Chess is a game shrouded in romanticism and grand allusions- a fun game that I’m not insulting- but a game that’s built up as some sort of symbol of genius or some grand context.
It is more cold numbers than anything else. Let me put it thusly- chess is a very complicated game of tic tac to with way more possible ways the board can develop. At its core though- for all intents picture chess like tic tac to. Any human that knows how the game works and is beaten in tic tac essentially got confused or mixed up and made a mistake. They either lost track of the moves or forgot to do X and did Y or they somehow mentally mixed up what they were looking at. Computers don’t really have that problem. They don’t really mix up where pieces are or somehow miss a move or miscount or misvisualize the board. For every move there is a move that leads to a victory or a defeat-
And numerous scenarios for each. In other words, you can plot out essentially every possible combination of moves that can…
Happen and which will lead to victory and which will lead to a defeat. At some point, if you have two opponents who each always make the “best move” to win, you can declare the game is over because all moves lead to defeat so long as each player plays the “best move to win.” When a player deviates and doesn’t always play the “best move to win,” that strategy (or luck) can unlock scenarios that don’t exist with a “best move.”
However- so long as opponent A keeps playing the “best move to win” based on a long term plan of contingencies, player B can essentially only prolong the game by doing this, as making sub optimum moves will almost always lead to defeat of the opponent is making all superior moves to counter.
If a well designed computer opponent is flawed or somehow experienced a “bug,” or if a players style somehow managed to exploit a flaw in the logic of the computer, they could win this way. But let’s remove computers for a moment. A human who is highly skilled- the best human chess player alive let us say- playing against a rookie opponent. There is always some chance of defeat, but it is highly statistically unlikely.
The advantage of the master player- the primary advantage, is they have board configurations and strategies and counter strategies memorized. If the rookie enters the game with a strategy, chances are it will be easy for the master player to counter. Of the rookie player enters the game with no strategy, they have very little chance to beat a master player because well… seldom in the history of war or any strategic enterprise has a rookie without strategy beaten a veteran with excellent strategy on merits of play.
What that means is simply that while poor weather of chance events can turn even the most unbalanced battle, while some thing that we can call “luck” can swing even the most unbalanced battle, where we remove those factors and can say each side has fairly even terms outside their specific strategies and preparations, haphazard tactics seldom prevail. Now, when two humans play there can be “bluffs” and lots of things beyond the board. Of a veteran doesn’t know they are playing a rookie they may be beaten if they can be tricked to believe that they may be playing a pro or are witnessing some masterful unconventional strategy. People can be distracted or intimidated or otherwise impacted by psychological tactics and such.
Our computer is like our veteran only generally much better. A poor strategy probably won’t beat it. A masterful strategy probably won’t beat it. It won’t try to read the opponent or get mixed up on what they are doing. It simply accepts what is on the board and determines the best path to victory from that point. In a sense we can say every move is a new game to the computer. It can throw away all its strategy and planning based on what the board looks like at any moment and come up with the next set of strategies. Even on the simplest form, if a computer contains a data base of all the moves and/or all the possible configurations of a board, and it knows the statistical odds of winning based on that set of configurations, if it merely chooses the set of linked configurations with the highest average chance to win and then chooses the move from each set so that at the end of its turn the board is the position which has the highest probability of victory and which the average victory..
.. probability from all possible moves to counter have the highest victor probability for the computer, then it would already most likely win. I won’t get into it, but you can often also increase the odds by simply programming a machine to not try to win but seek the highest pro abilities that it will not lose. A sophisticated machine could have the ability to exercise either of those strategies or both at once and alternate based on probabilities and progress. This severely limits the ability of even a skilled opponent playing met game tactics from exploiting the computers strategy and claiming victory. So chess is a game computers tend to be well suited for. There are many games which contain elements of chance or abstracts or elements or Social play which current computer technology or design are not well suited for.
When speaking of high end purpose built machines, a human opponent has officially beaten a computer opponent in organized chess in over 15 years. Early chess computers proved their abilities even when the best computers possessed less capability and processing than most phones or even some peoples watches might have today. We can say that it has been uncommon for a long time for a top ranked human player to beat a computer at all, but the most common way humans have been able to beat computers is exploiting flaws in the machine. As newer machines eliminated those flws and increased capabilities, it became much harder for a human to beat a computer. In theory a human could memorize all the information on every possible game of chess and possibly be able to averse it in a manner that rivals a computer though is somewhat different. However as computers increase in ability, even the greatest humans start to show gaps in what they are capable of consistently vs. the machine.
idk if guest_ mentioned it or not because i just skipped his stuff as im just checking the funsub before bed, but in the case of FPS shooters the difficulty isnt in making a bot thats good. With Chess making a better smarter bot is the primary point of advancements, with a bit of a shift nowadays towards "learning your skill level" stuff but as far as i can tell thats not the chess AI doing it, its a different AI matching you with a given difficulty bot to learn more about how you fair against various strengths.
