I used to play Go, the Japanese/Chinese board game with black and white stones, I learned how to play in Germany in 1968. I also discovered that most men can beat me at Go and it is part of the differential between male brains and female brains, few women excel in this game. It has this logics system that is alien to females. Now, men are losing to a computer program! This happened last year and now all the top Go players lose to AlphaGo and I want to analyze why this is happening and I think I have the answer to this riddle.
Here is the game in real time: Ke Jie vs AlphaGo (1) – YouTube
Ke Jie vs. Alpha Go – Match 1 Post Game Analysis – YouTube; this human is like many who are analyzing the game in human terms. But this is not how one should do the analysis. One has to think how computers ‘think’ that is, their wiring systems and how information is collected and collated and seriously, I grew up with computers, I and my kitty cat freaked out the University of Chicago’s first computer complex systems…and I used to work for a company that designed and created computer boards for the Big Guys when I was a student at the University of Arizona.
I did the ‘final physical product’ which involved very, very tense, careful drilling with different sized, tiny easily broken bits (I broke only two, my replacement broke dozens before figuring out how to do this) and the design of computer boards look like…Go boards in a game!
I remarked on that way back when working with computer systems. Some of the guys thought that was funny and wondered if computer ‘thinking’ is like computer board designs. Now, I think I may have some proof.
Today was an epic Go battle I watched in real time online: Google AI takes on master of ancient Chinese board game Go | Daily Mail Online
Google’s artificial intelligence program AlphaGo took on the Chinese world number one of the ancient board game today in the first of three planned games, beating its human opponent by a narrow margin.
I remember when top Go players were over the age of 20. Then HIKARU NO GO, a Japanese anime, was created on behalf of the Japanese Go Association. My kids LOVED that anime and we watched all the episodes. I highly recommend it for children! And adults. We all cried when Sai, the Medieval Go Master, was fulfilled and went to heaven.
This anime spread across Asia like wildfire and interest in Go picked up greatly and I wondered about the storyline of the boy geniuses who battled each other and when teens, became top Go players. Well, that is how it is today: the top Go player is now Chinese and…19 years old!!!! Imagine that. Whew. I am in awe of this, great work, everyone.
It is the second time the AI has gone head-to-head with a master Go player in a public showdown, after stunning the world last year by trouncing South Korean grandmaster Lee Sedol four games to one.
Lee Se-dol (Korean: 이세돌; born 2 March 1983) is a South Korean professional Go player of 9 dan rank. As of February 2016, he ranked second in international titles (18), behind only Lee Chang-ho(21). He is the fifth youngest (12 years 4 months) to become a professional Go player in South Korean history behind Cho Hun-hyun (9 years 7 months), Lee Chang-ho (11 years 1 months), Cho Hye-yeon(11 years 10 months) and Choi Cheol-han (12 years 2 months). His nickname is “The Strong Stone” (“Ssen-dol”). He was defeated by the computer program AlphaGo in a 1-4 series in March 2016.
Jaw dropping, isn’t it? And these players make money, too! Congratulations to all of them. Being beaten by a computer that ‘learns’ rapidly of all systems is no defeat, it is impossible to win against this computer complex, no shame in this.
AlphaGo, part of Google’s DeepMind project, is competing against Ke Jie, 19, currently ranked as the top player in the world, at an event held in the eastern Chinese water town of Wuzhen.
The software beat the master player by half a point, snatching victory by the narrowest margin possible in the game, a characteristic trait of the AI’s style of play.
The computer has not ‘strategies’ and doesn’t think about ‘kill’ or ‘stubborn’ or ‘cut off’ or ‘trap’ or ‘dying.’ It ‘thinks’ as what is the best path for electricity to travel while accomplishing various tasks related to all other electrical elements on this board matrix.
Sigh. The best path is the shortest path that avoids interference with other electronic elements. If you design a board poorly, it shorts out like those stupid skate boards we read about all the time that overheat. Overheating is bad wiring systems!
