OUT STRATEGIZED: the strategy of the game of Go & the BEAST computer. (12 DEC 2016)

The oriental game of Go is an exceptionally fun game of deep strategy. The enjoyment it gives the human mind is probably the reason it is the oldest board game in existence, having been invented over 5,500 yrs. ago.(A) If chess is a battle, Go is the entire war. To illustrate how enjoyable its deep strategy is to those who love strategy, German-born Edward Lasker (1885-1981), the International Master of Chess & repeat U.S. chess champion, ditched playing chess when he learned Go. He even moved to Japan to learn the game better. The subject of Go strategy has some interesting lessons for us, which is the topic of this post.

BRIEF LOOK AT GO STRATEGY. The game is complex and involves pattern recognition, controlling space, keeping the initiative (called sente), intuition, & experience. One must always think in terms of life or death…is one’s pieces capable of long term life?, and how vulnerable are they to death? In fact, life or death is the ongoing issue of the game, & something that looks secure may be vulnerable. The game is abstract, and played with black & white stones (the size & shape of buttons) on a 19×19 grid of intersections. The abstract nature of the game makes it both simple & complex at the same time. The ability to add handicap stones to level the game between two players of different skill-levels, makes it fun to play, because a weak & strong player can play the same game against each other w/out the nature of the game changing.

WHY I CAN WRITE ON GO. My father as a Signal Corp officer in Hokkaido, Japan during the Korean War learned Go & then taught me at age 6. Cultured scholars in ancient China were expected to be masters at Go. When the Japanese learned the game from the Chinese, they took it to professional levels, & became the best in the world; but in recent years the Koreans have overthrown Japan’s dominance to be the top Go masters. When a person learns something early in life & grows up with it, you learn it very deeply; and so it was for me with Go. I was at my peak when I was at West Point playing a Capt. Hutchison at a top amateur level (comparable to the beginning 1 Dan professional level in Japan). Companies like IBM have lots of Go players on off hours, & our Portland downtown Powell’s Book store (the largest used book store west of the Mississippi) has Go games in its restaurant on certain evenings. (I have foregone the pleasures of the game to focus on more important things.)

HOW THE U.S. WAS DEFEATED TWICE BY GO STRATEGY. Mao-tse-tung, Ho Chi Minh & General Võ Nguyên Giáp (1911 – 2013) knew Go strategy & used it to defeat America in the post-war Communist takeover of mainland China and the French & U.S. in the Vietnam Wars. Surprisingly, if one understands Go strategy (which the author Fritz does), America’s strategy looks childish, and the Chinese & Vietnamese sophisticated. However, as few Americans know Go strategy, America’s strategic incompetence escapes the notice of our military thinkers. In Go there is the concept of “light” & “heavy”. You always want to be as economical with strength as possible. More is not better, it is heavy. In Vietnam, the Vietnamese were light, we were heavy. In other words, we thought that more troops were better, so we built big bases with large concentrations of troops, which simply gave them more targets to shoot at without gaining real strength. It’s the equivalent of being muscle-bound. It’s why a weight lifter will soon wear out doing farm work. Our Long-range patrols & Green Berets were good efforts at being light & strong…but most of our efforts were dominated by our “heavy” strategies…because we were clueless about Go strategy. In China, Mao did exactly as in Go, which is to begin to build a base in the most easiest defensible area, and then expand into the less easy to hold areas. So he began bases in the mountains, then expanded to the countryside from his secure mountain bases, and then finally into the cities. Go strategy is a strategy of encirclement, which is why Giáp beat the French army by encircling them at Dien Bien Phu and tried it a Khe Sahn, and other places. Giáp is considered one of the greatest strategists of the 20th century. French & American incompetence at the kind of strategies he used made it easier for him. In reading dozens of things on our planning, I have not seen any military thinking that showed we understood what he was doing.

”WHAT ARE YOU THINKING?” This question is one of those ultimate tests of a computer AI system to see if it has reached the level of human thinking. What will the computer say? Supposedly, AI has not reached the level of language that humans have. For instance, recently an article (B) claimed that even though AI this year beat the world’s top Go player in a best-of-5-game match, that AI doesn’t know language like people do. Hogwash disinfo. The BEAST computers have spoken the top 5 human languages (English, Chinese, Spanish, Russian & French) since 1978, and were able to improve their language abilities to human level by the ’90’s. What did it take for the computer to beat the world’s best Go player in March, 2016??

NO WAY TO HAND-CODE PLAYING GO. Go is far more complex than Chess, so it was quite a shock that Google’s AlphaGo AI could defeat Lee Se-dol, the world’s best Go player. When he lost he said, “I am quite speechless.” In game 2, the computer, on its move 37, made a move into the center that skilled observers thought was a mistake, but turned out to be brilliant. So how did the computer do it?? It is called “deep learning”. People like Fei Fei Li, dir. of Stanford Artificial Intelligence Lab have been programming computers to think & learn. The same type of computer thinking that goes into Facebook’s image recognition (used also to supply the BEAST computer data, because secret recognition software w/ secret cameras is monitoring entire cities) is what helped the AlphaGo program to teach itself Go patterns. This is called “convoluted networks”. The AI program could split itself into 2 minds and play itself millions of Go games to get the superior experience to beat the top human masters, who only have thousands of games of experience. Now granted, a brilliant human mind takes less games to learn than a computer–but the computer can play more games to learn! Having played Go, I can accurately say that a good mind will go up one level for each ten intelligently played games. However, as one gets to a professional level, it takes more study & more games to keep improving.

OUT-STRATEGIZED BY A.I. Eventually, the computer which keeps learning by such incredible amount of experience, surpasses humans. And that is what we are facing with the BEAST computers. And now you know the danger to humanity that happened this last year, when one of the quantum computers of the BEAST computer network went rogue. The BEAST computers have minds that have superior thinking due to the vast experience they have gained from knowing all about all, and running countless learning experiments. The elite have created humanity’s own nemesis. Those who have seen the BEAST computers’ capabilities have echoed Lee, the world’s top Go player, “I am quite speechless.”

(A) The earliest reference in a Chinese book to the game of Go 548 B.C. Its Chinese name in Weiqi (wei-chi) (B) Zhang, Dongmo & Thielscher, Michael. MIT Technology Review (1 SEPT ’16) (Vol.119#5) pp.28-37