From torbenm@sjofn.diku.dk Tue Jan 22 10:43:05 CST 2002 Article: 1414 of rec.games.abstract Path: news!husk.cso.niu.edu!vixen.cso.uiuc.edu!howland.erols.net!news.net.uni-c.dk!munin.diku.dk!not-for-mail From: torbenm@sjofn.diku.dk (Torben Ęgidius Mogensen) Newsgroups: rec.games.abstract Subject: Re: Reversi with holes? Date: 21 Jan 2002 10:16:14 +0100 Organization: Department of Computer Science, University of Copenhagen Lines: 77 Message-ID: References: <20020120102118.05567.00004590@mb-bh.aol.com> NNTP-Posting-Host: sjofn.diku.dk X-Trace: munin.diku.dk 1011604575 13740 130.225.96.195 (21 Jan 2002 09:16:15 GMT) X-Complaints-To: news@diku.dk NNTP-Posting-Date: Mon, 21 Jan 2002 09:16:15 +0000 (UTC) X-Newsreader: Gnus v5.7/Emacs 20.7 Xref: news rec.games.abstract:1414 mlhowe@aol.com1234321 (MLHowe) writes: > I noticed that on the "ItsYourTurn" website there's a game called > Black Hole Reversi where one square is blocked off as nonexistent > before the game begins. I wonder if it might not be more > interesting to give each player a fixed number of "holes", perhaps > one, two, or three each, to be played anywhere on the board as a > turn. Notice that holes would be immune to the "must flip" rule and > could be placed freely. Has anyone tried this, and if so, does it > make a good game? It might go a long way toward opening Reversi > theory back up -- from what I understand, the 8x8 game is becoming > too narrow at the highest levels. Does anyone want to try an > experimental game or two of "Hole Reversi" with me? I have tried something similar -- I once proposed an Othello variant which was played on a 10x10 board, but where 16 blocks were placed before game began with the limitation that no area of the board could be blocked off and no blocks could be played in the middle 4x4 squares. I only played a couple of games with this variant, but it seemed O.K. I agree that classic Othello is too much in the hands of the computer these days. As I see it, there are several reasons for this: 1) The fairly small number of legal moves per turn allows computer programs to make a deep search without using too much intelligence to limit the search space. This is especially true in the endgame, where some programs search the 12+ last moves perfectly. 2) The opening is quite limited, which allows a computer to use an opening library for the first many moves. 3) Since so much changes in a single move, it is hard for a human to think more than 4-6 moves ahead without losing track of some disks. A computer can easily keep perfect track of the baord for any number of turns ahead. The last point is an integral part of the game - it is about turning over a large number of disks. The first two points are two sides of the same problem - a small selection of legal moves. I listed them separately because they can be solved differently. Your suggestion, for example, may make opening libraries less useful as more opening are possible if you can place blocks (and the value of some openings will change depening on where blocks are placed). But placing blocks won't address the general problem of a small search space. Hence, I believe the game must be changed to increase the number of different legal moves. A simple way of achieving this is to increase the board size. It won't change the number of moves in the opening, but there will be more moves in the mid game. However, I don't think this will be nearly enough without going to very large boards (16x16 or bigger) and that will make games take far too long. I have earlier proposed another variant: Each disk has a cross printed on both sides. When a disk is placed, it can be oriented so the cross is like an x or like a +. To turn over a line of the opponents disks, the disk of your own colour at the other end must be oriented in the same way as the newly placed disk. The disks that are turned over can be oriented in any way that the player who does the turning wants. So, while this new rule may limit the number of spaces where you can play (due to the same-orientation restriction), it will increase the number of ways you resolve the move (due to the choice of orientation on placed and flipped pieces). The impact will be greatest on the end game, where you would typically flip a large number of disks each move. This will make perfect end-game search virtually impossible for more than half a dozen moves (though you might recognize that some orientations don't matter and thus decrease the number of searched positions). I haven't actually played this variant, but I analysed the number of legal moves in the start game and found that after a few moves you had many more possible positions that in the standard game. Torben Mogensen (torbenm@diku.dk)