Check out Janggi (Korean Chess), our featured variant for December, 2024.


[ Help | Earliest Comments | Latest Comments ]
[ List All Subjects of Discussion | Create New Subject of Discussion ]
[ List Earliest Comments Only For Pages | Games | Rated Pages | Rated Games | Subjects of Discussion ]

Comments/Ratings for a Single Item

EarliestEarlier Reverse Order LaterLatest
Ultima. Game where each type of piece has a different capturing ability. Also called Baroque. (8x8, Cells: 64) (Recognized!)[All Comments] [Add Comment or Rating]
Michael Nelson wrote on Fri, Mar 26, 2004 03:31 PM UTC:
A point I've never seen in the rules for Ultima or Rococo: can an Immobilizer immobilized by a Chameleon commit suicide? Logic suggests yes.

Roberto Lavieri wrote on Fri, Mar 26, 2004 06:27 PM UTC:
I don´t know if it is written somewhere, but I have seen that the practice answer in yes, an Immobilizer immobilized by Chameleon can commit suicide. This rule is also understood in Rococo and Maxima, and it is also implemented in this way in the Zillions versions of these games. Some people enjoys Ultima and some of its variants, as some people likes FIDE-Chess and some its variants. Why not?

John Lawson wrote on Sat, Mar 27, 2004 02:55 AM UTC:
I dug out my copy of 'Abbott's New Card Games' (1963, Funk and Wagnalls)
and the suicide rule is stated thus:
'A piece that is immobilized does have one special move that it can make,
that of suicide.  A player may use a turn to remove from the board one of
his own pieces that is immobilized.'

carlos carlos wrote on Tue, Mar 30, 2004 03:10 PM UTC:Excellent ★★★★★
i have two questions about rules i am not completely certain on.

1/ can a pincer pawn capture more than one piece (in one direction)?  e.g.
if there is a friendly piece on f4, and enemy pieces on f2 and f3, can a
pawn move to f1 and capture both?  i think this is unlikely, but i want to
check.

2/ can withdrawers capture by moving away on the diagonal from an enemy
piece?  e.g. enemy on g5, can a withdrawer capture it by moving from f4 to
e3?

thanks.

Antoine Fourrière wrote on Tue, Mar 30, 2004 03:30 PM UTC:
1. No, a Pincer Pawn can capture up to three pieces in different orthogonal
directions, but it cannot capture two pieces in a row.

2. Yes, a Withdrawer can capture by withdrawing diagonally, a Long Leaper
by jumping diagonally, an Immobilizer paralyzes diagonally and though a
Coordinator captures in an orthogonal way, it may be through a diagonal
move (say King on d1, Coordinator moving from h1 to b6 captures enemy
pieces at b1 and d6).

carlos carlos wrote on Tue, Mar 30, 2004 03:30 PM UTC:Excellent ★★★★★
thanks for the swift replies.

that is how i initially thought the withdrawers must operate.  but when
playing my game of ultima in the tournament, i got in a position to take
in this (diagonal) manner, but typed in my move and the piece remained on
the board.  so i changed my move.  until i saw ben's next move, i did
not
realise that i had to manually (as it were) remove the piece from the 
board myself with a separate command.  never mind!  thanks again for the
clarification.

Ben Good wrote on Wed, Mar 31, 2004 05:57 PM UTC:
carlos, the answers to your questions are: <P> 1) no <BR> 2) yes <P>

George Duke wrote on Thu, Apr 1, 2004 04:52 PM UTC:
Ultima design analysis:
# squares: 64
# piece types: 7
Piece-type density: 10.9%
Initial piece density: 50%
Power density: 84/64 = 1.31
Long diagonal: a1-h8
Est. piece values: P1, K2, W3, Co 3, Ca 4, L 5, I 8
Exchange Gradient: G = 0.505; (1-G)=0.495
Ave. Game Length: M = 3.5(Z)(T)/(P)(1-G) = (3.5*64*0.109)/(1.31*0.495) =
38 Moves
Features:  Unusual Pawns (pincer) may cohere with the chosen piece mix
Comments: Prosaic values across the board confound evaluation.

Dan Baisden wrote on Mon, Apr 5, 2004 06:07 AM UTC:Excellent ★★★★★
Is there anywhere to play the game online, other than against applets? I love the game, although when I used to play it the pawns moved one step orthoganally, could not be captured, and had still the (in that case) fairly unused ability to capture in a pincer fashion, one piece at a time. They mostly just got in the way. Civilian pedestrians, as it were.

Roberto Lavieri wrote on Mon, Apr 5, 2004 11:53 AM UTC:
You can play this game using the Game Courier System, inviting someone (open or direceted) or accepting invitations. The first thing you have to do is register as user. Click on 'PLAY!' (main page of 'What´s New'), Click then on 'Game Courier System' and follow clicking on 'Register', It is free!. About the way you play the game, it is not adjusted to official rules. Certainly, you are an atipical ULTIMA player, usually this kind people is particularily resistant to changes in the game, and it is played preferently with the original rules. One 'improvement' of the inventor was rejected by the fans, and new 'improvements' to the original game are not usually very welcome, aparently.

carlos carlos wrote on Tue, Apr 6, 2004 12:28 PM UTC:
a couple more questions.

