Some Games (up to V1.4)

There are a lot of Reversi or Othello applets or CGI servers out there. I played most of them, and in the few cases where I thought I had found a strong program, I even let my applet play against them. On this page, you'll find a couple of games I or my applet played on the web.

To save bandwidth, I have opted for a textual representation of the game record, similar to chess: the colums on the board are numbered from A to H (left to right), the rows are numbered 1 to 8 (top-down). So, field A1 is the upper left corner, A8 the lower left corner, and H8 the lower right corner. Passing is indicated by "--".

Thomas Wolf vs. Web-Othello

Web-Othello is a CGI server for Othello. I played it on level "Adept" (seems to be a 5-ply look-ahead), with black. If you replay this game, you'll see that this program doesn't play well: I had an easy 56 - 8 victory. The program played a very bad edge fight on the 8th (bottom) row (its 10th move was already very bad, and after my 13th move, this edge was bound to become black sonner or later. To defend the edge, Web-Othello even had to play on the X-square G7 in the 20th move, giving me access to the H8 corner, which I got in the 25th move. Combined with the long download times for the images (your mileage may vary depending on where you are), this program isn't worth a visit.

Magoo vs. My Applet

The Java version of Magoo played on "Expert" level, but that didn't help much: it made a terrible mistake in move 13, which after another 3 moves lost the lower right corner. After that, the game was a clear 6 - 58 victory for my applet on "Amateur" level.

Jothello 3.1m3 vs. My Applet

Jothello 3.1m3 was the first and so far only other Reversi program I found on the web that could beat my applet once. Its evaluation function is similar to the one my applet uses. It also takes about the same time to compute the next move. Playing on level "difficult", it uncovered an error in my applet's algorithm and won easily. (The error made my applet make essentially the same bad mistake in an edge fight as Magoo did in the game above. Unfortunately (or luckily?), I forgot to keep a game record for that game, but my applet looked just as bad as Magoo did in the above game.) After I had corrected this error, a second game was played (again, my applet plays on level "Amateur"), and this time, Jothello didn't stand a chance: my applet won 13 - 51:

A rematch with my applet playing black (i.e. making the opening move) even resulted in a 53 - 11 victory (Jothello played on level "difficult", my applet - as usual - on level "Amateur" without timeout).

I haven't analyzed this game yet, but on a first glance it seems that Jothello prematurely gave up the H8 corner. Instead of playing 44. G8, it could have tried 44. D2...

One caveat: Jothello 3.1m3 is rather slow on level "difficult": it may take up to two minutes to compute its next move!

A Japanese Othello Server vs. My Applet

I played this server twice, but forgot to keep game records. The first game I played myself and won 38 - 26, the second game was played by my applet on "Amateur" level and won 17 - 47 on Sep. 24, 1996. Both games are in that server's score book.

My Applet vs. Yuh-Jye Chang's Othello Applet

In this game, my applet played on level "Beginner", no timeout. Yuh-Jye Chang's applet has only one level, so there was no choice there. The game ended by a clear 44 - 19 victory by my applet. My applet correctly sacrificed the H1 corner in move 42: in the following moves, it captured the whole rest of the H column, got the H8 corner, and proceeded to capture the bottom corner.

It seems that Yuh-Jye Chang's applet plays too greedily...

Thomas Wolf vs. Swanlake Othello

They say on their web page that they believe to have written one of the strongest Reversi applets on the web. Feh! Sorry to desillusion you guys, but it plays just lousy! Don't waste your time with it! I had no difficulty winning 61 - 0 on its "Expert" level:

I have the impression that the Swanlake Reversi program doesn't do any look-ahead, even on the "Expert" level. Why? Well, first, it's too fast - a Java applet can't do any look-ahead that fast, and second, it plays exactly the same moves on all levels. This smells fishy - all the more as they are trying to sell that applet's code and promise the buyer will learn something on the "Minimax method" and "alpha-beta pruning". May be, but I don't buy it! (Pun intentional)

Copyright © 2000 by Thomas Wolf. All rights reserved. TW, Aug 02, 2000