#101 · FatPhil
18 Dec 2008
What else do you connect with edges?
If you can't see a cycle in the undirected connections between cells, then words fail me.
Public forum discussion.
18 Dec 2008
What else do you connect with edges?
If you can't see a cycle in the undirected connections between cells, then words fail me.
18 Dec 2008
Phil, cells have edges. Cells alos have interiors. A figure is not merely its exterior.
Would you argue that a cube is a loop all by itself? If not, then four cub es resting on the same plane and that share a vertex also do not topologically form a loop, even though one could make a cyclical path that traverses them. Because the four cubes are just as much a sinlge solid object as one cube by itself.
More to the point, if A touches B, B touches C, and C touches A, they form a loop unless they all touch at the same point.
In this game, a piece is not its surface, it is the it's interior as well as its exterior. One wins by forming a shape akin to a donut, where the game board itself is taken to complete the donut, if necessary.
18 Dec 2008
In Havannah, a triangle of three adjacent cells does not form a ring, right? Even though Phil's argument says that it forms a loop.
18 Dec 2008
You're weird. In no game I'm currently playing is the concept that cells have edges or interiors necessary. The cells are the indivisible atomic components from which a graph of connectionsis made, i.e. cell are the vertices. Why do you think they have to be two- or three-dimensional rather than 0-dimensional?
Given the context of a cube-stacking game, why on earth do you think I would be more likely to say that a single cube is a loop than I would say that a single hex grid is a loop?
4 cubes resting on the same plane and that share a vertex _do_ clearly form a loop. As you're currently in a plane you could immediately replace the cubes with 2-D tiles, and the position's loopiness would not change one little bit. So the dimensionality of the playing pieces is _irrelevant_. Cubes, tiles, hexes, are all just vertices.
Have you not noticed that the hex playing pieces themselves do not even ever touch? It matters not if they are balls, hexagonally-shaped, circles, or flat hexagons. They do not need to have a volume, a surface, or even any edges in order to be used as playing pieces. Single points would do just as well as any other shape in all these games. And single points are vertices. Which ties in nicely with connections being edges, as edges connect vertices.
You're chosing to view the playing pieces as continuous regions in a 3-manifold. I think that's an artifice based on an attachment to living and thinking in meat-space. You're getting all obsessed with both its interior and its surface, which is a hell of a lot considering that the difference between the piece's presence and absence is just one bit of information (plus a bit more to identify which colour it is). That's because at its core, the playing pieces are, like all the examples above, just vertices in a lattice. A computer playing the game would not need to model continuous 3D space in order to model the game, would it? It instead would model the lattice, the vertices in that lattice, and the connections between those vertices. And it would need to know that not all loops in the graph of connections are winning configurations. (What it would probably do is repeatedly reduce such subgraphs to new vertices until no more remain. If what remains is a loop, you've got a winning configuration.)
18 Dec 2008
wc - 3 mutually adjacent cells do form a ring. < {A, B, C} , {AB, BC, CA} > has a cycle, right? I can't believe someone with a graph-theoretical expertise such as yours could make such a blooper.
However, it's not a winning configuration in havannah. However, the rules of havannah have never, as far as I know, been an influence on the laws of mathematics.
A winning loop configuration in havannah seems to simply be one which contains a set of vertices V={v_i : i \in I} for some finite index set I and connections C={{v_i, v_j}: v_i, v_j \in V} which has a closed loop, and such that for every pair of distinct vertices v_i, v_j \in V, if {v_i, v_j} is a permissible connection (i.e. an adjacent cell in the board), then {v_i, v_j} \in C, and there are no distinct vertices v_i, v_j, v_k \in V such that {v_i, v_j}, {v_j, v_k}, {v_k, v_i} \in C.
Fortunately, surrounding yourself is equivalent to surrounding anything else, so the above can be formulated purely graph-theoretically (and not even graph theoretically, that's set theoretically, nothing more), and doesn't need the introduction of the topological concept of an 'inside' and an 'outside'.
18 Dec 2008
Phil: let me clarify! In the rules of Havannah the word "ring" is formally defined. What I meant with my post was just "3 mutually adjacent cells do not form a ring in the sense of the rules of Havannah". I'm not making any other assertion. In particular I am aware that there are other definitions of "ring" for which it might well be the case that 3 mutually adjacent cells do form a ring :-)
18 Dec 2008
Santorini, Chinese Chess and Shogi get my vote.