1.  can a king take by co-ordinating with the co-ordinator, or does it
have to be after the co-ordinator's moves?
2.  i assume that a king cannot move into check, and you have to tell your
opponent if they do?  i was playing against the applet linked off this
page, and you have to actually capture the king rather than checkmate it. 
which is it?  

if white's co-ordinator is on f1, white's king on b1, and black's king
moves onto say b7 from the c file...  if the white co-ordinator has a
clear run to f7, is black's move therefore illegal?

Roberto Lavieri wrote on Tue, Apr 6, 2004 12:51 PM UTC:
Answers:
1.  can a king take by co-ordinating with the co-ordinator, or does it
have to be after the co-ordinator's moves?
No, King only captures by replacement. Co-ordinate capture is performed by
the Coordinator, or by a Chameleon to capture an enemy Coordinator. King
can´t do that.

2.  i assume that a king cannot move into check, and you have to tell
your
opponent if they do?  i was playing against the applet linked off this
page, and you have to actually capture the king rather than checkmate it.

which is it? 
Oficially, you can´t move into check. The object of the game is checkmate
the enemy King. Zillions implementation adopts the 'capture-the-King'
rule for technical reasons, the main reason is because if you immobilize
the King and you use the checkmate rule, it is interpreted by Zillions as
a Check, and it is not. 

3. if white's co-ordinator is on f1, white's king on b1, and black's
king
moves onto say b7 from the c file...  if the white co-ordinator has a
clear run to f7, is black's move therefore illegal?
Yes, this move is illegal, you are moving into check.

carlos carlos wrote on Tue, Apr 6, 2004 01:51 PM UTC:
thanks roberto, that clears everything up.

George Duke wrote on Wed, Aug 4, 2004 05:32 PM UTC:
In April 2004 comment this same page I estimate those Ultima values in
Design analysis: P1, K2, W3, Co3, Ca4, L5, I8, keeping integers. I
haven't done C++ programming this decade, so don't know whether King's
offensive value would be needed, depending on structure of program. 
Refining these estimates, maybe L closer to 5.5, and W>Co. [Further, I would 
enter P1, W3.1, Co2.9, Ca4.3, L5.3, I8.2, and see how that plays]

Roberto Lavieri wrote on Wed, Aug 4, 2004 08:26 PM UTC:
Those values are close to the subjective values I have stimated, although evaluation of positions in Ultima is more complex. Immobilized pieces values are significatively reduced, but it depends also on the vulnerability of the Immobilizer that is freezing the pieces, and in the fact that the Immobilizer is also immobilized or not, or it CAN be threatened to be immobilized. Pince-Pawn values strongly depends on the whole set of the own Pincers and other pieces, if there are more squares where Pincers can take enemies after hypothetic moves, higher is the Pincer value, and in some situations the value of a Pincer can be as high as the value of some major pieces, if the mobility of enemy pieces is reduced enough. Ultima tendence is to be a very defensive game, so this must be considered in the evaluation of positions. Long-Leaper value is high if the position is sparse, without many clusters of pieces, but it is low if there are massive clusters and/or the enemy pieces are positioned on squares of the edges, usually Long-Leaper value is high when Pincer values are high. Coordinator value depend strongly on the position of the King, if there are many enemy pieces on the orthogonals of the King, the Coordinator value increases a lot, regardless the imminence of a capture. Chameleon value increases if the Chameleon can attack positions where enemy pieces can go if that can hurt a lot after the hypothetic move, also if Chameleon can Immobilize the Immobilizer, and this value can be increase or decrease depending on the vulnerability of Chameleon. The Chameleon value is high if he can threaten the enemy King safely. This is a game not very easy for position evaluations, but I can suggest that good defensive moves must add something, and risky moves that allows counter-attacks or augment the enemy mobility must substract points.

Anonymous wrote on Wed, Aug 4, 2004 08:56 PM UTC:
I doubt whether all those contingencies of position will be worked into a program, unless someone wants to buy another Deep Blue. Instead, it just goes 3-, 4-,5-ply, whatever, based on set of values, as I remember games programs. Those values are somewhat compromises and estimates, also for what move it is, 10th, 20th, 30th. In Falcon Chess, Falcon starts at 7 points, and falls below Rook's 5 by the time there are fewer than 10 pieces/pawns. It is important to normalize, if possible, Pawn to 1 for a values table, like in my Design Analyses, even if, like a Cannon Pawn, it is 1.5, or 1.8, if actually matched with standard P; all values are relative.

