I think the 13k default rank is doing harm [Closed]


#101

@Tofoon That makes no sense. New (i.e. provisional) players would still be 13k glicko, but “29k” - for the sake of argument - decorated. Non-provisional players are (essentially) those whose rating uncertainty has dropped to acceptable level, say, 100 points).

Provisional players who turn out to be actual newbies would both only see their humble rank - and they would never get (auto-)matched with actual 13k opponents as long as their humble rank isn’t even close to 13k.

Jade’s misunderstanding probably stems from the fact that humble rank only makes sense for provisional player matchmaking, not for everyone. We would merely stress the fact that a provisional rank is extremely volatile. No one claimed the established ranks were similarly volatile. For new accounts, we have no match history to base an estimate on. However, to apply higher uncertainty to established players with a history of matches to base their rating on makes no sense.

In the form of a short play:

Glickman to Newbie: "Hi. I don’t know you, for now you’re labeled average. If you aren’t, you’ll find out soon which kind of not average you belong to."
Humbledore to Newbie: “Hi. I don’t know you, but you’re probably not much worse than these people.”

Glickman to Regular: "Good to have you back. All in all, this is how you’ve performed until now compared to your peers."
Humbledore to Regular: “I was just about to say that. You improved so quickly, too, you must be talented!”

Newbie to Regular: "Wow, I just got here. I am not this bad!"
Regular to Newbie: “Prove it. Jubango, you and me, now.”

Newbie1 to Newbie2: "Okay, I guess I am this bad. But you are about my level, want to play?"
Newbie2 to Newbie1: “Sorry, I just got here, too. From another town. I’m pretty good actually.”


#102

LOL one of us is completely misunderstanding.

I believe that Tofoon’s propsal is that we always use the humble rank for everything. The glicko rank just becomes the calculation that leads to the humble rank. When uncertainty is low enough, they are almost indistinguishable.

It’s elegant.

GaJ


#103

Newbie1 to Newbie2: “Your reputation does not precede you. Let’s see just how good you are, by playing… [dramatic pause] a game of Go!

*** A couple of games later… ***

Glickman to Newbie2: “Indeed you are pretty good! You have now achieved a reputation for skilful play. I acknowledge this by awarding you a suitable rank.”


I realise I made a couple more assumptions:

  • Most new players would prefer to win against several actually-weaker opponents and see their rank increase, rather than lose to several actually-stronger opponents and see their rank reduce.

  • Strong players are happy to prove it.


#104

I didn’t read all the above posts, so if this has already been said please bear with me.

Has anyone considered the possibility that new and/or clueless players might not abandon the game because of a few losses due to less-than-accurate rating, but because Go at the base level of understanding just isn’t much fun?

I honestly didn’t enjoy my first couple games before I learned about making territory, life and death, etc.
Fortunately I was motivated to keep digging and the work I’ve done has been rewarding. But I bet there are plenty of people who ditch it after a game or two of monotonous stone placing without any strategy or purpose.


#105

Yea, it’s been mentioned. The main problem is the same: It’s hard to get hold of quitters to interview about their reasons, because they aren’t around anymore.


#106

True that. Now everyone please forgive me if this seems impertinent but does the reason for their departure really matter? I mean ultimately the quitters gonna quit, players gonna play.

And as far as the ratings thing goes, I noted something similar on chess.com when I first enrolled years ago. Got my butt whooped, and did some whooping as well until the score settled after a handful of games.

Find your place either above or below average and you’re golden. But passing through the standard entry level is very volatile, and that’s just how it is.


#107

It appears that there are overwhelming number of negative responses, and yet we are told this is a better system. Perhaps it works well from a mathematics point of view, but obviously the emotional response is negative.

I will no longer recommend OGS as a starting server for newbies who barely know the rules. I do not wish to unknowingly play a 19x19 game against someone who does not recognize atari. Nor do I believe they wish to be slaughtered.

I have noted playing auto challenges, more ? marks and some of these people play strong, and sometimes it is unbearable.

Definitely not a winning combination. I am looking forward to being able to play on KGS when I do not have to download java.


#108

¨quitters gonna quit players gonna play¨

I do not think this is necessarily true. There are people exploring GO who have varied reasons and committment to learning the game. There is no advantage in leaving a bad taste immediately in somebodies mouth. There are so many options with what to do with ones time. Many new players would like to play new players not get slaughtered. And not everyone will take into account how the ranking system works. Not having an option to give a preset of your rank is a very poor approach to new users.


#109

Congratulations on the New baby!!!


#110

Fair enough. I understand and agree with the idea that a rating preset system could improve rating accuracy and let new users skip the butt-whooping sessions before they get automatically paired with users on their level.

I also think that there are plenty of reasons for noob turnover beyond initial rating though. And honestly, I haven’t felt anything but nurtured since I started here. So I guess what I don’t understand is the idea that OGS is somehow inhospitable for new players, or “doing harm” to those trying to find their place. It’s a great environment, ratings be d@mned.


#111

Have we thrown aside GaJ’s statement? I still find it a good solution as it deals with new high level players and low level players:

To repost:

Now then, Glicko does NOT require the initial assumption that all players need to start at an average rating, but it is more accurate if it does: which makes this solution all the more beautiful.


#112

To be honest, this discussion doesn’t start from

“Gee we have high noob turnover I wonder why?”.

It starts from “wow, I just went through that, and it sucks, that can’t be good”.

That being said, overall OGS is a great environment - the best place to play - and this conversation is about making it the best place it can be.


#113

Understood.


#114

Yeah. I really like tofoon’s simple and elegant suggestion, but I doubt that it takes more than half a dozen lost games to bump you down from 13k to 25k in the first place (compare this to the adage of “lose your first FIFTY games as fast as possible”). And as someone who’s 13k herself right now, I actually enjoy playing against absolute beginners and trying to do my best to explain and teach the game to them (I play very passively in those instances). So I feel that the problem is vastly overstated, but I think tofoon’s suggestion, if it can be implemented, would be a beautiful solution.


#115

As a 13k I HATED playing against new players… I just did it out of principle so that they wouldn’t end up rising for some reason… I wouldn’t show any mercy either (and I know I’m terrible for that, but I wasn’t (and still am not completely) sure how to actually do teaching go. It felt like a waste of time and a ratings spike that would only get dropped later.


#116

But …you only addressed one part of a problem with at least four bullets in the list.

I recently signed up a new account (to play and rank live separately from correspondence) and the first games I got on automatch were cancelled out because the person “doesn’t play beginners”.

That sucked. That’s the experience that new middle-DDKs get when they come here. And that is only one more out of the bullet list of current problems.

But - I’m glad we’re only debating whether the problem is “vastly overstated” or not :slight_smile: At least there’s good concensus that there are problems, and more than one candidate solution that have legs!

GaJ


#117

Pls forgive the [off-topic]

Oh, has s/he arrived already? Can we have a photo, @anoek? Wishing you and family all the best!


#118

Lee Hajin is 13k at OGS. Isn’t it a little bit inpolite?


#119

Last I checked (just now) she’s 4p OGS despite having retired.


#120

I just remembered playing Dominion on another website which also uses Glicko-2 as their ranking engine. Interestingly they do use an equivalent of what is called ‘humble rank’ in this thread (without calling it humble) and it works pretty well:
http://forum.shuffleit.nl/index.php?topic=1679.msg5891#msg5891

I’d say that climbing up the ranks in there was quite rewarding, although it did have a little of ‘sandbaggy’ feel in the process :wink: