The KGS is dead, long live the KGS ?

[Kirby]

Let me try again the purpose of the tilde: It indicates that the user exhibits a tendency not to play weaker players. As explained in the kgs faq, if everyone did this, it would be hard for weaker players to improve, so it is discouraged on the server.

It's impossible for everyone to do this, because in order to get a tilde, the guys playing you have to be willing to play weaker players

[gamesorry]

I have to say that ? is one of the major reasons I don't play many games on kgs. It's so hard to get a game when I have the question mark, and even if I got rid of it once, a few months later when I'm back it's there again

[BlindGroup]

I have to say that ? is one of the major reasons I don't play many games on kgs. It's so hard to get a game when I have the question mark, and even if I got rid of it once, a few months later when I'm back it's there again

I agree that their policy of having accounts "time-out" seems like it doesn't work too well. I guess I may be one of the few people who don't avoid people at my rank with a "?". I've played a few. I think many of them were people who were just playing elsewhere for a few months and got demoted. The net result is that they usually beat me badly and seemed to be "under-ranked". But this is just my impression. I never checked back to see if they ended up with a higher non-"?" rank. (And everyone who beats me seems stronger than I am

I think the practical issue is that the ranking formula only considers the games played in the last 180 days. So, if some one hasn't played in that period, they don't have a rank? It seems like a better solution would be to initially assign the returning player their last rank (without a "?") but treat the player like a new player in the ratings formula so that their rank adjusts quickly up or down until they have enough games for a reliable estimate.

[ez4u]

I like this discussion. I understand the intent behind all of the marks and so on. I do not even disagree with any of them. However, this is a good time to look at their effect in practice and try to improve on that. I haven't played actively online in a number of years. My server of choice would continue to be KGS, except for the difficulty of starting out again with that '?'. I like the idea of just leaving the last rank in place but 'behind the curtain' treat it as a newbie '?' rank. The KGS rating system adjusts fast enough for players with only a few games that the effect will be minor.

The other unfortunate effect of the '?' was to relegate '?' players to the bottom of the 'sort by rank' list. Old timers will recall that this was not true initially. Originally the '?'s were sorted directly into the list according to their rank. After some time (a couple of years?) this was changed to put all the '?'s together below the firm ranks.

As for the '~', just turn it around. Stop being so negative! Achieve the same distinction by giving a 'hero' badge to those who play more games with weaker players instead. (Doesn't DGS have something like this these days?) If you really are into the idea, then you can restrict yourself to only games as white against heroes.

[Kirby]

Hero idea is really creative, and I like it. Players shouldn't be punished for wanting to play stronger players- who doesn't?

But rewarding people that teach weaker players is a great incentive.

I don't know why this wasn't thought of before. Great idea.

[jeromie]

I agree with Dave: rewarding positive behavior instead of shaming behavior perceived as negative could have a significant positive impact on the server culture.

As a side note, I've gotten a ~ once. I went through a (relatively short) phase where I played the majority of my games against bots, and the bots that were available were all a little higher ranked than me. I sought out games against lower ranked players, so it didn't stick around long. I don't pay much attention to the tilde when looking for a game, except that I've come to expect those with the tilde who are around my rank might be more likely to refuse a game.

[Fenring]

In reality the ~ have no bad effect.
Ans as he concern a very little minority,seems no sense to give a hero badge to 95% of user or more

[Kirby]

In reality the ~ have no bad effect.
Ans as he concern a very little minority,seems no sense to give a hero badge to 95% of user or more

I viewed it as a higher percentage than 95%. The point of a reward is not to give it to everybody - it's to award people that take more-than-usual time to teach weaker players.

That's the point of the system, isn't it?

[KOCMOHABT]

I think the practical issue is that the ranking formula only considers the games played in the last 180 days. So, if some one hasn't played in that period, they don't have a rank? It seems like a better solution would be to initially assign the returning player their last rank (without a "?") but treat the player like a new player in the ratings formula so that their rank adjusts quickly up or down until they have enough games for a reliable estimate.

It seems to me that the best solution will be the following: if the user stopped playing on the server and it passed six months, then to calculate his rating it is need to use his last 20 rated games no matter how long ago they were (therefore, his games archive will have at least 20 rated games all of the time, exception is a new account). In this case there is no need to change the formula for calculating the rating in general and the system as a whole, and the user will never get "?" If he stops playing for a while. And it will be comfortable for him to return on the server.

[ez4u]

I think the practical issue is that the ranking formula only considers the games played in the last 180 days. So, if some one hasn't played in that period, they don't have a rank? It seems like a better solution would be to initially assign the returning player their last rank (without a "?") but treat the player like a new player in the ratings formula so that their rank adjusts quickly up or down until they have enough games for a reliable estimate.

It seems to me that the best solution will be the following: if the user stopped playing on the server and it passed six months, then to calculate his rating it is need to use his last 20 rated games no matter how long ago they were (therefore, his games archive will have at least 20 rated games all of the time, exception is a new account). In this case there is no need to change the formula for calculating the rating in general and the system as a whole, and the user will never get "?" If he stops playing for a while. And it will be comfortable for him to return on the server.

I think this is not appropriate for the KGS rating system. AFAIK it is not just that the system only uses the games from the last 180 days, it also has the games' weights decay throughout that period. It might be possible to keep the last win and the last loss at the minimum weight and use them together with new results to calculate a new rating. However, we would want the new results to quickly dominate the rating calculation so as soon as a player has at least one (?) new win and one (?) new loss the out of date games could be dropped. KGS should get some technical advice from WMS on that.

[ez4u]

In reality the ~ have no bad effect.
Ans as he concern a very little minority,seems no sense to give a hero badge to 95% of user or more

If the qualification level for heroes is the same as that for villains (~), there should not be many more heroes singled out than the existing villains. If they switched from badging one to the other and we suddenly see a large number of heroes, it might indicate how many people took their villainy and moved to another server.

[HermanHiddema]

Instead of letting the server decide, how about letting the user decide?

Give the user some settings to automatically mark players as heroes/villains depending on what that user finds important in an opponent (plays weaker/stronger players, grants/requests undos, talks a lot in the chat, resigns often/never, willing to review games afterwards, has "sai" in their username, whatever tickles your fancy)

[Fenring]

In reality the ~ have no bad effect.
Ans as he concern a very little minority,seems no sense to give a hero badge to 95% of user or more

If the qualification level for heroes is the same as that for villains (~), there should not be many more heroes singled out than the existing villains. If they switched from badging one to the other and we suddenly see a large number of heroes, it might indicate how many people took their villainy and moved to another server.

i dont understand the logic of your whole post.
especially "villainy" or "heros" for ~ mark, the mark is just an objective information you give to KGS users.~ is not a punition,just an information.
And in this case, play only with weakers players is as bad, for find a match, than play not enough against weaker player,it just inverse the problem.
i really dont understand the problem: the ~ is hard to obtain(even in one month training sanrensei i didnt have, i just had when i play ranked only against stronger and pedagogic against weaker), easy to leave.
i have the feeling its just some people who have ~ ,who feel guilty to have this one,but in reality they feel guilty for their behavior,no?
i really dont understand.
When i had the ~, i know why and it was never a problem.

[Kirby]

When i had the ~, i know why and it was never a problem.

To be clear, why do you want to keep ~ around?

[jeromie]

the mark is just an objective information you give to KGS users.~ is not a punition,just an information.

This may be true in an ideal world, but it's not true in practice. When the developers choose what information to display in a prominent position they are implicitly assigning a value judgment to that piece of information. At the very least, that information is considered important enough to report. Even if some people treat the tilde as a simple data point, many in the community will automatically conform to the value judgment offered by the developers. Conformity and shared values are an inescapable property of human communities.