Life In 19x19

I know I shouldn't be thinking about this, but an unusual rank dive on my "I'm awake and have enough time to play a serious game without being disturbed" account prompts me to ask: how exactly are older games weighted?

I heard chattering over the weekend that there had been a server wide rank adjustment which was downward for most people. I'm not sure if this is what's happened with your account.

I also wondered what caused my rank to drop the day I haven't played. Needed 5 straight wins to be back at the same "level".

Yes, I noticed this too. It's irritating when they keep moving the goalposts like this. After all the effort people put into getting stronger, only to have their graphs drop sharply without having lost games - that's discouraging. (An upward adjustment may have no more actual meaning than the downward one, but in contrast it would not be perceived as a well-aimed kick in the teeth.)

My solution is just to play free games on my 1d account. I get the practice I`m after, and no worries about losing the rank. Of course, it'll turn unstable again eventually, but from now on I will keep my accounts relatively unstable so that such adjustments won't have an excessive effect anyway.

I did feel that KGS ranks shifted somewhat upwards in the past few months or so (though, I haven't played much on KGS in between, so maybe it's just perception). Perhaps the downward adjustment is meant to offset the shift?

Shift in rankings happened probably because anchors changed their playing. I'm not sure how rankings are defined. Does the system have hard divisions, like assumption that only 0.1% of player base will reach 6d ranking?

Tami - you can't "lose" a ranking same way as you lose a game. It represents your level of playing and is accurate only when you play ranked games against fair number of different opponents.

I also don't like those rank-shifts, but I wouldn't like to be AGA-like either: An 3d AGA is like a 1k on IGS and that's humiliating. So I encourage the KGS stuff to keep those ranks in a way to be comapred to IGS or EGF. An 1d KGS should be a 1d all over the world, server or real games. That is my point.

I haven't knowledge about the rank calculations of KGS, IGS, EGF & Co., but from what I hear many very good players are not satisfied how KGS calculates ranks & ratings.

Pippen wrote:I also don't like those rank-shifts, but I wouldn't like to be AGA-like either: An 3d AGA is like a 1k on IGS and that's humiliating. So I encourage the KGS stuff to keep those ranks in a way to be comapred to IGS or EGF. An 1d KGS should be a 1d all over the world, server or real games.

So a KGS 1d should be 1d in the US, the EU, Korea and Japan? I'll let the national organizations know that they need to change their ranking systems.

hyperpape wrote:

Pippen wrote:I also don't like those rank-shifts, but I wouldn't like to be AGA-like either: An 3d AGA is like a 1k on IGS and that's humiliating. So I encourage the KGS stuff to keep those ranks in a way to be comapred to IGS or EGF. An 1d KGS should be a 1d all over the world, server or real games.

So a KGS 1d should be 1d in the US, the EU, Korea and Japan? I'll let the national organizations know that they need to change their ranking systems.

a lot of european 4 kyu's would be so happy!

@hyperpage: Yes, that would be best, if all Go players around the world are compareable within one rank-system. Would u like if there were several world records over 100m dash according to the continents or assosiacions? I wouldn't.

I'm all for KGS ranks matching to EGF.

Pippen wrote:I'm all for KGS ranks matching to EGF.

Why choose EGF?

Just to reiterate, my question was not whether the KGS rating system is good or bad, but rather how exactly does it work? There is a page about this here, but all it says about weighting older games is: "This system is further refined by assigning each game a weight based on how old the game is." Does anyone know any more details?

There's a bunch of stuff here: http://senseis.xmp.net/?KGSRatingMath

Ortho wrote:There's a bunch of stuff here: http://senseis.xmp.net/?KGSRatingMath

Thanks, but if you look carefully, my post above links to the same page.

I believe WMS has stated that the half-life for games is 45 days.