Debian Codegraph @

src/gamazons.html

raw
1 <html>
2
3 <head>
4 <title>Gamazons Help</title>
5 </head>
6
7 <body>
8
9 <h1>Gamazons</h1>
10
11 <a href="#rules">Rules of the Game</a><br>
12 <a href="#strategies">Strategies</a><br>
13 <a href="#buttons">What the Buttons Do</a><br>
14 <a href="#settings">Settings Menu</a><br>
15 <a href="#future">Future Enhancements</a><br>
16
17
18 <a NAME="rules"></a>
19 <h2>Rules of the Game</h2>
20 An amazon moves in a straight line in any direction, just like a chess queen.
21 When an amazon moves to a new spot, she's required to fire an "arrow", which
22 again travels in a straight line in any direction. The square the arrow lands
23 on becomes an impenetrable wall, which neither amazons nor arrows can move
24 over or land on. You also can't move or fire over other amazon pieces.
25 <p>
26 Game play alternates turns, starting with the white player. Each turn an
27 amazon moves and fires an arrow, thus each turn another square becomes unusable.
28 The winner is the side that moves last.
29
30 <a NAME="strategies"></a>
31 <h2>Strategies</h2>
32 One way to judge a good move or position is by determining the number of
33 squares you can move to vs the number your opponent can move to. You can
34 exert good control over your opponent by limiting his options. The bad news
35 is, the computer can calculate this a lot faster and more accurately than you
36 can. If this is your only strategy, you're going to get wupped.
37 <p>
38 I've had good success by starting out with a full out attack on one amazon and
39 then working on positioning the rest in good locations. Try not to put your
40 amazons too close to each other, try and move them to separate places so they
41 own different squares.
42 <p>
43 The computer AI is a pretty decent player. It has a decent board evaluator,
44 but since there are so many different options to evaluate, it can't look very
45 far ahead until late in the game. It also can't plan strategies, like, "Ok,
46 I'm going to try and claim this area for myself, so I'll block this guy here and
47 then block that guy over there..." You can. Use it to your advantage.
48
49 <a NAME="buttons"></a>
50 <h2>What the Buttons Do</h2>
51 <b>UNDO</b>
52 <br>
53 <blockquote>
54 Sometimes you may accidentally drop a piece in a spot you didn't mean
55 to. It's especially touchy on laptop touchpads if you're not familiar
56 with them. If you've made an accident like this, you're in luck. Just
57 click on Undo and your amazon will move back. If, however, you fired
58 your arrow, you're out of luck. Once you've done that, the turn is
59 officially over and there's no going back (at least for now. If I hear
60 from enough whiners^H^H^H^H^H^users who feel this would be a useful
61 feature, I'll raise the priority of multiple undo.
62 </blockquote>
63 <p>
64 <b>FORCE MOVE</b>
65 <br>
66 <blockquote>
67 Ok, it seemed like a good idea when i first designed the game, but it's
68 probably not all that useful right now. It's designed for the impatient
69 user who gets tired of waiting for the AI to move. Right now, you can
70 specify the maximum time for the AI to think each turn, so if you don't
71 like waiting, just set that to a low value. Eventually I'd like to give
72 the AI a variable time limit, perhaps within an overal set game time
73 (ie - it has 5 minutes to play its entire game, so budget that time as
74 best as possible). Then this button <i>might</i> come in handy if it
75 gets stuck on one move too long and you get impatient.
76 </blockquote>
77 <p>
78 <b>AUTO FINISH</b>
79 <br>
80 <blockquote>
81 Ah, one of the most annoying parts about Amazons in general is finishing
82 the game. Once you get to the point where all the pieces are in their
83 own little areas, the game is pretty much decided. It's just tedious
84 filling in the squares until one side runs out. This button is meant to
85 remove this boredom. Technically, all it does is set both sides to AI
86 with only 1 second to think. Generally, filling in squares is pretty
87 brainless stuff. The only problem is, it doesn't always find the
88 optimal fillin method. It's entirely possible (and in weird complicated
89 areas, likely) that it will seal off squares you could've used. If
90 you've got a close game, you could lose because of such an error.
91 Eventually, I'll fix this to fill in the squares perfectly. But for
92 now, use at your own risk.
