I think the 13k default rank is doing harm


#201

Since anoek is back and active (yay! :slight_smile: ) and humble rank seems to be instated (did I get that right?), I’ll shamelessly use this opportunity to talk about MY experience as a new user for a moment and what MY biggest challenge to enjoyment of the game was.

I joined OGS in August 2017 with no prior knowledge of the game other than “stones with no liberties get taken off the board”. Being matched to players of equal strength was never an issue for me. I was beat down to 25k in no time at all - five games at the most. (Unfortunately, my games history only seems to go back to September for some reason, and the rank progression graph is severely bugged for me, so I have no exact numbers here.)

The lowest my Glicko rating dropped to was 623 +/- 82. OGS converted this to 25k. I played and played for about a month, around 120 - 150 games, before my rank finally became 24k (this happens around a Glicko rating of 1030). For a month, I was left in despair because I had no idea at what point I would finally “escape” 25k. I had no sense of progress and wondered if I was improving at all. I definitely came close to quitting at some point.

This was due to OGS’ arbitrary rank limit of 25k. If the Glicko conversion formula was applied consistently at the lower levels, my initial rank of 623 would have converted to 40k and I would have been able to observe the gradual rise of my rank from 40k through 24k. Even at these low skill levels, the Glicko deviation works so well that wild rank variance from game to game would not have been an issue.

Hence my suggestion to remove the arbitrary 25k limit and let the Glicko conversion formula do what it does best - relevant thread here :slight_smile: Please (re)instate ranks beyond 25k and 9d

(It seems that since the last post in that thread, players stronger than 9d are now marked 9d+, which currently seems to concern less than ten accounts, anyway, so feel free to skip over that part of my proposal.)


#202

Looking at the OGS rank histogram (Thanks @S_Alexander) there may be a case for extending the lower limit to 30k.


#203

With a lowest rank of 30k, I would have been stuck in limbo with no sign of progress for only about half a month or ~80 games :slight_smile:

Seriously - it seems the lowest rank one can be is around 41k. Why have an arbitrary cap at all, whether it be at 25k or 30k?


#204

I think the idea was that a handicap stone did not mean much in those ranges, or something along those lines


#205

It’s not realistic. I doubt the 600-rated players would lose to regular 25k players 50% of the time when receiving 16 stone handicaps.

The whole 1-stone-per-rank thing is already quite frail, but I also think it’s better to have some beginners be a little frustrated they don’t get out of 25k than to have them be really frustrated they’re at 41k when other beginners are 25k.

Lastly, it would make matchmaking a lot worse due to the 9 stones limit. They’d never find a match. :stuck_out_tongue:


#206

It’s funny, I never really think about handicap since, as you’ve stated, one-stone-per-rank probably isn’t accurate until you’re like 15k or something - especially since one-stone-per-rank was only ever meant for 19 x 19, and most beginners stick to 9 x 9 and maybe a bit of 13 x 13 for a while. I doubt that more than a small percentage of players think of their rank compared to the rank of others as a “difference in handicap stones”.

I can only speak for myself, but even when you’re told you’re 25k, it’s already pretty overwhelming to think that there are 24 more kyu ranks before you become a “strong” amateur - much more so if you can’t even get out of 25k for over 100 games. At any rate, as a 40k, why would you compare yourself specifically to 25k players to begin with, once we’ve done away with the arbitrary cap? :wink:

As I’ve stated in the relevant thread, matchmaking already treats you as a 40k for this purpose, it’s just not displayed that way. You won’t get matched against “true” 25k players until you’re 34k (within nine “stones”). Matchmaking is based around your Glicko rating, not the kyu/dan rank converted from that.


#207

I agree with the fact that people should not be auto placed into 13k.

My idea (a combination of many other ideas) is that a person can choose 25k (beginner), 15k (intermediate), and 5k (advanced). In addition, there will be a very high uncertainty, so each game played changes your rating by a factor much higher than regular. Also, unrated games do not change an opponent’s rating as much as usual.

For instance:

  • 5 dahn chooses 5k as their option, plays and wins 5 games, and is placed in 4 dahn
  • 25 kyu chooses beginner, as they have never played before
  • person trying to inflate their rating chooses 5 kyu, loses 8 games, and gains a rating of 23 kyu
  • person tries to underestimate their rating, chooses 25 kyu, wins 8 games, gets placed 13 kyu
  • 15 kyu person choose 15 kyu, wins 3 games out of seven, gets 16

#208

Automatch is one of four ways to get a ranked game and it would arguably be the slowest. Also, the 9 stones rule does evidently not apply to 25k, otherwise you could not have played this ranked game. Your rating was 652 (39k), your opponent’s rating was 1228 (19k) - a ‘20 stone’ difference. :3 Or are you saying you just wouldn’t have been automatched with that player? Same conclusion holds: you would not have been able to play that game. ^^

I believe it’s useful to have an upper bound - just like these guys:
30k - KGS, China
25k - (OGS,) AGA, Japan
20k - EGF
18k - Tygem
17k - IGS (BC for provisional newbies)

Let people have some tangible idea of where they stand. That, or explain to them the reasoning behind the rating to rank conversion formula. :slight_smile:


#209

Huh - I guess I was mistaken about this. I vividly seem to remember seeing ranked game offers by other 25k and not being able to accept them, but I suppose the causes of that lay elsewhere…

Regardless, I wonder how hard it would really be to find ranked games against players within nine “stones” even at 40k.


#210

Let us assume that 2000 users are active at any given time (based on the figure in chat) and 10% are playing a game or looking for one (there are currently 86 live games going on). Let us further assume that the bottom 0.4% of those (which 41k-32k would clearly fall into) are super-eager to play, then we already have 0.8 people about to duke it out. :stuck_out_tongue:


#211

Hmmm. If my assumption about matchmaking using Glicko rather than displayed kyu rank was wrong, and finding games would really be an issue for the weakest players, I humbly retract my suggestion. Maybe a cap at 30k might be a good compromise, then?


#212

You wouldn’t want to play anyone stronger than 20k anyway, so that seems like a good place to settle.


#213

Here’s an interesting problem. If we going to lower starting rating it would be harder for newcomers to find games. Because there’re fewer 20k players than 13k.


#214

Quick match finder defaults to ±3. As 30k you can wait forever.


#215

Expand the quick match +/- in the TPK range?


#216

He’s talking about the defaults.


#217

Who’s “he”? I was talking about the defaults, yes.


#218

:point_up:
.

:slight_smile:


#219

(retry)

How about expanding the default range for TPKs?


#220

For instance;

    if player.rank <= 25k;
       default match.finder.range = ±5