clarity. Is= there any more detail regarding these value-systems and how to effectively= apply them to any situation?

------=_Part_27791_19971398.1173001883893--