Check out Symmetric Chess, our featured variant for March, 2024.


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

Comments/Ratings for a Single Item

Later Reverse Order Earlier
Arimaa. Board game playable with standard chess set, hard for computers. (8x8, Cells: 64) [All Comments] [Add Comment or Rating]
Kevin Pacey wrote on Tue, Sep 20, 2016 11:28 PM UTC:Good ★★★★

Like Bombalot, Arimaa is a variant that's very unlike chess, in that there are no matable kings, but it can make use of standard chess equipment. Arimaa might have spread over-the-board better were it not for the licensing requirements for e.g. literature or running tournaments. It's perhaps too bad world championship level human Arimaa players finally lost in a challenge series vs. a bot (engine) in 2015.


George Duke wrote on Wed, Feb 12, 2014 06:50 PM UTC:
Apply the four-move Mutator to Melee, a random CV not in the Multi-move list, http://www.chessvariants.org/large.dir/melee.html, as one example which has mostly short-range pieces. There should be 5000 or 20000 alternatives per ply, in the same range of choices as arimaa.  In Melee for this purpose, special design must limit power of the three long-range pieces, or substitute mid-rangers having the same radials.  Would three or four movements per turn, "elevate complexity of [the CV] to a higher level without affecting the theoretical size of the game tree?" Or is it necessary always to add condition other than extra move per turn, to get complexification?

That quotation above, http://en.chessbase.com/post/a-new-challenging-chess-variant, is wording at the Chess Base article on Switch Side Chain -- the Chain there is excellent new Mutator in spirit of Variantism not exactly seen before (although here was another Chain worth re-looking at as a Win Condition: http://www.chessvariants.org/other.dir/fools.html.)

George Duke wrote on Wed, Feb 12, 2014 06:50 PM UTC:
http://www.chessvariants.org/index/mainquery.php?type=Any&category=Usual-MultiMove&orderby=LinkText&displayauthor=1&displayinventor=1&usethisheading=Multiple+Move+Variants.  Daniil Frolov has the right idea, and his referencing 007 Chess actually has to do with Arimaa.  That is, what gives arimaa its 15,000 or so average alternative moves per turn comes from the plural moving each turn.  Just find suitable multi-move CVs from  above index, and those CVs arguably present the same difficulty for Computer as arimaa; research can then find which ones actually do. Approximation of how many moves are available in some sample CVs from  multi-move category is instructive.  Argument can also develop two other ways: (1) use CVs only from Progressive sub-category; (2)  Apply Mutator of requiring up to four moves per turn on any CVs having mostly short-range piece-types, as Frolov is implying in mention at 007 Chess.

George Duke wrote on Mon, Feb 10, 2014 09:00 PM UTC:
Patent proceedings are supposed to be conducted with the utmost decorum and integrity in any country.  http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.htm&r=1&f=G&l=50&s1=6981700.PN.&OS=PN/6981700&RS=PN/6981700.  Arimaa does not disclose any material published in Chess Variant Page, yet appears to have drawn on ideas from other CV inventors [each individual responsible for own work]. Moreover, some referenced patent documents would seem to be irrelevant. It's reasonable over-all to patent this usp6981700 as unique, but the novelty is not quite honestly disclosed in the description.  Since Arimaa by Variantists' high standards is subpar CV and game anyway, this critique will be intermittently developed.  Some AI articles constantly kneejerk referencing Arimaa as landmark drone without rigour, and their authors unschooled in CVs should be embarrassed.  

http://en.wikipedia.org/wiki/Arimaa.

Like as just one example, Carrera-Bird-Capablanca-Seirawan BN and RN compounds are periodically returned to as topic, to the extent Arimaa continues to be overplayed in Chess press undeservedly, the critiques will revive and deepen.

George Duke wrote on Sun, Feb 9, 2014 09:59 PM UTC:
Warp Point, http://www.chessvariants.org/diffmove.dir/warp-point-chess.html, was invented in connection with CVPage April 2002.  This CVPage article says Arimaa was invented around 2002.  Arimaa USA Patent has filing date October 3, 2003 and priority date October 3, 2002: http://www.patentbuddy.com/Patent/6981700.  That would indicate Arimaa is invented fall or winter 2002-2003.  Though not unprecedented Warp Points are unusual in CVs, and they do have functionality related to later Arimaa Traps.  Unfortunately Omar Syed has never spoken here about Chessic influences and input, but arimaa sites tell of distress over 1996 Deep Blue-Kasparov match prompting new design.

http://www.chessvariants.org/large.dir/contest84/transporterchess.html -- Transporter squares, four of them, were November 2002 in another CV.

