Life In 19x19

i had this wild idea..
what if this forum gives out dan or kyu certificate to our members.
i think there are many members who are willing to earn their certificate.
How to assess their rank is another hurtle we might have to discuss but it seems to be an intresting idea.

Interesting idea, but can I ask why? Most of us already have KGS/Wbaduk/national Go association ranks.
Perhaps this could be used for an actual "Malkovich Rank" of sorts? Sorry, I am a little confused about what you are suggesting here.

It can be defined by how many stones Magicwand gives you in a Malkovitch game ^^

More seriously, I have thought in the past that it would be interesting to run some sort of server where people listed all their handles (KGS, IGS, wbaduk, etc) and ID#s (AGA#, EGF, etc), and the server constantly scanned all those sources for games and ran them through the best available rating algorithm (WHR at the moment, IIRC). Two problems: It would need a lot of people to sign up to be useful, and I don't know if it's possible to to retrieve game results automatically from all of the different sources.

Nikolas73 wrote:Interesting idea, but can I ask why? Most of us already have KGS/Wbaduk/national Go association ranks.
Perhaps this could be used for an actual "Malkovich Rank" of sorts? Sorry, I am a little confused about what you are suggesting here.

it can be a soruce of income?
edit: to elaborate more.. i spent $1000 to get 3 certificate from Microsoft. MCP, MCSE, MCSE+i
i can brag that i am a MCSE. well i did it for job search...
anyway.. it is a certificate that proves that you are certain rank.
some people might pay $20 for that piece of paper.
right now it's name value is not as high as Korean Baduk Club..or Japanese Baduk Club..
but who knows..it might become valuable as theirs.
then it will be a soruce of income.

daniel_the_smith wrote:It can be defined by how many stones Magicwand gives you in a Malkovitch game ^^

...

Maybe I really can make it to Malkovich 17k, then!

The biggest problem is simply finding someone strong enough to evaluate the level of play for those interested. Actually, Eric Dai's forum did something like what you suggested, and for them it was plausible because...well, they had Eric Dai.

The "certificate" they gave out was really just a picture which provided an estimated rank along with a pentagon graph that showed some stats on a player's attack, defense, etc. etc...kind of like Pokemon lol.

Araban wrote:The biggest problem is simply finding someone strong enough to evaluate the level of play for those interested. Actually, Eric Dai's forum did something like what you suggested, and for them it was plausible because...well, they had Eric Dai.

The "certificate" they gave out was really just a picture which provided an estimated rank along with a pentagon graph that showed some stats on a player's attack, defense, etc. etc...kind of like Pokemon lol.

Go stats sound cool. It would be awesome if there was some accepted way of measuring different parts of a players strength, then we could get cool graphs like those from the chunnin exam in Naruto

Jedo wrote:

Araban wrote:The biggest problem is simply finding someone strong enough to evaluate the level of play for those interested. Actually, Eric Dai's forum did something like what you suggested, and for them it was plausible because...well, they had Eric Dai.

The "certificate" they gave out was really just a picture which provided an estimated rank along with a pentagon graph that showed some stats on a player's attack, defense, etc. etc...kind of like Pokemon lol.

Go stats sound cool. It would be awesome if there was some accepted way of measuring different parts of a players strength, then we could get cool graphs like those from the chunnin exam in Naruto

we can base our rating off other server's rating. something like "if you have kgs 3d account then you are 3d so we will certify you as 3d."

Magicwand wrote:we can base our rating off other server's rating. something like "if you have kgs 3d account then you are 3d so we will certify you as 3d."

But it is easy to get a KGS 3d rank if you are really only 1d by having a lucky run with a small number of total games, so no rank based on internet games should be based on less than 100 games.

Not only that. Due to all the variables involved - time limits, time of day, level of drunkenness or seriousness (yours and the opponent's) - it is highly unlikely that any dan rank can be guaranteed within 2 ranks or any kyu rank can be guaranteed within 5 ranks.

DrStraw wrote:

Magicwand wrote:we can base our rating off other server's rating. something like "if you have kgs 3d account then you are 3d so we will certify you as 3d."

But it is easy to get a KGS 3d rank if you are really only 1d by having a lucky run with a small number of total games, so no rank based on internet games should be based on less than 100 games.

Not only that. Due to all the variables involved - time limits, time of day, level of drunkenness or seriousness (yours and the opponent's) - it is highly unlikely that any dan rank can be guaranteed within 2 ranks or any kyu rank can be guaranteed within 5 ranks.

