![[Welcome to Sensei's Library!]](images/stone-hello.png "Sensei's Library")
Frarugby: This question has probably been asked a thousand times, but i'm new to this wonderful site, give me a break. :) I've been playing the great game of Go for a while now, but for the most part my games have been limited to the internet. I intend to join a club in the future, but currently i personally don't know anyone else who plays seriously. Is there, somewhere on the internet, a program or person or site that can analyze/look at my game and give me an idea of what my rank is? i'm very curious and i have no idea how to find out approximately where i'm at. Often when i play against others, they tell me they are 15 kyu, or 18 kyu or 5 kyu or so on.. are they estimating? how does one find out? thanks a lot
(moved from QuickQuestions)
HolIgor: I guess you've been playing on some internet server without a rating system. Try to create an account at a server with the rating system such as IGS, KGS, NNGS. After playing a number of games you'll get your rating. They are quite close to the real life ratings. IGS gives approximately European rating, KGS is anchored to the American rating system.
Charles Without any data to go on, I can say that 15 kyu is the top end of the 'social player' range: without a bit of serious study, or contact with more experienced players, most people find it hard to progress further than that.
IanDavis You can also try the Osmotic method - I have begun to notice several players now who don't actually play; they just watch high dan games for months, then they start playing at about 15k. Could just be urban myth though.
(moved from QuickQuestions)
Just some thoughts on rankings.
If rank is determined solely by comparing your results against ranked players, wouldn't the skill represented by a given rank 'drift' over time in a kind of random walk? Which suggests the question: Is a 1 dan from 1900 considerably different ( better/worse ) because of this than a 1 dan today? And, if by game analysis we see that yes, 1 dan today is far better than a 1 dan of 1900, doesn't that mean the idea of 1 dan is a bit vague? In the physical martial arts, ranking varies across clubs, but there also exist specific sets of skills which must be demonstrated to advance. ( Currently, go rankings are like saying a green belt is simply a guy who can beat up a yellow belt. Nothing wrong with that except that the idea "green belt" has meaning only for the members of a particular dojo at a particular time. )
If anyone wanted to associate quantifiable absolute skill levels with rank levels, perhaps this would be a good task for go computer programs, which don't get better or worse as they practice, and will be around forever. A body of experts might assign 'AI rankings' such as associating 10k with being able to consistently "beat GNU GO 3.6 at Level 5", or whatever. This might provide an objective measure of ability across servers, clubs, countries, and decades. It would only work for kyu levels, given the state of computing, but that's probably the people who might want such a system.
I guess the drawback would be that the programs maybe aren't creative enough - people might train against the particular handful of tricks that tend to beat that computer program, instead of developing well rounded skills.
Tim
It took me a weekend to get used to GNU go's style. The first few games it was consistently beating me by 20-30 points, but about 10 games later I had worked out which moves it could answer and which it could not.
I didn't study it per se, I just played it for a bit.
The problem with current computer go is that it doesn't play all parts of its game at the same level. I guess GNU go was about 8-10k on average when I last played it, but some of its moves were taken from a database of 9p moves whereas others would have made a 15k laugh.