The annual Kyoto Artificial Intelligence Conference usually attracted little attention on the global news cycles. More prestigious local outlets such as the Yomiuri Shimbun, Japan's largest newspaper, traditionally reserved a mere paragraph for one of the biggest gatherings of AI enthusiasts on Earth. Yomiuri Shimbun's Western counterparts such as the New York Times and the Chicago Tribune never left a hint that the event even existed. Not that the respective editors didn't feel the AI conference wasn't important, it's just that politics and sports appealed much more to the readership.
However, the meeting in 2002 was different. For the first time, the conference's organizing body actually commissioned a committee to promote the event. Commercials were aired, flyers were posted. By three days before the meeting, the Kyoto International Conference Center had run out of room and the committee began to book nearby convention centers to accommodate the anticipated crowd of 20,000. Two days before the event, an army of cars and bicycles occupied all the available space within a four mile radius of the convention center. And one day before the event, the attendees had already worked up a frenzied atmosphere redolent of Woodstock or Live Aid. Excited scientists began to line up for the main talks half a day before the doors were to open. They chatted amongst themselves about how the new discovery would change the face of machine learning and algorithms forever. Most of them understood the technical jargon that saltated between tongues and ears, but some of them did not because they were Go players. These Go players brought a different species of anxiety to the conference center. "Nodes" and "permutation trees" and "pruning heuristics" were of no concern to the Go crowd. They just wanted to know if the claims that Go had been solved were really true.
At 5:00 PM, right when the minute hand reached the twelve, a dozen attendants in three piece suits opened the auditorium doors from the inside, and a deluge of bodies filled the presentation room almost all at once. Scientists jockeyed for seats and then for places to sit on the walkways. The men in suits simultaneously tapped on hidden earpieces and then began to close the doors. They were opposed by the crushing force of hundreds of desperate latecomers who gathered one final collective effort to enter the main presentation hall. Only a few stray bodies slipped into the auditorium before the ushers – after a superhuman effort – finally slammed and locked the doors.
It was 5:12 PM now, and the room simmered with the sweltering fog of body heat. A throng of cameramen directed dozens of lenses at the stage, the wiring from their devices twisted in an ugly jungle of vulcanized rubber. Low-decibel discussions dissolved into silence as the conference chairman, an old man in a starchy suit by the name of Dr. Hideo Kobayashi (Ph.D, of course) limped to the podium at stage right. He tapped on the microphone twice and the room thumped twice in response.
"Ladies and gentlemen," he announced in an English viscous with Japanese flavoring, "welcome to the 10th annual Artificial Intelligence Conference in Kyoto."
The audience broke into delirious applause, half in response to the speaker and half as a reward to themselves for making it into the main lecture hall.
"As you all know by now, we are running a marathon session today from 5 PM until midnight. We have allotted the entire first day to the work of Dr. Samuel Peoples and his associates. Dr. Samuel Peoples attended the University of Wisconsin at Madison as an undergraduate and received a MacArthur Fellowship at age 22 for his work on quantum computing. He then attended Stanford University, where he completed his Ph.D on the foundations of Computer Go. Afterward, he landed a faculty position at MIT, where he is now a full professor of computer science and works on superconductor applications in Turing machines, quantum decoherence, and, of course, computer Go. Please join me in welcoming our first speaker, Dr. Samuel Peoples."
The audience erupted into seismic peals of ovation. Half the guests were already standing. The room rocked with the force and noise of an earthquake. Among the cheering and whoop-whoops, near the anterior of the lecture hall, Kouyou Touya watched the new figure on stage with unflappable serenity.
The man was tall, his head brimming with gray hair. Wrinkles marred his cheeks, a few fresh ones from the anticipation of the conference, and several deeper ones only possible from years of tireless toil, very much like the creases on the Meijin's own face. The presenter set a glass of water on the podium and dabbed the perspiration from his neck with a handkerchief.
After all the audience's mania had converted into attention, the man on stage smiled and took a sip from the glass. A large, gray screen slowly descended from the ceiling at center stage, and a projector at the back of the auditorium flipped on to reveal the title screen for the talk:
"THE HAND OF GOD: WORKING TOWARD A PERFECT COMPUTER GO ENGINE"
The presenter fiddled with a laser pointer and then cleared his throat. The intensity of the lights bespoke the likes of a circus event rather than an academic presentation. The man's forehead was mired in sweat again.
