Are OGS rankings inflated, deflated or neither?

Thay was not the conclusion of the thread on this opened before.
It takes like 6-8 games to get some wrong rating and like 12-15 to get a beginner rating (like 20-25k)

1 Like

@Groin:

I believe the thread you’re talking about was before the provisional rank was changed back to 12 kyu.

EDIT: I was wrong, the true provisional rank is 6 kyu: explanation

Anyone can update this thread if necessary.

It seems I did misremember a little. Assuming that all of the rank changes were applied retroactively, and cautioning that these are “actual ranks” rather than displayed ranks, it took me one game to drop from 12k to 15k, and then another thirteen games to drop to my lowest rating, 24k.

I don’t think this is unreasonable at all, especially considering that one of the first things you’re told as a go player is to “lose your first 100 games as fast as possible”. Plus, the displayed “humble ranks” would have been even lower, right?

You started way back in 2017! That’s at least two rating systems ago! :laughing:

As I’ve said, this is assuming that all rank changes were applied retroactively, which I’m pretty sure they were? It does line up with Groin’s assertion that it takes twelve to fifteen games to attain a “beginner rank”.

Yes, but the retroactive calculations are not reliable, especially because the rating updates still reflect the old system :+1:

Fair enough, I suppose. In any case, I don’t think losing 12 - 15 games to drop to an absolute beginner rank is unreasonably “humiliating”.

It’s especially not welcoming to have to be crushed 15 times to be accepted as a rated beginner. Note that is not the only problem as if you just want to play with other beginners the system will refuse it as you start too high ranked.

2 Likes

Hmmm. I have to say 12-15 would start to worry me for the beginners. Hopefully I’m right about the current system being faster than that, though.

Edit: after a quick search, this player is the worst counter-example I could find to my hypothesis. They reach 25 kyu (645 rating) after 2-3 games, but their true rank is way below at about 212 rating, and it takes them, I guess about 20 games to reach it.

How come? If you’re an absolute beginner like I was, you will be crushed many, many more times after attaining your “true beginner rank” before you win your first game(s).

The only thing that was demoralizing about it for me at the time was that my rank was displayed as “25k” for more than 150 games, and I wasn’t sure if I would ever reach 24k. Thankfully, the newest rank update has completely solved this issue.

This seems impossible. I haven’t looked at your link yet, but the lowest I ever dropped as someone who knew NOTHING about go was 640, which is about the same I’ve observed with other newbies. If someone’s rating drops below that, I’m pretty sure there must be something funky going on.

EDIT: I guess that link looks kosher, actually


Well, people have different talents. Go beginners come in all shapes and sizes. Some Go beginners surprise me by playing moves with purpose and forethought, some play mostly random moves, like playing out a lost ladder even though they know about ladders.

Well, I talked about my first experience on a Go server. Though thinking back, I’d say it might have been a psychological experience caused by the presence of the rating itself.

As a beginner, I was assigned a specific rating, and without knowing anything about rating systems, I just assumed “oh, this is an appropriate rating for someone starting out!”. So I kind of immediately adopted the rating as part of my identity. Seeing it drop game after game, realizing how much better the players I was being matched with were, was quite the jarring experience.

So my conclusion is that a) taking too much to correctly rank a beginner can be a problem if they’re not educated on what to expect, and b) seeing the rating from the start is potentially very bad for the psychology of their first experience.

1 Like

I won 5 out of my first 47 games. Unless I’m a complete exception to the rule, I wouldn’t call this proverb a myth (unless you take it literally).

1 Like

If they know about ladders, IMO they’re not a complete beginner. :stuck_out_tongue:

On a similar note, I still remember how devastated I was when I learned that my opponent was allowed to capture my group even if it had an eye


1 Like

Allerleirauh has created his own strawman myth. The saying about losing 100 (or 50) games as fast as you can is not about how long it takes to win. It is about encouraging beginners to play a lot of games quickly, even if they lose them all, because experience is the best teacher. We might well dispute whether that kind of experience is best, but it has nothing to do with how long it takes to win, and no one I know ever imagined it did.

1 Like

I was showing that beginners are allowed to have ranked games (restricted to ±9 ranks) with 1d players. So something about matchmaking assumes they are at 6k; it’s not just a display bug.

2 Likes

It was just pointed out to me in the other thread that the uncertainty in the provisional rank is apparently one-sided. 11.9 kyu is actually the lower bound of one deviation.

So I was wrong, in some sense there is a “hidden” true provisional rank at 6 kyu, but the graph displays it at 11.9 kyu artificially - as a prov player loses against a 1d, their true rank slightly lowers, and the deviation shortens, making their graphical rank increase.

My freaking God. That’s REALLY weird.

2 Likes

Well, you see in places people say that it’s ok for complete beginners to lose a lot. That it’s fine to start them at 6k or whatever because, oh, they’re just gonna lose a bunch and it’s fine and so on. That it’s fine to have them lose 9 out of 10 games as Maharani because, oh, well, they’re beginners. I don’t believe it. I think it’s best to pair them with other beginners so they have a fair chance at a game and expected to win 50%. And although I don’t remember exactly, I had an impression some teachers recommend to let the beginner win their first 9x9 game.

I mean why are we pretending that losing 9 out of 10 games isn’t a discouraging experience. Do you want go playerbase to consist only of masochists?

5 Likes

We could make this one for, hmmm, like players who end up lower than 20k if there’re any?