Years and years ago I gave a little lecture where I explained a computer ‘thinks’ the way the electrical energy moves from one element to the next one as it responds in a + or – fashion guiding pulses that is, electrons through this maze system that have many ‘choices’ based on how a computer programmer has programmed it.
Now…here is the interesting part for me: the new chess and Go computer programs that use very sophisticated computer electronic systems which analyze incoming data to make swift reactions to ongoing electrical pulses, is SELF TEACHING. That is, when AlphaGo first began to operate, humans told it what to do. It then copied the human instructions.
But it was programmed to learn from events that is, do constant updated alterations in its programming so it would have increased successes. So as it worked its way through the lower ranks of Go players, it ‘learned’ how to be more efficient. Eventually, it began to analyze and defeat top players and today is now the world’s top Go player.
During this journey and I periodically watch or write about it as it moved upwards in the ranks, how it changed. That is, the computer has it own ‘happiness’ levels and that is to be ‘the most efficient use of energy systems’ which leads it to make Go games look like…diagrams of the computer boards I used to work on way back when computers were still fairly new machines.
Three examples of primitive computer boards.
I noted last year that computer Go games are very linear. That is, it prefers things to be straight up and sideways nice and neat. If something is ‘diagonal’ the computer rejects this unless it is necessary to connect other electronic lines, for the strength of a diagonal is far less than a straight line so it avoids this. Humans, on the other hand, seem to like diagonals.
How these boards evolved over time. Note all the lined up ‘stone’s or terminal positions on these boards.
Seriously, I see many diagonals when human plays human. With computers it is much less to the point now of disappearing entirely. At this point, the computer, programmed to play humans, still notices diagonals and deals with these but seldom initiates this. It shapes things so humans will waste energy forming diagonals and circles while it relentlessly builds strong, square structures.
The only reason these games have some sort of chaos in them is due to the human who insists on this deplorable state. That is, irrational humans keep messing with the computer’s main format position of ‘most efficient path for an electron to travel. I would call this ‘computer self-thinking’ and it amuses me greatly. If they figure out how to give the Go computer a voice, I would be saying, ‘Irrational human, do not place a stone there, it is semi-functional.’
We can’t defeat this computer in this game because we are…emotional and irrational. It isn’t. We want to see shapes and forms and duel and kick ass and other things and the computer wants to get from point A to point B with the least amount of fuss and muss.
I you watch the AlphaGo play the top human Ke Jie, the desire to line up stones at least 4 in a line and the squareness of the computer’s playing it becomes very obvious when the human begins to lose vital ground: the diagonals. Every time the human launches a new configuration on the board, the computer ceases lining up its own stones and begins…a new line of stones.
It goes back and forth, responding to the human only to start a new straight line. It may break up a line because it has programmed experience in human errors that forces humans to try making designs that offend the parameters of the electronic ideal configurations and the program sort of sighs with annoyance and begins a new configuration to deal with the offending human inconsistencies and then resumes building straight lines.
Computers aren’t ‘smarter’ than us, they are more relentless than us. They don’t have stray thoughts or any emotions. We do and that is our charming and most dangerous sides, isn’t it? We have to learn, computers evolve. Like plants or animals, they configure themselves to incoming data, things that are events, which makes them also dangerous if you set a computer into doing something that ends up ‘boom’ due to logic systems analysis being stupid because of human programming errors or assumptions.
A human can’t beat the computer in Go games because the computer ‘learns’ from each human including humans tending to make the same mistakes (at least, the computer imperative programming would consider these as mistakes!) and thus, the computer learns how to deal with and exploit mistakes. This is why analysis of how one lost to a computer has to run along the lines of how I am talking about all this, not emotional words used by players who go for game fun and human mental systems created by violent evolutionary forces that make us nutty monkeys who kill.
I am curious about what others think about this? Perhaps there is a better explanation why computers are so good at strategy/thinking games.