93 </blockquote>
94 <p>
95 <b>REPLAY</b>
96 <br>
97 <blockquote>
98 This button will allow you to load a saved history file, and will let you
99 watch a game that the history file describes. In theory, you should be
100 able to load games from major tournaments, and be able to watch them in
101 Gamazons. Minor massaging of their notation might be necessary for Gamazons
102 to recognize it though.
103 </blockquote>
104 <p>
105 <b>STOP</b>
106 <br>
107 <blockquote>
108 This will stop a running replay of a game. This will allow you to go back to
109 a game that might have turned out differently if you did something else.
110 </blockquote>
111 <p>
112 <b>REPLAY DELAY</b>
113 <br>
114 <blockquote>
115 This sets the delay between moves in a game being replayed.
116 </blockquote>
117 <p>
118 <b>MOVEMENT SPEED</b>
119 <br>
120 <blockquote>
121 This sets how fast the pieces slide across the board. For slower machines, the
122 sliding might be painfully slow. Incrementing this value may make it more
123 tolerable.
124 </blockquote>
125 <p>
126
127
128 <a NAME="settings"></a>
129 <h2>Settings Menu</h2>
130
131 <b>PLAYER</b>
132 <br>
133 <blockquote>
134 Here you have the ability to choose whether a side should be a human or
135 AI. You can also set some strange settings for how the AI works: how
136 long he thinks, how far ahead he should look, and how many options he
137 should evaluate. Really, all you need to worry about is how long you
138 want the AI to think. However, if you really want to see how the other
139 setting affect its gameplay, here's what they mean.
140 <p>
141 The AI works by generating a search tree out of all the possible moves
142 it can make. First, it gives evaluates all of the moves it can make
143 this turn. Then it sorts them in a line, best move first, worst move
144 last. Then it looks at what its best move and what kind of moves its
145 opponent can make in response. Then the second best move, etc... all
146 the way down the line. A move's true value is not based solely on the
147 your move, but also what kind of move your opponent can do in response.
148 <p>
149 The max search depth means how far deep it will look (ie - moves it
150 can make, then moves its opponent can make, then moves it can make
151 from there ...).
152 <p>
153 The search width specifies how far down the line it will look evaluate
154 response values for. For instance, if there are 3000 possible opening
155 moves, what are the odds that the absolute worst opening move will
156 really turn out to be the best? Some theories say it's best to
157 evaluate all possible moves just to make sure there's not some trap in
158 ignoring something further down. So the default search width is 3000.
159 However, by evaluating the response of fewer moves, say 15-50 you reduce
160 the amount of work done at each level of the tree, thus you can see much
161 further ahead. Obviously, there's a balance to be made. And the
162 options exist so one can research the trade offs between searching wide
163 vs searching deep.
164 <p>
165 </blockquote>
166 <p>
167 <b>THEME</b>
168 <br>
169 <blockquote>
170 Now you can select from a variety of themes that show give your gameboard
171 a different appearance. The theme files are very simple and just
172 contain the type of image and its location. If you're feeling creative,
173 go ahead and experiement with your own themes. If it looks halfway
174 decent, send it to me and I'll probably include it in the next release.
175 Note: Square images are 40x40 pixels, and the game pieces should fit
176 within that size.
177 </blockquote>
178 <p>
179 <a NAME="future"></a>
180 <b>FUTURE ENHANCEMENTS</b>
181 <br>
182 <blockquote>
183 Soon you will see different difficulty
184 levels for the AI. If you're new to the game and sick and tired of
185 getting your can kicked every time, you'll be able to choose between
186 several levels of difficulty (I'll even throw in a level that plays
187 completely random so there's almost no way you could lose. Speaking of
188 which, remind me to add a feature that will detect if someone loses
189 against a random AI and have it email me the loser's name & email so I
190 can come back and taunt him). I would also like to break the AI up into
191 two different parts. So you could have black at level 9 with a
192 high max depth & high max width search play against white at level 6
193 with a low max depth & max width search. It's really hard for me to
194 tell if a new heuristic is an improvement unless I can play it against
195 the old version, so I'm not even going to try to make a smarter AI until
196 these features are in.
197 </blockquote>
198 </body>
199 </html>
200
201