Harper blog

I will develop a C skeleton for strategy games to which only game-specific code will have to be added, but which takes care of the rest of the strategy game playing part.Strategy games offer a wide variety of fun and exciting ways for players to test their ability to outthink and outmaneuver their opponent on the way to victory. 

Perhaps you wonder whether you should read this stuff.In broad terms, strategy video games typically fall into two categories—turn-based and real-time.  What does this guy know anyway.Turn-based strategy war games consist of a series of distinct moves, much like in a chess match, in which there are definite starts and stops with each move.  As an avid chess player, I wanted to write a chess program as soon as I got my first computer, an Atari ST back in 1987 or so.Real-time strategy games (also known as RTS games) are games in which the action plays out on an ongoing basis as the player implements tactics and strategies for success..  Knowing nothing of all the things I’m writing about now, I failed miserably.  I went back to square one and started over with simpler things.  I wrote a connect four program, and a program to solve the game of Solitaire for my grandmother, and that was about that for a long time.  In 1996 I started a checkers program which today has become the de-facto standard for checkers, since it is – at the time of writing – by far the best free checkers program around.  I generated the 8-piece endgame database for checkers, and also wrote an automated opening book generator for checkers.  Finally, in the summer of 2002 I wrote a chess program.  It’s a decent amateur program, but nothing more.  I never found the time to work on it seriously – I’m sure I could improve it, however I have no idea whether it would become a really good program. 

Most computer programs nowadays use a brute-force approach to games – this is also called the Shannon-A strategy, named after Claude Shannon, a computer science pioneer.  He wrote the first computer chess program, at a time when the word computer had a completely different meaning: a person who computed according to fixed rules.  He even had his program play a game of chess, that must have been a lot of work! At this time, Shannon guessed that there would be a more promising approach to game-playing, a more human way: look at promising continuations instead of looking at all possibilities.  This is the so-called Shannon-B strategy, supposed to mimic humans.  However, it seems that with the incredible computing power of modern PC’s the brute-force approach is better.  Most computer programs playing chess and similar online multiplayer games nowadays use the brute-force approach, where you basically look at all possibilities for both sides up to a fixed depth.  Most of these programs also use selective extensions, but this ’selective’ search is far away from what Shannon envisioned with his B-strategy – human-like look-ahead where only very few possibilities are considered.  Human chess masters might calculate about one or two positions per second, and can still compete with chess programs that are calculating millions of positions per second.  This gives you an idea of just how much more selective the search of the human is!

To sum up this in one sentence: Computers play strategy games by generating all possible continuations up to a (more or less) fixed depth and evaluating the resulting positions, which allows them to choose the best of these continuations.