18 Dec 2008
A loop as defined for this game is the eye of a needle. Does that help? Or the bow (loop) of a shoelace? Or that thing (called a loop) that one can weave potholders out of, if one has about 30 of them and an appropriate small plastic loom.
As you say, seeing that objects have a physical presence comes naturally to those of us who live in meat-space. Since this game's rules are rather intuitive when one uses that natural view, I don't know why you're resisting it.
Oh, and in most Hex implementations I've seen, whole hexes are colored, so adjacent hexes of the same color do form a continuous region, except for the thin membranous borders whose color is distinct from unplayed hexes.
18 Dec 2008
"except"
snigger
18 Dec 2008
What singger? In JHex, the cells are incompletely filled. That's the true exception, as it obscures the contiguity of a filled region. In other implementations, there is only enough border to be able to read the grid. The cells themselves are completely filled. At least as I see it, the borders are no more than the grid lines on graph paper. That grid lines remain visible when colored pencils are used to fill a region on paper does not mean the region is broken into disjoint sub-regions.
So if JHex's implementation were standard, your graph theory visualization could be assumed to be what most players (of at least most interface implementors) found useful and/or intuitive. But since the hexes are completely filled, I suspect the simple topological conception is what most use, at least at first. Not that the graph-theory visualization is inferior; I suspect it's superior. But it is not as natural.
When trying to learn a new game, I suggest not being stuck in any one habit of visualization.
19 Dec 2008
In defense of FatPhil, i have to admit that teh rules of Lazo are a bit vague where the 3-D loop is described.
19 Dec 2008
A 3-D loop could perhaps better be terms an arch.
19 Dec 2008
Have you looked at the third page and the fourth page?
I welcome any feedback on how I might improve these pages, to make them easier to understand. Details would help. Please, what is vague?
19 Dec 2008
My votes to:
rock, paper and scissors
Crack the code
Perudo (bluff-dice game)
and perhaps classic board games such:
Stratego
Risk
Monopoly
22 Dec 2008
How about "Arimaa"? believed to be the second most complicated game after go:)
23 Dec 2008
I see Go, Shogi, Draughts (Checkers), Chess (and possibly Xiangqi) in the same complexity class (EXPTIME-complete) in that page.
Arimaa, Lines of Action, Chinese Checkers, Fanorona and Nine Men's Morris are in EXPTIME, which means they are a bit lower (although any of them can possibly be EXPTIME-complete, and we may have not found the proof yet).
23 Dec 2008
If you like havannah, you might want to trust me on dameo.
23 Dec 2008
@yper: Maybe this isn't the best measure of complexity though: checkers has been solved and AFAIK e.g. Chinese Checkers hasn't...
23 Dec 2008
True, maybe not. But have you seen the cpu-time used in solving the Checkers case?
23 Dec 2008
Speaking of checkers, one of the most enjoyable books I've ever read is "One Jump Ahead: Challenging Human Supremacy in Checkers" by Jonathan Schaeffer.
Despite not being a checkers player, I enjoyed it because it was well written, exciting, and it had a lot of chess and computer programming references. I can recall RE-reading it in between rounds of the last CHESS tournament I entered. (I'm sure I was the only one reading a checkers book at a chess tourney.)
7 Jan 2009
Irensei looks pretty complex and fun. Simple rules. I'd love to see Grand Chess, Epaminondas, Emergo, and Tori Shogi. Standard Shogi can go on forever.
8 Jan 2009
I propose Epaminondas. Robert Abbott's modern classic that deserves a wider number of players. http://www.logicmazes.com/games/epam/index.html
It is different, with simple rules and elegant deep and games takes about 30 moves though one can think that games can be very long because of the numer of pieces.
9 Jan 2009
A variation of Stratego, as already suggested, would also be interesting.
9 Jan 2009
may be antireversi??
20 Jan 2009
as posted twice in 2004: yinsh!
20 Jan 2009
Gesundheit!
30 Jan 2009
6x6 reversi - almost half the number of moves of 8x8, still challenging, and not many computer programs play it.
30 Jan 2009
But the programs who do.....
4 Feb 2009
that's true, and i wonder how a reversi variant would be where every move did not need to be a capture. being able to move to move to any open square would increase the complexity of the game so much that it would rival GO, all things being equal (ie board size)
5 Feb 2009
The first thing that comes into mind is that the players would play first at corners, where their stones would be secure. Perhaps you can modify your suggestion with the rule that at the corners (or more at the external lines), the only allowed moves would be the capturing moves.