In 1996 Transporters, http://www.chessvariants.org/dpieces.dir/transport.html, one per side, move one step, the same as all the pieces in arimaa.  Follow-up will look at prior art the patent itself cites.  Also to locate are CVs where multiple movements, two or more, in one turn get distributed over two or more p-ts.

George Duke wrote on Sat, Feb 8, 2014 04:41 PM UTC:Poor ★
Draw-less Arimaa unacceptably represents that our beloved OrthoChess on the same 64 is unsalvageable.  Arimaa foregoes Chessic moves except lone Wazir one-step.  Arimaa six piece-types are really one piece-type, since they all move that same Wazir-like one step at a time cardinally.  Well, 1.5 p-ts since Rabbit cannot step backwards (no nibbling).  Call 21st-century 
Arimaa the recidivist Game of Wazirs, http://www.chessvariants.org/piececlopedia.dir/wazir.html (befitting certain times?).

It would understandably appeal to small minority of programmers themselves for its strict queued prioritization, Elephant > Camel > Horse > Dog > Cat > Rabbit, to keep track of.   Which pieces so childlike 
Beast-named seem to speak comfortably for the entire 20 million  threatened species outdoors in the real world.

From design standpoint, actually drab as it is, there would be umpteen subvariants of Arimaa, as many as one likes, 1000, 10^4 you name it.  There is nothing compelling about the particular Rules for freezing (immobilizing), pushing or pulling, or serial p/p.  There is nothing sacrosanct about where the four Traps are placed, their variability in number and location making thousands of A-subGames.  Here's a mere one subvariant: Mortal Arimaa lets captured pieces entering or in Traps be held for later drop.  Here's another: Rabbit Roadkill Arimaa requires Rabbit forward one-step to dis-lodge any Elephant only there.  That specialized move, whether 1st or 2nd or 3rd or 4th in a turn, is to be called Hare-Alfing and is always a Push, but any number of squares not just one along legal available orthogonal.

Arimaa so many Rules is inelegant.  There are 8 or 9 Mutators overlaid; any awkward designer can similarly arbitrarily complexify in mediocrity; that enlarges the game tree surely in giant uglification.  Try programming in a month the hugest war-game, http://www.chessvariants.org/index/msdisplay.php?itemid=MSwaroftheroses , and then try playing it skillfully.  Or Arimaa is like a different Lewis Carroll syllogism,   http://www.math.hawaii.edu/~hile/math100/logice.htm, every turn just to define which moves are legal per go.  Or series of tongue-twisters: He Smells She Tells Sally Sells Sea-Shells.

George Duke wrote on Tue, Feb 4, 2014 06:09 PM UTC:Poor ★
Poor insofar as it is a CV.  Well-promoted with award announced for successful AI program, Arimaa has no displacement capture as such, and everything is along orthogonals, no diagonals.  It's a shame Arimaa is habitually being mentioned in the same breath as great oriental classic Go as hard for computers. (Games Magazine and the rest must be influenced by the "Beasts" for the piece-types) Arimaa has unaesthetcially too many win conditions available; it would be more tolerable on larger boards, for example the way Maxima has three possible win conditions.  There are far better race games those brilliant by Parton and Betza -- isolating the happenstantial Arimaa win route to get Rabbit to rank 8. Further, non-intuitive Arimaa (invented just before 2004)  won't improve natural Chessic thinking for design or for play.


Think of the whole business as a sort of pico-Rithmomachia phase http://www.chessvariants.org/misc.dir/rithmomachia.html, http://en.wikipedia.org/wiki/Rithmomachy. And well-conceived historic Rithmomachy actually has many a worthwhile variation to speculate and test, unlike the above listless distraction of gargantuan game tree.

(zzo38) A. Black wrote on Tue, Jan 8, 2013 03:21 AM UTC:

Should it be played on the intersections (like xiangqi or go), using flat pieces with roman numerals I to VI? And then mark the trap squares with stars or something, which is large enough to see even when covered by a piece. At least to me, this seem better than the chess board with animals on, due to the way this game works.

Is the name "Arimaa" made by "Aamir" spell backward?


George Duke wrote on Mon, Jan 7, 2013 08:41 PM UTC:
Arimaa was invented over ten years ago and this additional explanation of rules went up a year ago. There are six piece-types like OrthoChess but no capturing as such, that is in cv terms no displacement capture. But pieces pushed onto a trap exit, if not protected, as if captured. If losing all Rabbits that way, the opponent side wins; and Arimaa is a cv with several alternate winning conditions built into official rules, Checkmate not among them. The main objective to win is to get any one of eight Rabbits to Rank 8.

9 comments displayed

Later Reverse Order Earlier

Permalink to the exact comments currently displayed.