RamSteelwood wrote:I would like chess more if it wasn't possible for anyone to remember prevous games/tactics etc. like if every game was fresh where you know the rules but other than that you're making it up. not that i've ever been good enough to learn any tactics or played anyone that good, but it does put me off.
acemuzzy wrote:It's that this? https://play.google.com/store/apps/details?id=com.chess If so, happy to try, but I'm really not very good!
davyK wrote:There's a doc about the 2nd Kaspersky match which generates a bit of controversy about that. There was a move in one game (game 2?) that Kaspersky thought looked dodgy - had signs of human intervention in his opinion. He said he never got over that move as he was never satisfied by IBM's explanation. And so his focus was gone. IBM won by a single game in a 6 game match. Maybe that was sour grapes. But maybe not. Chess matches at the highest level are all about psychology - something a machine is obviously impervious to. Kasperksy played a certain way against computers - abusing their weaknesses and that's how we won the 1st match. Maybe that was his undoing in the rematch? It could have been a development in the tech that flummoxed him. But maybe it wasn't as IBM were embarrassed by the machine's performance in the first match where Kaspersky won easily. The machine was disassembled afterwards and IBM enjoyed a share price increase. At the end Kaspersky was still asking of the diagnostic logs of "that" game. Deep Blue was a work in progress even during the matches and Kaspersky was able to detect this. So as an actual match it wasn't valid really - the machine was changed overnight usually. But as an exercise in AI dev, obviously it was extremely valuable. Could have been "cheating", could be IBM protecting their IP. Deep Blue was a stepping stone toward their cloud AI "Watson" service. It's academic now as chess has been mastered by AI now anyhow. It seems even Go has been toppled now. Watched a doc about that and a match against a Go master which the machine won.
Facewon wrote:Ace just hit me with a good move. Cat is hanging with pigeons.
Roujin wrote:That's covered in this documentary. Basically the IBM guy says that Kasperov was correct in the first game, he was still just playing computer chess, so deep blue as good as it was, still acted along the same risk/reward probability lines as any other computer, it just had the power to calculate waaaaaay more outcomes when it was deciding which move to make. However for the second game, the IBM guy says they trained Deep Blue extensively to play from the position that it wound up in in game 2 before it made the suspicious move. They identified that as a weakness after the first game and specifically made the system play from those kinds of positions so that it would be better at defending itself from them, so it made a move that it had either seen during the training, or that it had given additional weighting to during the calculations for the move because of the scenarios that had played out after the first game.There's a doc about the 2nd Kaspersky match which generates a bit of controversy about that. There was a move in one game (game 2?) that Kaspersky thought looked dodgy - had signs of human intervention in his opinion. He said he never got over that move as he was never satisfied by IBM's explanation. And so his focus was gone. IBM won by a single game in a 6 game match. Maybe that was sour grapes. But maybe not. Chess matches at the highest level are all about psychology - something a machine is obviously impervious to. Kasperksy played a certain way against computers - abusing their weaknesses and that's how we won the 1st match. Maybe that was his undoing in the rematch? It could have been a development in the tech that flummoxed him. But maybe it wasn't as IBM were embarrassed by the machine's performance in the first match where Kaspersky won easily. The machine was disassembled afterwards and IBM enjoyed a share price increase. At the end Kaspersky was still asking of the diagnostic logs of "that" game. Deep Blue was a work in progress even during the matches and Kaspersky was able to detect this. So as an actual match it wasn't valid really - the machine was changed overnight usually. But as an exercise in AI dev, obviously it was extremely valuable. Could have been "cheating", could be IBM protecting their IP. Deep Blue was a stepping stone toward their cloud AI "Watson" service. It's academic now as chess has been mastered by AI now anyhow. It seems even Go has been toppled now. Watched a doc about that and a match against a Go master which the machine won.
It looks like you're new here. If you want to get involved, click one of these buttons!