I have been developing A.I's to beat games for 4 years now, (tic-tac-toe, connect 4) mostly games with a set way to guarantee a win or draw. But I gave my self a challenge, make an A.I. (or in this case an engine) to beat chess, this was easily my biggest challenge yet. Here is the journey.
I originally started out with analyzing Hikaru Nakamura and Magnus Carlson games, by taking them and having them analyzed by my A.I. (That's what it was at this point) but I ran into a couple problems, 1. It took, a LONG time just to analyze a single game would be upwards on 30 minutes, 2. Even tho they are grandmasters, they will make mistakes. (This would eventually work itself out, but I wasn't gonna sit and wait) eventually I gave up on the idea of an A.I.
Next I went with the idea "what are all the possible moves, that will get me to checkmate my opponent." basically It would analyze as many moves as it can and find what percent of those moves would return mate for me, draw, or mate for my opponent, and picked the one that had the most check mate for me percentage. This led to my engine getting scholar mated, because it found checkmates for it later down the line.
So I switched it to " All possible moves that will get me to checkmate my opponent, WITHOUT them having a forced mate" and this is where my engine is now after a bit of tweaking, it calculates as many possible moves as it can, that lead to it getting a positive checkmate ratio, without allowing the opponent a forced mate. Now its time to face stockfish
7 games were played (because my engine absolutely EMBARESSED Stockfish 10)
So am I a better coder than the Stockfish devs? In short. No.
Why did my engine win? 1. Time Odds, My Engine took roughly 3-6 times the amount of time Stockfish took to make a move.
2. I'm playing BROWSER Stockfish 10, which is not tournament Stockfish
3. I wasn't stuck to an opening. When i played stock fish, they had a set opening that they used when I had free reign
If you guys want to see an updated battle between me and Stockfish let me know, this will be different because ill allow my engine 3 seconds a move, and use the computer championship opening book.
TLDR; I made engine, it beat Stockfish, I beat because of potential weaknesses to the Stockfish I played, should I rematch without weaknesses
I have been developing A.I's to beat games for 4 years now, (tic-tac-toe, connect 4) mostly games with a set way to guarantee a win or draw. But I gave my self a challenge, make an A.I. (or in this case an engine) to beat chess, this was easily my biggest challenge yet. Here is the journey.
I originally started out with analyzing Hikaru Nakamura and Magnus Carlson games, by taking them and having them analyzed by my A.I. (That's what it was at this point) but I ran into a couple problems, 1. It took, a LONG time just to analyze a single game would be upwards on 30 minutes, 2. Even tho they are grandmasters, they will make mistakes. (This would eventually work itself out, but I wasn't gonna sit and wait) eventually I gave up on the idea of an A.I.
Next I went with the idea "what are all the possible moves, that will get me to checkmate my opponent." basically It would analyze as many moves as it can and find what percent of those moves would return mate for me, draw, or mate for my opponent, and picked the one that had the most check mate for me percentage. This led to my engine getting scholar mated, because it found checkmates for it later down the line.
So I switched it to " All possible moves that will get me to checkmate my opponent, WITHOUT them having a forced mate" and this is where my engine is now after a bit of tweaking, it calculates as many possible moves as it can, that lead to it getting a positive checkmate ratio, without allowing the opponent a forced mate. Now its time to face stockfish
7 games were played (because my engine absolutely EMBARESSED Stockfish 10)
6W, 1T, 0L
Results of each game, and links to move list
So am I a better coder than the Stockfish devs? In short. No.
Why did my engine win?
1. Time Odds, My Engine took roughly 3-6 times the amount of time Stockfish took to make a move.
2. I'm playing BROWSER Stockfish 10, which is not tournament Stockfish
3. I wasn't stuck to an opening. When i played stock fish, they had a set opening that they used when I had free reign
If you guys want to see an updated battle between me and Stockfish let me know, this will be different because ill allow my engine 3 seconds a move, and use the computer championship opening book.
TLDR; I made engine, it beat Stockfish, I beat because of potential weaknesses to the Stockfish I played, should I rematch without weaknesses