On kyudan.net, you are shown a partial game record of dan players. Your task is to judge their level from their moves. Once you submit your estimation, you are presented with their actual level.
This site may be interesting to Go players who would like an unbiased opinion on their progress in understanding the game, as displayed by judging the skill level of other players. If you often get it right or very close, you probably have a good grasp of the situations that occur in the record, and were likely able to spot the players' mistakes.
This kind of excercise is probably a tad difficult to come by for most people, as there are, to my knowledge, no game records available in which the player level is only displayed at the end of the game. Alternatively, your eyes might unintentionally betray you by glancing (even unconsciously) at the displayed ranks, which would skew the results.
As for the idea itself - that came from my experience on the KGS go server, and the very particular 'kibitz culture' it features. It was surprising to me how some observers were (no offense intended) more than ten stones weaker than the players, yet appeared to be able to judge accurately that the players could only be '7 dan, but not 8 dan'.
Could they have gotten it right? This question should not be dismissed too quickly. Even weaker players have an understanding of (local or global) consequences of certain actions, and though they may not always find the right moves themselves, they should be able to tell good from bad results.
It is unclear how this ability is distributed among players of various skill levels (can an 8 dan player always tell a 2 dan from a 3 dan?), if it is equally inherent in everyone or if some players are distinctively more accurate than their peers with similar skill level (which would indicate a very different individual thinking process). Depending on preliminary results from analysis of the votes cast, it may be possible to hypothesize a formula linking the involved variables together.
The kifu were extracted from public, license-free sources and with permission from all involved Go servers and administrators. The games were completely anonymized to protect the players.
| Rank (dan) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|---|---|
| Number of games | ? | ? | ? | ? | ? | ? | ? | ? | ? |
Note: The rank distribution of the games shown to be voted on is roughly even, i.e. not based on the distribution of ranks in the database. That means you won't be shown tons of 4 dan games just because the database contains many 4 dan games.
Right now, the database only uses KGS games and the according KGS ranks. I am aware that this is only accurate within a certain margin. It should be close enough though - I manually checked various games (by confirming that the ranks of the players were solid), and found that a very large fraction of them were using appropriate ranks.
Anyway, I am currently working on providing a much better source. This will take several months. I will announce when the data source changes.
Depending on the feedback by the Go community, I may further improve this site and/or add new features, content, participation possibilities and analysis tools for you to enjoy.
This section got large, so I moved it to a separate page. Read the privacy statement.
The server is 100% powered by green renewable energy sources, consisting mainly of Scandinavian water power plants.
I'm Sebastian Heuchler, a German CS student. Creating kyudan.net in early 2016 was mainly an attempt to get my hands dirty with 'web programming'. As this was the first website I created from scratch, I had to learn a lot: