Entry point is (and will stay) at 1500±350 (12.3k). The high deviation makes the drop fast. There is already a length thread about the entry point somewhere.
3 Likes
Ok, but I’m still wondering if people seeing their rank drop fast, and constantly, is more of a deterrent to new players…
Again, I genuinely don’t know if new players are sticking around, how they feel about it, etc. I genuinely do not care about the admin side of stuff, what glicko is or does, etc. I just want to know what new / beginner level players feel. I’m not asking to change anything, nor am I making recommendations.
I only want to know how new players feel, and if there is any data on how long new players stick around if they lose a lot of their first games, if there is a correlation between those that win a lot of their first few games and those that keep playing, etc.
1 Like
Correct, however, this is solved by what you said earlier…
Since this huge, fast fall in rank happens during the player’s provisional games and the “starting rank” of 25k isn’t revealed to the player until the [?] is removed, there theoretically shouldn’t be any difference in the experience for new players.
The only difference would be that when the rank is revealed it would be whatever the new classification is and they could see themselves making a bit of progress as they learn.
Edit: That’s my take on it anyway please correct me if I’m wrong
7 Likes
Sounds about right to me.
1 Like
It just occurred to me that the “real ranks” are not more real than the “humble ranks”. Why not call them “mean rank” instead? 
So true! It is not and can never be a perfect system. Some people get too attached.
Yes, straight to the point!
Traditionally, there is no rank below 20k.
It is hopeless to attempt to grade the kind of strategic ideas that form in the head of someone who doesn’t know the basic rules of the game they are playing.
… in the form of Glicko points?
Since kyu/dan are just decoration anyway, the BC label is perfectly sufficient. “Progress within BC” is going from 500 points to 900 points. Anyone who cares enough to look that up will swiftly rise to DDK.
3 Likes
(clapping excitedly) Oh that is awesome! Thank you so much for preparing this version 
I hadn’t even considered that this data was ranked only. Most of the 25K players I meet have profiles with AI games being played before any ranked matches. I imagine that the vast majority of newcomers play most, if not all, of their first games unranked. And it would be in this “testing” phase where they are most likely to quit if they can’t wrap their head around the game.
I wonder if there is a way to try and model the retention rate of new players. Or what the average number of games played is before an account ceases to be active anymore. This data could be very useful in coming up with ways to improve player retention.
A lot of [?] players ask when they will get their rank. Adding a tooltip to it that specifies a rank will be determined after 5 to 10 ranked games would be helpful. For the player that doesn’t play ranked games, they will have no idea what progress looks like. When I start a new game, I never hop into ranked play. I wait until I feel I understand how to play the game. This seems to be true for a lot of OGS accounts.
Having some way to know you are making progress, when the rank is the only numerical indicator of progress in Go to be found, definitely seems worth considering. Especially since those playing bots don’t even have a rank to look at 
1 Like
Amybot and GnuGo both accept ranked games. The problem with users not playing ranked games is the same for them. The users simply don’t want to know if they are making progress.
1 Like
I think that is too presumptuous. How did you come to this conclusion?
I certainly did not avoid ranked games because I did not want to know if I was making progress. I avoided ranked because I did not want to have a bunch of losses on my official record that were accumulated during a time I was learning something brand new. Starting off in ranked matches would make no sense, especially if I intended to take playing ranked matches seriously later. Given that new users have no idea how rating actually works, they simply cannot make an informed decision about whether playing ranked games is a good or bad.
Exercising common sense, concerning playing a ranked vs an unranked game, would lead a person to error on the side of caution, instead beginning with unranked. In this case I consider “common sense” to represent a person gauging whether playing ranked vs unranked in any new endeavor is wise. They would consider what the average experience of “ranked” offerings were in the various games that they’ve played in the past and then decide if playing “ranked” (in general) is wise when they do not understand how ranked matches work or what they effect.
At the very least they understand that their losses will be tracked on some sort of official record that is used to evaluate their capabilities as a player. Realizing they are not yet competent in the game of Go, playing a ranked match first seems nonsensical. The only folks who hop right in are those who don’t generally care about ratings, those who are overconfident, or those who are impulsive and don’t give the matter any thought before beginning.
1 Like
I don’t see how this is connected.
I care about ratings. Precise ratings can improve my experience by getting paired with equally strong opponents, makeing the games interesting.
Starting with ranked games right away has also the advantage of setting my starting rank to what it was at the beginning, giving me the possibility to see how I improve (if rating is visible).
The only problem with starting with ranked games right away are bad rating systems, in which you are stuck with your rating even if you improve. But for those, playing ranked is never really a concern.
4 Likes
I think both starting with ranked and unranked make sense.
I started with almost all ranked games. Naturally I don’t remember what I was thinking at that time but maybe I treated ranked as main mode of play, default, and unranked are casual not-serious games. Motto: “I should play raked unless I want to try doing something silly”.
But people who start with unranked also make sense, for them casual unranked games are main mode of play and ranked games is something you play when “you’re ready”.
Probably a big hint that ranked games are normal is the play page.
But maybe players who start with bots simply don’t see it.
I think it would be interesting to watch beginner playing habits. I’ll maybe think about it next week.
9 Likes
I jumped straight into ranked games when I started. But I have also written Glicko2 software successfully so am fully aware of what the ratings mean and do and I know from other games (starcraft 2) that people have an irrational fear of ‘ranked’ games.
Progression is important though so I would be much happier with ratings down to 35 and an improvement. The matchmaking doesn’t have to change.
As an aside on the ranked/unranked one of the best writers about ranking system was Sirlin and he wrote some wonderful articles on ratings years ago (about SC2 mainly). The article linked below has some interesting comments on ranked/unranked games as well so I recommend it to those interested in ranking discussions. It’s only on wayback as far as I can tell now… Just for those fascinated by ranking and Glicko though!!
10 Likes
My comment (quoted above) is a supporting statement of my primary point to your original comment (quoted above first). And my primary point was:
I understood you to be saying that people who do not play ranked games simply do not want to know if they are making progress. I disagree with that statement, believe there is a clear and logical reason why most people would want to avoid starting any new endeavor by playing ranked matches.
As a competitive gamer I understand this notion very well. As someone who cares about their win/loss record, I very much avoid doing anything ranked before I feel competent in any activity. Otherwise, what is the point? Then again, I do not play ranked games constantly. I play when I’m mentally fit, prepared to compete, and within an environment that is distraction free.
Only then can I be assured that my performance is “clean”, which helps me to identify weaknesses with certainty, as well as be assured I am able to manifest the best version of my competitive self. OGS is no different for me. As one who has spent a lot of his life competing in serious competitions and competitive culture, I realize I am the norm for competitors, not an exception.
This is very true. Though rarely do I covet the progress I made when I was learning the basic concepts of a pursuit. I covet progress made once I feel I understand what I am doing. If I am a professional cyclist, I don’t think back to when I was learning to ride my bike with training wheels. I begin caring about my progress once they come off.
In the world of competitive gaming, where you are tracking overall kill/death ratios and number of deaths in general, this early data can be particularly detrimental.
Many rating systems care about statistics. This is a solid reason to avoid filling a record with junk data from your training wheels days. Not all pursuits in life can be beautifully modeled by advanced algorithms. Nor are stat based pursuits inherently shallow or unworthy of being concerned about.
In the end this all comes down to different strokes for different folks. But if we’re talking about the law of averages here, I still believe that a person is better suited to get acquainted with something before they begin pursuing ranked matches.
2 Likes
If the rating system is crap, you have to create new accounts regularly anyways. If they keep using outdated data, they are unable to adjust at any rating.
1 Like
I do not disagree with you. Though, I would consider them primitive or their developers lacking the ability to implement something truly impressive like glicko; versus pure crap. I think crap has zero value. I, personally, would still rather have some form of rating system than none at all. Ratings are one of the ways that I often find fun in activities that otherwise would grow stale or empty.
2 Likes
Is this humble rank, or glicko rating-rank? If it’s humble rank, then the glitch at 19k is newcomers.
Real ranks. Plus, you can still see the notch on the green part which is <100 deviation - humble rank doesn’t change anything there.
1 Like
Green and blue are both totally unaffected by humble rank, humble rank is only active on grey bars.
I quietly suspect it was people who battled their way out of TPK (Twenty Plus Kyu) and called it quits after that (or hit a new wall).
You can see similar rises at 9k and 1d although not enough to fully overcome the downward trajectory of that side of the mean, both of them are much closer to the rank before them than the rank after. (Seems to be another one at 4k?)
That’s very cool 
3 Likes
9 Likes
we seem to have gained a substantial number of players, and apart from a spike at 24k the new graph looks a lot smoother
2 Likes