In the first in a new series of AlphaGo video commentaries, Michael Redmond 9p, hosted by the AGA E-Journal’s Chris Garlock, reviews Game 1 of the amazing AlphaGo vs. AlphaGo selfplay games. The 50-game series was published by Deepmind after AlphaGo’s victory over 
world champion Ke Jie 9p in May 2017. Games 2-4 will be released this week, leading up to this year’s U.S. Go Congress in San Diego, which starts on Saturday.
“In the Master series earlier this year, AlphaGo first showed its big shimaris, and often played to dominate the center of the board in the early opening, using its famous shoulder hits to do so,” says Redmond. The Master version “had an early advantage in almost all of the games,” Redmond says, “and I was impressed with its ability to simplify complicated middle game positions and bring the game to an early outcome. Less convincing was the way Master handled complicated joseki. It also had a disturbing habit of losing several points in the endgame to win by the smallest possible margin.”
“With this new self-played series,” says Redmond, “I wanted to see how these traits had survived AlphaGo’s evolution; five months is a long period of time for a self-teaching AI. The exciting news is that Alphago has changed dramatically.” Click here, to find out how in the video commentary and see below for Redmond’s extensively commented sgf file. The videos are produced by Michael Wanek and Andrew Jackson; sgf editing support by Myron Souris.
Redmond’s AlphaGo-AlphaGo commentaries launched
Tuesday August 1, 2017
In the first in a new series of AlphaGo video commentaries, Michael Redmond 9p, hosted by the AGA E-Journal’s Chris Garlock, reviews Game 1 of the amazing AlphaGo vs. AlphaGo selfplay games. The 50-game series was published by Deepmind after AlphaGo’s victory over
world champion Ke Jie 9p in May 2017. Games 2-4 will be released this week, leading up to this year’s U.S. Go Congress in San Diego, which starts on Saturday.
“In the Master series earlier this year, AlphaGo first showed its big shimaris, and often played to dominate the center of the board in the early opening, using its famous shoulder hits to do so,” says Redmond. The Master version “had an early advantage in almost all of the games,” Redmond says, “and I was impressed with its ability to simplify complicated middle game positions and bring the game to an early outcome. Less convincing was the way Master handled complicated joseki. It also had a disturbing habit of losing several points in the endgame to win by the smallest possible margin.”
“With this new self-played series,” says Redmond, “I wanted to see how these traits had survived AlphaGo’s evolution; five months is a long period of time for a self-teaching AI. The exciting news is that Alphago has changed dramatically.” Click here, to find out how in the video commentary and see below for Redmond’s extensively commented sgf file. The videos are produced by Michael Wanek and Andrew Jackson; sgf editing support by Myron Souris.
[link]