Greg Strong wrote on Thu, Aug 5, 2004 12:14 AM UTC:
<p>Thank you. This information is most helpful!</p> <p>The previous anonymous poster does make a point about sliding piece values, and ChessV already supports this. In Orthodox Chess, for example, I evaluate the Rook starting with the Speilmann value of 4500 (relative to a 1000-point pawn.) The value then is scaled up by 50 points for each capture of any piece, starting with the 10th. So, when the board is down to only 12 pieces, for example, the rook is then evaluated at 5000. I don't know how much this applies to Ultima, but since the previous poster brought it up, I thought I would mention it.</p>

Roberto Lavieri wrote on Thu, Aug 5, 2004 01:44 AM UTC:
I have mentioned some basic conditions that can add or substract points in evaluation of positions, and sliding values of pieces based on positions and conditions may be important in this game, if you want a very strong program. How can slide these values up or down?. You can begin using gross estimators of adds and substracts based on the stablished conditions/positions, and then the values can be tunned by essay-and-error, running a few game tests (I´ll be happy offering part of my time doing it). Deep Blue?, I don´t think so, but it is possible the construction of a REALLY strong Ultima player, expert level, if CHESS-V supports sliding values of pieces and global evaluations based on positions/conditions. How good can be?. At least, it should be easy the construction of a program that can beat Zillions without difficulty (Zillions plays some variants superb, -if you don´t think so, try one of my games, DENEB-, and others very poorly, if you don´t think so, make the attempt of implementing Amazons... Zillions ULTIMA or ROCOCO level of play is not that of a novice, but I can say is not very strong playing these games, although it plays MAXIMA and FUGUE clearly better, in my game MAXIMA the piece values were artificially adjusted, but I´m not sure about all the reasons)

George Duke wrote on Thu, Aug 5, 2004 02:21 AM UTC:
Of course I posted anonymously about sliding values based on things objective, i.e. especially Captures and Number of Moves, just standard programming. More mathematically, to go with Move Equation under Game Design topic, I also developed a Positional Advantage Equation. It measures a game from its rules for positional advantage potential. [Achernar], being Orthodox Chess with interesting but bizarre rules overlaid, would expect any basic program to play well.[I switched Achernar and Deneb, sorry]

Roberto Lavieri wrote on Thu, Aug 5, 2004 02:39 AM UTC:
One of the problems with the game play level of some Zillions files is just that you don´t have tools for adjust evaluations, adding or substracting points based on positions/ conditions, probably, the program does that, but it is a closed evaluation made by a general-purpose game program, and it is a black box for the programmer. Zillions plays some games better than others, and it is not easy seeing why. By example, Zillions plays Grand-Chess clearly better than Eurasian, and Shogi original ZRF plays like a novice (Fergus improved a lot the Shogi game play using Zillions, but he have had to tune some values and drops, and it was not easy). I expect the next version of Zillions can add new tools to help the programmer work, it is a need for some games implementation, like Amazons and others. The idea is not a Deep Blue, I don´t like extremely strong computer opponents because it can cause some frustration many times, but if you want play a particular game, at least you expect it can offer a good challange. ChessV is fine, it plays really well the most of the variants, but it is not unbeatable. For me, it is an excellent challenger, regardless I have to lose not so few games, I can draw sometimes and once in a while I can beat it!.

Roberto Lavieri wrote on Thu, Aug 5, 2004 02:55 AM UTC:
Why Zillions plays well Deneb?. I´m not sure, but I think one of the reasons is because in the ends are many 'royal' pieces, and the program can visualize in a good manner complex tactics in which more than one 'royal' piece is attacked after some moves, but the characteristics of the game may also help. No, I don´t think so any basic program can play this game well, Zillions A.I. seems to be, coincidentially, well adapted to this game.

Roberto Lavieri wrote on Thu, Aug 5, 2004 03:26 AM UTC:
Achernar?. Yes, George, I coincide with you, any basic program plays this game as bad as me or any other not very experienced player, this is a very complex game in which I have medited in the last times, it is nice, but I´m convinced it needs some important reforms, because I don´t want this game for people that have to spend years training with it, I want the game for happy people that wants have fun once in a while trying it. I´ll work on it in the next future.

Greg Strong wrote on Sat, Aug 7, 2004 07:25 PM UTC:
<p>Dear Mr. Duke,</p> <p>In your recent comments, you mention a Positional Advantage Equation, the details of which may be found under the Game Design topic. I am interested in <i>anything</i> related to mathetmatical analysis of positions, but I cannot find this Game Design forum...</p> <p>Sincerely,<br> Greg Strong</p>

John Lawson wrote on Sun, Aug 8, 2004 11:13 PM UTC:
Try this link.  It was a little awkward to find:
http://www.chessvariants.com/index/listcomments.php?subjectid=Game+Design

George Duke wrote on Mon, Aug 9, 2004 04:52 PM UTC:
I have not published Positional Advantage Equation, taking Mark Thompson's advice to write an article. Move Equation is M = 3.5N/P(1-G). I don't know where they index special topics; you see it scrolling back any of the talkers.

25 comments displayed

EarliestEarlier Reverse Order LaterLatest

Permalink to the exact comments currently displayed.