I reckon AlpaGo has come largely after the new ranks, hasn’t it?
Where before I think there was this concensus that around 9d it pretty much topped out, how could you rank AlphaGo if you can’t go about 9d?
This alone tells me that it’s time to drop the ceiling. Play can be much better than 9d…
GaJ
Or just call AlphaGo 9d. Whoever plays worse is 8d or less.
Afaik AlphaGo has been granted an honorary pro rank.
Right, but it doesn’t make a lot of sense does it?
If we call Alphago 9d (or 9p) then we have to drastically scale everyone down. That’s basically redefining the system.
Whether or not Alphago is “pro” still leaves open the question … is it 9p, or … more?
The point I was making is not so much about whatever rank we give Alphago.
It’s the fact that Alphago made it abundantly clear that the skill levels to far beyond our current pros.
Prior to that we sort-of-conceitedly made a system that assumed that the pros were the pinacle.
Now that we know this isn’t remotely the case, it seems we need a ranking scale that is allowed to go beyond the previous “top”.
There is Ing ranking scale with 8 (one stone), 4 and 2 points granularity and three classes instead of just two.
You are complicating things a bit here i think 
. To me the question is only wether it is necessary to represent the difference in elo within the 9dan rank on ogs in name as well.
why do you assume we have to somehow fix the current pro ranking system because AG arrived? AG isnt competing with other pros. Its title is only honorary.
There is a seperate league for AI and there is a reason for that. Chess AI s also compete against each other and have their own elo system.
Of course feeding a lot of strong computers into an online elo system will make it harder to gain points, but ranks here have the main purpose of allow matchmaking, which would still work i think. Should that turn out to be a problem, there might have to be changes a lot bigger than adding a few ranks at the top.
I feel having a differentiation below 25k is great.
When i started playing Go and registered at OGS i was 51 Kyu. It was rewarding for me to see a progress in rank, first starting out at 13x13 boards. I experienced a difference between 35k and 25k. And i believe having an early recognition, and therefore reward, of player progression is especially important in a game with a very high entrance skill barrier.
I am kind of in agreement with GaJ’s original post… Maybe you have different experience bur from what I have seen bellow 20k the differences in rank are more only cosmetic in my opinion. It seems normal that 20k is wiped by 24k and the progress is very fast.
It can feel good to progress quickly from 40 to 25 for beginners but it can then feel very discouraging to fall back to 35 just because of streak of bad luck… Also after such a quick progression it may be depressing when the progression rapidly slows to “normal” rate. And there is also tradition to consider…
Not that I would mind it, just another opinion. And I am always scared to screw with something well established just because it seems good at first glance… 
Hey Adam,
i disagree with the argument that bad luck could depress the newer player. At least not more then the ‘catch all’ below 25k group. With ratings luck should be reduced, though i agree that variance in skill level is less accurate in these regions. Still better then the big group.
I feel feedback is better then no feedback.
This seems similiar than the “forever 30k” problem we had till 2014.
OGS used to have it’s lowest rank at 30 kyu, and huuuge pool of beginners who were stuck there without being able to get out, i also was one of them. Problem was caused when all the beginners were registered as 30k, and our elo quickly drop to negative.
When i ‘lost my first 50 games’ and started to actually learn how to play this damn game, i was still playing with all the other beginners who were in same situation as me, but some of them had a head start. So no matter how much i played or learned, i was still getting beated by other 30 kyus, many of them were able to give me 5 handicap stones, even if were the same rank!
It would not have been even that bad sitewide, unless OGS would have not been so beginner-friendly server. Huge proportion of new users are/were new players, and when we keep throwing “new meat to the grinder,” the problem just gets worse and worse.
How this was fixed in 2014? Mass adjustment. Our ranks were raised all at once by using some secret(?) algorithm, some of by 5 stones, me by 10 stones, and some by 20 stones. There literally was guys who got pushed from 30k to 10k in one night, but that did let us all find better games.
Is there any good reason why it shouldn’t?
Yep, since matchmaking is based on ELO, anyway. But why not differentiate the displayed ranks accordingly as well?
As people have stated in this thread, that simply isn’t the case See also this post almost right above you: Please (re)instate ranks beyond 25k and 9d - #27 by Vicarios
Yeah, but the whole issue is, for instance, 30k beating 40k, and 40k taking forever to make it to 24k (all the while being displayed as 25k the whole time). Making it “out of” 25k to 24k can take a long time if you join OGS as someone who’s never played go before, as evidenced by my own experience as well as that of other people in this thread.
This is not a rhetorical question, I’m genuinely curious: Is there any tradition outside of OGS of having a “lower rank cap” of 25k?
Given all the data is available, refitting glicko points into the 25k-9d should be easily doable in a way that maintains reasonable bucketing.
Current scaling is linear, but does the “correct” scaling need to be?
Why insist on having those arbitrary rank caps, though?
Given all the data is available, refitting glicko points into the 25k-9d should be easily doable in a way that maintains reasonable bucketing.
I’m not sure if I’m understand what you mean, but if you mean “maintain 9-25 as the maximum range, and fit everyone inside it”, then two massive consequences of that would be:
Current scaling is linear, but does the “correct” scaling need to be?
Similarly.
Sorry, but i fail to see where you got the idea i was arguing against anything??
In fact I liked your idea. I kindly ask that you reread my reply to you, where i said this:
Ive already heard that many people find the skill gap inside the 25k rank is too big. Thats why im all for adding lower ranks. It could make climbing the ranks more enjoyable for beginners.
Im am a little less convinced (but not against it), the same is merited for the top end. Ive heard no complaints by a player of that strenght and ranks above 9d arent used anywhere else, making them (hypothetically) an OGS-only thing. Thats also ok of course, but only if the people on OGS want it. Thats why I simply posed the question.
In my reply to @GreenAsJade i felt that we didnt need to include alphago in this.
Maybe this made my intentions a little clearer.
Sounds good to me @kickaha
I think I didn’t get my point about Alphago across clearly. I was responding to:
I was trying to show how the emergence of Alphago makes this a lot less of a laughing matter, because Alphago shows us that just like in the below 25k range, in the above 9d range there is a whole world of skill levels.
I think that prior to this there was some sort of implicit “by the time you are 9p, it really doesn’t get much better than that”.
The idea that 9p or 9d is “the best you can be” might now be the laughable concept…
GaJ
That’s totally fair enough, of course, and I wasn’t trying to suggest that you were arguing against me Just genuinely wondering why 9d+ players shouldn’t be differentiated. Your answer seems to be “because it would be an OGS only thing, and so far no one has complained about it”.
I think what adds to the confusion is that professional ranks work differently than amateur ranks. I’ve heard it stated that the difference between a 1p player and a 9p player is approximately one stone (I hope I get corrected if this is totally incorrect). Can it even be unequivocally stated that 1p > 8d?