I never think of ranks as being more accurate than 2 ranks, ever.

daniel_the_smith wrote:It can be defined by how many stones Magicwand gives you in a Malkovitch game ^^

More seriously, I have thought in the past that it would be interesting to run some sort of server where people listed all their handles (KGS, IGS, wbaduk, etc) and ID#s (AGA#, EGF, etc), and the server constantly scanned all those sources for games and ran them through the best available rating algorithm (WHR at the moment, IIRC). Two problems: It would need a lot of people to sign up to be useful, and I don't know if it's possible to to retrieve game results automatically from all of the different sources.

There are few things that bait me, but "best rating algorithm" is one that normally manages it

Same with the ranking system for the Insei league and everything else, there is no best. Mathematicians spend far too much time looking for numerical purity, but every algorithm and system is only as good as the ability it has to fit to the design of the system. Professional ranks in Japan for example are awarded largely on accumulated performance, but they don't measure actual current playing strength. Therefore a 7d isn't necessarily better than a 5d, or even a 1d. But that's fine, because the ranks aren't designed to test that.

I think player inconsistency (like playing 30% of the time slightly drunk, or too tired to concentrate but ok to relax etc), time control inconsistency, and other random influencing factors affect someone's ratings far more than the difference between WHR, Glicko2, or anything else. It cannot easily be avoided that amateur ranks are completely arbitrary, and generally reflect only your actual performance against the various people you have chosen to play. KGS for example has a number of issues with its rating system, but it remains remarkably good internal consistency, thanks in no small part to the idea of rank anchoring.

It's no secret that ranks in the UK have deflated considerably over the last 7 or 8 years (by over a full stone looking at the regular tournament attendees). This is largely because an influx of new players with rapidly improving strength have pulled said points off the established players. Being able to reset one's rating helps, as to some extent does the EGF algorithm that is designed to deal with a small amount of system wide inflation (WHR doesn't have this hack does it?), but it isn't enough. We have a few dan players that will never again reach their peak ranks, and it isn't because they've got weaker.

I think all the top rating algorithms, including the basic ELO system, are good enough. The signal to noise ratio caused by player improvement and ability deviation far outweighs the merits of the improvements on predictive accuracy I think (although this is probably not the case if you play a very large number of games, 10+ a day or something, as the algorithms get much stronger with more data). If I wanted a shodan diploma, I'd go to Japan or the US, and I certainly wouldn't go to Holland or France. No one seems to have a "standardised" rank for any of the grades, so if we wanted to do L19 certificates, we could. I don't think anyone would want them, but we could do them

Let's issue certificates for the best word/move ratio in Malkovich games.

Joaz Banbeck wrote:Let's issue certificates for the best word/move ratio in Malkovich games.

Hey, there's a thought: We could add a button to flag good moves, and only show it in the Malkovich forums. Separate from the 'good post' in general, kind of a golf clap after an insightful move.

This post is unrelated to the current topic.

topazg wrote:There are few things that bait me, but "best rating algorithm" is one that normally manages it

There's a difference between "best" and "perfect"

Two things, first-- the point of that wasn't the rating algorithm as much as it was combining all those different sources of games and rating them all together. (Play your drunk games on a different account and don't tell the system about that account, hehe.)

Second, I mostly agree, with one exception. Say I lose to a 4k in an even game in a tournament. In subsequent rounds, the 4k beats a 3d and a 4d in even games. AGA's algorithm notes that I lost to a 4k in an even game and docks me a massive amount of rating points. KGS's algorithm realizes that the 4k was probably not 4k and retroactively changes the impact that game has on my rating. KGS's algorithm is a massive improvement in my mind, especially if you don't play that many rated games.

(KGS's algorithm could be improved further to reduce the drift it is famous for, I think. If it did that, I think it would be nearly perfect.)

daniel_the_smith wrote:Second, I mostly agree, with one exception. Say I lose to a 4k in an even game in a tournament. In subsequent rounds, the 4k beats a 3d and a 4d in even games. AGA's algorithm notes that I lost to a 4k in an even game and docks me a massive amount of rating points.

Not quite. All games in a single tournament are rated simultaneously, so if that 4k goes on to beat a bunch of strong player you're okay. The real problem shows up when that bogus 4k beat you in one tournament, and then beats up the dan players next week in a separate tournament. The AGA ratings don't propagate data backwards in time, but it is on the to-do list.