The European Go Congress (EGC) is by far the largest Go event in Europe. It has been held yearly since 1957, usually in a different country every year. In 2017, the EGC came to Germany once more, organized by the DGoB, the German Go Federation. The venue was located in Oberhof in Thuringia in the midst of a beautiful landscape best known for winter sports, especially ski jumping and biathlon.
The main parts of the EGC are the European Go Championship tournament and an open main tournament. These two principal competitions are flanked by numerous smaller tournaments and side events, e.g. lectures by Go professionals. The EGC 2017 was the largest in history with well over 1.000 attendees. As codecentric was sponsoring this year’s EGC, in particular the Rengo tournament (more about this later), I attended as official representative of the sponsor. My codecentric colleague Thomas Jaspers was at Oberhof, too, and has written down his own impressions of the EGC in this blog post – German only.
Where now, one might ask, is the connection to codecentric and the world of IT? Well, Go and IT are sort of natural companions. For both, one needs clear, logical and combinatorial thinking and discipline. Thus it is not surprising that many Go players work in IT. Add to this that Go used to be a supreme challenge for AI. Just two years ago the seemingly unshakable truth was human superiority over AI in Go. My old boss at the University of Jena, Prof. Ingo Althöfer, had the idea to document the progress of computer Go to a professional level – expected to be long and slow back then – by means of a yearly competition. Ingo suggested a best of five matchup between strong German amateur players and powerful Go programs and asked if codecentric would sponsor such an event. Thus, in 2014, the codecentric Go challenge was born.
But DeepMind aka Google had other, more immediate plans for computer Go. In spring 2016, DeepMind’s program AlphaGo smashed old certainties about human predominance in Go to bits and pieces with its victory over 9 dan professional and Go legend Lee Sedol. Since this historic event nothing is as it was before in the world of Go. The Go-world-shattering hand of AlphaGo permeated the EGC 2017 thoroughly. Books, lectures, talks – AlphaGo was omnipresent. The organizers even managed to win Fan Hui to give two talks about the games of AlphaGo. Fan Hui is a multiple European Go champion, the first victim of AlphaGo in fall 2015 and now a representative for DeepMind.
But the EGC 2017 demonstrated another, much more profound influence of AlphaGo. It changed the way Go players – amateurs and professionals alike – think about their game. Literally centuries of knowledge what constitutes good and bad play went straight to the trash can. But in many cases the moves of AlphaGo are simply beyond human comprehension. Even professionals just look look baffled at them and can only utter “I don’t understand this”. One could witness this countless times in Oberhof.
But not understanding why AlphaGo’s moves work does not prevent human players from trying and experimenting with them. The “AlphaGo style” of playing Go has conquered the Go boards everywhere and Oberhof was no exception. Moves that would have made strong players cringe or been attributed to bloody amateurs just two years ago were seen at all levels.
In the wake of AlphaGo, two exhibition matches between strong computer Go programs and professionals were played at Oberhof. They were organized by Prof. Althöfer, and it was Ingo, too, who asked codecentric to sponsor the EGC in Oberhof. codecentric decided to sponsor the Rengo tournament of the EGC and thus I got my ticket to visit another European Go Congress.
Rengo is a team variant of Go, in this case three players per team. They take turns in making their moves and are not allowed to discuss their plans or communicate in any way except with their moves on the Go board. Naturally, this leads to a completely different type of game. The Go skills of single players are still important, but to cooperate successfully as a team, other faculties are needed, too. As the Rengo tournament of EGC 2017 was played with full handicap (adjusted by certain bonus and minus points), all teams had in principle the same chance to win the competition. Being in harmony as a team was the deciding factor, not mere playing strength.
Fitting to the size of the EGC itself, the Rengo tournament with 69 participating teams also was the biggest one anyone could remember. Therefore the five planned rounds did not suffice to determine a winner. A final, sixth game between the two teams, KLT and SAT, with 5 wins and no losses was necessary. Team KLT, consisting of a 7 dan, a 5 dan and a 1 dan player, had a 4 dan rating on average and had to give three handicap stones to 1 Kyu rated team SAT, consisting of a 5 dan professional, a 7 dan and a 15 kyu.
This final game was fascinating to watch. It was not the professional player Zhang Tao, who by the way won the open main tournament of the EGC, or the other high dan players who dominated this final match. Instead, it revolved around the weakest player, Team SAT’s 15 Kyu Anna Razymova from Ukraine. Team KLT tried to exploit her weakness by posing difficult problems for her on the Go board to solve when it was her turn to play. Her SAT team mates, on the other hand, tried to keep the game simple and calm. Sometimes they even sacrificed something by playing actually bad forcing moves just not to give team KLT the time to pose hard problems to Anna. In the end, team SAT’s strategy paid off as team KLT was unable to overcome the three stone handicap.
At the end of the game I raised to applaud both teams. As representative of the sponsor I also had the privilege to hand over the cups and medals at the central prize-giving ceremony.
I hope that codecentric’s commitment to Go will continue for many years to come. Nevertheless its format needs to be revised: See you at the next codecentric Go challenge!