With FPS games the difficulty is in making a fair bot. If the devs wanted to make the bot play as good as possible it would always hit you in the head every time it fired, never missing always firing the frame that its possible to kill you every single time. interestingly some FPS games do actually do this, but the damage the player deals is a vastly larger percentage of the total health of the enemies compared to the enemies damage percentage of the players total health (cont)
games that dont use that method generally have the difficult task of figuring out what a "fair and fun" balance of aim quality is, if the enemies use the same weapon profiles as you do, etc. Positioning in a fps is also far more complex than in chess. Theres a whole 3d world with cover, concealment, sight lines, distance, etc to deal with.
I didn’t touch this topic, mine was more chess bot specific. Game “AI” is interesting stuff. As you say, the bigger challenge in many games including chess is creating a computer opponent that is fun to play against and doesn’t just win. Many older games especially would just leverage the fact that the computer knows your movements and due to delays between input of a command and video output, the computer can generally “react” to your actions before you’d have a chance to even see the result of your command. In other words- if you decide to move left or jump, the computer receives the command before your “avatar” actually moves, and in many scenarios it can thusly process the reaction of an “enemy avatar” to that input before your “avatar” even moves. Since any movement from a jump to a run or a dodge is all
Plotted out in numbers ahead of time so that the graphics on screen and hit boxes etc. are where they need to be, where a movement ends up or any point between are known to the
Computer. If it was only programmed to win, something like jumping a projectile or dodging an attack would become almost impossible, or a follow up attack would land with surety. A real challenge is creating a virtual opponent that isn’t infallible but also isn’t woefully inept. What many trying to design enemy behaviors struggle to achieve is a computer opponent that can feel more like a “human opponent.” The scripting of computer opponents makes it difficult to create behaviors that aren’t easily “exploitable,” where a player can anticipate with certainty or near certainty how a computer will react to a given series of events. There are different ways to achieve this- in our FPS example, programming a computer so that it will miss X% of the time is one example. So this enemy theoretically has 100% accuracy but a “dice roll” determines wether that shot actually hits; or it may “always hit” but a dice roll determines if the shot is not counted and does 0 damage etx.
It’s all really neat stuff and when we start to think about and look into how even seemingly simple things aren’t so simple it gets neater. Computer opponents may “take cover” or “run” or perform other actions to help them seem more “natural” or “realistic.” Just like the player, the computer doesn’t actually have a “real” fear of death in a virtual world. A computer opponent controlling an army of “thugs” for example, if given decent hardware and programming, could easily figure out that simply “mobbing the player” was the “best” solution without care for trying to preserve any individual “life” of the thugs. Simple computer opponents may use a path finding to reach a similar behavior- they just direct enemies to take the most direct path to the player and the enemy types effective combat distance.
We see this a lot with older games and poorly made games- an enemy might be on the other side of a wall where they can’t reach the player, but be trying furiously to run through the wall or attack the player from the other side of the wall. Players have often exploited such enemy AI by using calculated movements to “corral” enemies where it benefited the player. Scripting to give enemies alternate path finding when “stuck” or to avoid routes with obstacles between them and the player etc. can help there- but that’s the thing that complicates it all- the computer has almost omnipotent control and knowledge of the game world. For example It knows the entire map, even those areas a player cannot access. So part of making a “fun” computer opponent is in defining what information and what conditions under which the computer may use, and how it is used, in controlling the enemy. Lots of fun stuff.
It is more cold numbers than anything else. Let me put it thusly- chess is a very complicated game of tic tac to with way more possible ways the board can develop. At its core though- for all intents picture chess like tic tac to. Any human that knows how the game works and is beaten in tic tac essentially got confused or mixed up and made a mistake. They either lost track of the moves or forgot to do X and did Y or they somehow mentally mixed up what they were looking at. Computers don’t really have that problem. They don’t really mix up where pieces are or somehow miss a move or miscount or misvisualize the board. For every move there is a move that leads to a victory or a defeat-
And numerous scenarios for each. In other words, you can plot out essentially every possible combination of moves that can…
However- so long as opponent A keeps playing the “best move to win” based on a long term plan of contingencies, player B can essentially only prolong the game by doing this, as making sub optimum moves will almost always lead to defeat of the opponent is making all superior moves to counter.
The advantage of the master player- the primary advantage, is they have board configurations and strategies and counter strategies memorized. If the rookie enters the game with a strategy, chances are it will be easy for the master player to counter. Of the rookie player enters the game with no strategy, they have very little chance to beat a master player because well… seldom in the history of war or any strategic enterprise has a rookie without strategy beaten a veteran with excellent strategy on merits of play.
With FPS games the difficulty is in making a fair bot. If the devs wanted to make the bot play as good as possible it would always hit you in the head every time it fired, never missing always firing the frame that its possible to kill you every single time. interestingly some FPS games do actually do this, but the damage the player deals is a vastly larger percentage of the total health of the enemies compared to the enemies damage percentage of the players total health (cont)
Plotted out in numbers ahead of time so that the graphics on screen and hit boxes etc. are where they need to be, where a movement ends up or any point between are known to the