"Before I get to the technical stuff, I'd like to first present … a proposition.
"I'm sure that you've all heard by now about Shusaku, which is the name of the Go engine I've developed this year. First, a bit of history."
The man tapped on the laser pointer, and the screen switched into a black-and-white contour of a young man. On the bottom was a caption that read: "Honinbou Shusaku, June 6, 1829 – August 10, 1862."
"As many of you know already, the engine's namesake was probably the greatest chess players of all time. Honinbou Shusaku was born in the early 19th century in Onomichi, Japan and quickly rose to prominence as a Go prodigy. His style was kinetic and ambitious. He aimed to claim gobs of territory early and defend them ferociously until the end. This style of play demoralized weaker opponents and frustrated stronger ones. Either way, Shusaku's calculating ability was second to none, and he boasted an impeccable sense of offensive timing."
The next slide showed Honinbou Shusaku across from another player, a wooden Go board between them.
"He went on to win nineteen straight castle games, which were official games held before the shogun and defined the reputation of the Go houses at the time. You could say they were the Super Bowls of the Edo era."
Light chuckles from the audience. The presenter clicked into the next slide, which depicted a single point mushrooming into hundreds, thousands, millions of white branches.
"Even today, many claim that Shusaku was the closest to possessing the 'Hand of God.' By 'Hand of God,' I mean making the perfect move. If you took a 19x19 board and extended the game tree to every possible sequence of moves, the perfect move would be the one that leads to the most wins and fewest losses. The optimal move, you could say."
The next slide showed checkers and chess matches taking place between humans and computer consoles.
"All of you know that computers have already vanquished checkers. Chinook, a checkers engine developed at the University of Alberta by Dr. Jonathan Schaeffer, plays checkers perfectly."
He placed a hard emphasis on "perfectly" as if to knock it out of the sentence.
"In chess, which is a bit more complex than checkers, computers have already claimed dominance over human grandmasters under tournament conditions. A quick glance at their history will enlighten you. Garry Kasparov, humanity's prime delegate against AI chess, was defeated by Deep Blue in a six game match in 1997. Nineteen ninety-seven! That's five years ago! Since then, chess engines have only gotten better. While our current chess grandmasters barely scrape past an Elo rating of 2800, the best chess engines have already broken the 3000 mark!"
An audible gasp from the audience. The next slide was of a simple Go board in Yose.
"But of course, the Holy Grail is to solve Go. Unlike chess, Go only grows in complexity as the game progresses. Furthermore, a move made in the beginning could have direct implications on the board position dozens, maybe hundreds of moves in the future. This is relevant in chess, but the effect is much more pronounced in Go.
"My research has produced an admittedly complex but also beautiful set of heuristics that have been the basis for the Shusaku engine. It was developed in 2000 and has undergone countless revisions to sharpen its tactical and positional abilities. We have tested its strength against numerous skilled amateurs and a few Western players on the professional circuit. As of today, this program has yet to be defeated in a Go match under standard tournament settings."
The crowd erupted in spontaneous applause. The presenter wiped his forehead with a damp handkerchief and took a sip of water. The canyons on his face formed and reformed under the fiery lights.
"That is why today," he proclaimed over the still-cheering crowd, "I issue a challenge to the world: with each side given exactly 24 hours to make a move, defeat Shusaku and claim the Hand of God for yourself!"
The cheers loudened and then loudened still. The soundproof walls quivered. It was a rock concert gone academic. Not surprisingly, no one noticed Kouyou Touya rise from his seat amidst the din and navigate past the forest of cameras to the back exits. The doorkeepers informed him that he wouldn't be able to come back in once he left. The Go master nodded, the men shrugged, and then opened the doors for him. Kouyou walked past a crowd in the lobby that applauded as it watched the presentation on a mammoth TV screen fixed on the atrium wall. From here, the commotion inside the lecture hall came to the Meijin as a jet engine wrapped in thin foam.