Worldwide Rank Comparison List

[karaklis]

I have created a new rank comparison list that I'd like to discuss. The major differences to the established rank comparison list is that it uses intervals instead of single ranks and that it uses clearly defined sources.

I thought it was necessary to create a new one because I heard many voices (mainly from the German go forum) that said that the old comparison list was quite useless. One interesting finding when creating the new list was that there is a very high bandwith when comparing the ranks of different servers/organizations.

Unfortunately I have no information about the national go associations from China, Japan and Korea. Do you have any information how to obtain and add these data?

Any commments are welcome.

[schilds]

Perhaps the kyu ranges could also go from weaker to stronger, so instead of 17 - 23k, 23 - 17k.

[daal]

Interesting idea, and nice work. It seems on the one hand to be more correct than the old system, in that it produces less wrong information, but on the other hand it appears to increase the fuzziness of the result because a single known rank (such as my KGS 6k) falls into several (in this case 5) ranges.

It might be nice to see the chart in the form of a program in which one could input his known rank on a given server, and see the corresponding range on another server as the output.

[Kirby]

...
It might be nice to see the chart in the form of a program in which one could input his known rank on a given server, and see the corresponding range on another server as the output.

I think that it would be difficult to get accurate results from this. For many ranking systems, for a non-improving player, playing more games means that your rank will converge to a particular number, but in reality, this isn't the case for many people. To name a few discrepancies, some rank systems are much more volatile than others, some depend on future results of your previous opponents (and some don't), and systems are affected by wins or losses at different rates. You can add in the fact that some players are actively studying, and their "real" strength is more volatile than somebody playing casually, and it becomes very difficult to compare different ranking systems.

One of the merits of the KGS ranking system is that it is somewhat stable in terms of strength to rank. It's not always the case, but usually if you play two different 1d players, you can feel a similar amount of strength (although, this might be psychological). On some other servers, this isn't the case. When this happens, it's hard to even have a real comparison of ranks, even on the same server.