Otherwise, the game seems worth a try at least. It would be quite different from Reversi and I have no idea how hard it would be for computers. Probably harder that Reversi though.
Oh, and you have to find a name for this variant. What about Wax-Reversi ?
5 Feb 2009
i like the name freeversi, because it's "free-form" compared to conventional reversi. i definitely don't want to name it after me, i don't own it, it's just a tweak of the existing rules. if it turned out i had any say in the matter, which i don't think is likely, i would want it to be in the public domain.
i had thought of the corners being taken first, but i dont think it's a problem. if the board began empty, the corners opening could function as a replacement for the standard center starting position. the game could even begin with the corners filled in alternating black and white, but leaving it empty to begin with would maximize complexity, and also leave room for experimentation with other possible openings, maybe on the sides.
5 Feb 2009
\* or, since the natural flow would go from corners to sides to center, "obversi."
5 Feb 2009
Excuse me, wouldn't the openings just revolve around building stable masses of discs from the corners outwards? It does not seem like an improvement to me.
It reminds me, though, a little bit of a game that I and a couple of friends developed many years ago: Pleiadis. It starts with an empty board with e.g. 12×12 squares. The two players take turns putting a piece of their colour on an empty square for which the neighbours (up to 8) do not contain more of the opponent's pieces than their own. The first player who can not make a legal move loses. The second player should get 4 points komi to even out the advantage of moving first.
I now think of this as a game that is not particularly good as it seems to lack depth.
5 Feb 2009
A good game is a mechanism, but feels like an organism. Why not have a shot at a mechanism in its purest form instead of tinkering around with one rule or another in known games. I think very few games offer so much with so little as Emergo, see our blog.
Like all strategy games it has a treshold- most likely you'll offer 'a kingdom for a plan' in the first few dozen games. There's no way around it.
But beyond it there's the beauty, the simplicity, the complexity, the purity of a quintessential game.
5 Feb 2009
Correct link to Littlegolem's Blog
5 Feb 2009
Thanks Ypercube :)
5 Feb 2009
\*\*\* Unsigned note pinned to Christians door with a dagger ***:
"We'll support your claim to implement Emergo if you agree to include Swap in Havannah"
;-)
5 Feb 2009
Well, I made the suggestion as much for the benefit to the players as for the game's future, if any,
And as for swapping, I'm still undecided and awaiting more reliable stats :)
5 Feb 2009
i haven't given up. a starting position of black and white stones around the corners and sides, i am almost certain, would work. effectively it would render 8x8 into 6x6, and 10x10 into 8x8 etc. a starting position around the outside is the obverse to starting in the center. obversi. if someone can show why this doesn't work, i'll shut up.
5 Feb 2009
a ctually, i do give up. the board looks ugly like that. nobody would want to play it. thanks though.
6 Feb 2009
although it does look better on a checkerboard ...
6 Feb 2009
game with ataxx-like capture rules (to capture, go requires covering 4 or more liberties, reversi and pente require 2, ataxx and this game require 1). game begins on empty board, and pieces may be placed in any open square. capturing moves are not required. any neighbouring adjacent (vert horiz diag) discs are flipped to your colour. this kind of capture solves the corner problem, corner pieces are capture-able. it also seems to be the case that taking a corner early will result in its being controlled by your opponent.
6 Feb 2009
Ataxx on a 10x10 or 9x9 board with randomly selected blockages and bidding to go first sounds very good to me.
7 Feb 2009
the variant of ataxx i mentioned would work well with those ideas and would be harder for computers to play than conventional ataxx, i think, due to the far greater number of possible moves at every turn. i especially like the random blockages, which ypercube also mentioned as an idea for reversi.
bidding to go first might only matter much on an odd-numbered board, where the first player has a slight advantage in terms of overall number of moves.
7 Feb 2009
Shogi is possible?
7 Feb 2009
i think it's possible... I don't know nothing about how probable is, though...
7 Feb 2009
Go only requires covering 2 liberties.
Reversi and pente don't have a meaningful concept of liberties. Please keep your unhelpful go terminology on the go board where it belongs.
8 Feb 2009
Why would you want bidding for Ataxx?
8 Feb 2009
I suspect we would come to learn what sorts of starting positions are good for the first player, and by about how much.
8 Feb 2009
That may be the case in many other games too.