I know you’re working on the beta site with a large goal to get the rankings more meaningful.
I was thinking that maybe things would improve on the non-beta site if the range of ranks was increased. For example, let it extend down to 40k or whatever. And leave the default 30k newcomer alone (or change it to 20). The numbers would be deflated compared to AGA, but at least if I was ranked 35 kyu I’d know where I stood in relation to another 35 kyu or a 36 kyu.
Would this be very difficult to implement?
we definitely want to change something with the enrollment of the new update. It’s something similar to what you suggested: Established players will start at 20k, while only total beginners will start at 30k.
You can already test it on the Beta site, and it will be the normal system on the non-beta site soon!
Feedback is, of course, appreciated.
would be nice if kyu’s from 30 to 25 level up ones per win (level down from 3 or 4 lost) and after 25 kyu your ELO/GOR system would kick up.
if any of you have additional comments and feedback about the ranking system of the Beta, just post them here in this thread, so we don’t have them spread over many topics.
Thanks. I’m looking forward to it.
One of the problems with KGS is that the addition of a 9 Dan player does not change everyone else’s ranks.
I would like to see something elastic.
A 7 Dan with however many rating points would change the scale of what it takes to get the next rank.
I don’t understand why Rank is so important. Isn’t Rating a superior metric? I started the group Low-Rating 30k’s specifically because “30k” does not have the fidelity to accurately describe a player’s skill level. There is a whole world within “30k”. Why can’t matchmaking be performed based on Rating instead of Rank?
I guess I could also mention: I play chess at chess.com and there it’s all Rating.
Hi Therms. I see rank as being a scaled, truncated, and shifted version of rating (ignoring what happens around 0):
rank = round(a*ELO+b);
Of course the problem here is that there is a lower limit:
rank = limit( round(a*ELO+b), -30,+9)
So a huge range of ELO gets bunched into the -30 rank. I say, “Get rid of the limit.” Let rank go to -50 if it needs to. Then handicaps will be appropriate if you play anyone within 9 stones of your rank.
So I’m with you on rating being more important (and having more information in it). But rank tells you how many handicap stones to use. Unless your rank is 30, then it’s not so helpful.
Ah. Handicap stones. I endorse your proposal completely.
On Beta I registered as 1d then won against 19k, lost to 2k and become 2d. So I think something is wrong here.
Still Beta: Some days ago, I also won (luckily, by time) against AQJG in game 85, and it seems to me like AQJG ranked up from 5d to 6d in conclusion. But it seems like my rating increased correctly in both cases.
Still Beta: My opponent lost by time, and (apparently) ranked up from 10k to 9k. I think this is a very funny bug, you should think about keeping it
Addition: This time, it seems, I didn’t gain any rating from the game, even though it was rated, and not annulled. I talk about game 135.
Meanwhile I noticed that it’s valid that I didn’t gain rating since my opponent has a provisional rating.
When will the update occur?
yesterday I offered some open challenges for handicap games. One of my opponents is ranked 30 kyu. He’s new to the site and seems to be a nice guy. He told me his real life rank used to be 5 dan amateur but he hasn’t played for a while. So I didn’t exactly get what I bargained for!