Well, there are differences in a couple of ways:
1.) In some cases, you would not play identical opponents, but stronger opponents. Playing stronger opponents helps you to improve, so there is benefit to it. It's not that uncommon of a story that someone makes a new account on KGS because their rank is stuck.

2.) This assumes that you eventually get a win. You never get the win in some cases (actually, I've had a handful of escapers, but I don't recall ever having received credit for a forfeit - it's possible that I've missed one, but I'm sure there are others in a similar situation). If I play 100 escapers, who each escaped 3 times and then quit KGS, there are 100 wins that I should have that I never get credit for.

_________________
be immersed

This is slightly off-topic, but I wonder how difficult it would be for KGS to implement a "noescape" option like they have on some chess servers?

You just have a menu item called "noescape" on your account settings that you can toggle on and off. When it's toggled on, you are only matched with other people who have "noescape" toggled on, and any escape/disconnection counts as a forfeit. No one complains about this because they have intentionally agreed to this rule. If "noescape" is toggled off, you are only matched with other people who have it toggled off and any escape/disconnection just suspends the game. A "noescape" option would allow the players to voluntarily sort themselves into groups that are bothered by escapers/disconnections and those that aren't.

Another useful feature might be to allow either player to disable communication during the game. (A "censor talk" feature rather than a "censor player" feature.) It might be designed so that it can't be turned on until after the first few moves have been played, so as to allow the players to exchange greetings, and to turn off automatically after both players have passed to allow post-game discussions. If your opponent starts talking smack during the game you could just toggle "disable talk."

Of course, it may be harder to implement such features than it seems at first blush; but they would give the players more control over their own playing environment with respect to escapers and trash talkers.

All that having been said, as someone who has spend hundreds if not thousands of hours playing on internet chess servers, I can tell you that KGS is an Eden of courtesy and good will compared to the chess servers, and I personally find KGS's present system sufficient for just this reason.

Of course, (1) is why I intentionally included that caveat in my statement in the first place. I must admit I find (2) a little curious, since the claim is discussing how important getting one additional win is, when the problem is that getting many additional wins doesn't seem to have an effect on rank (=

Nevertheless, a nice solution to the stuck rank is that if you simply play games ranked at where you think you should be the KGS system is kind enough to move you accordingly (provided you can actually earn that rank).

As far as (2) goes....if you are playing escapers that only play very few games, and never get a solid rank, a victory against them is virtually meaningless for ranking calculations.

Why doesn't getting many additional wins seem to have an effect on rank? Getting many additional wins improves your rank.



Why is it meaningless? I have spent the time to play a game with them, and I have won. I should get credit for the victory.

Perhaps the example was too extreme. Surely it can occur that, since the system allows it, players can escape, say, 4 times every 6 months - at a frequency rare enough not to get penalized. Take 100 of these players, and have their escapes be against me. I want my wins against these players.



Yes, when you enough games, most ratings systems converge to some level.




OK, you put a caveat in your statement. This implies that you see that, in some cases, the current escaper policy does not produce the same result as if the games were counted as losses.

So I will ask for a third time:


Actually, I'd even say that Bartleby's solution would work as a viable alternative, as well.

_________________
be immersed

There has been situations I have encountered in the past where my opponent had something urgent come up during our game and he tells me he has to leave so he resigns instead of escaping even though he is ahead. It seems that it would be unfair for such a person to come back later to find that he has been deranked as he had resigned a "won" game. From the perspective of the two players, it's a matter of courtesy. From the perspective of KGS, it is intentional malicious conduct which affects everybody elses ranking. It would certainly make it hard for the deranked person in such a case to find KGS as a friendly server.

Perhaps the only elegant solution left to a player in such a case is just to go away from the keyboard and let their time run out. Then they have not escaped nor resigned and cannot be punished for either misconduct.

_________________
http://tchan001.wordpress.com
A blog on Asian go books, go sightings, and interesting tidbits
Go is such a beautiful game.

first, what does deranking exactly mean? is it removal of one's rank and giving him a '-'? because deranking sounds to me a bit like demoting someone and that would be really ridiculous punishment

second, i suppose that topazg meets quite few escapers and so intentionally resigns only few games and that there is quite few people like topazg. similarly, only few people have to left their game and so resign out of courtesy (which i personally value quite much). so ranking system should be robust enough to withstand influence of such minority of corrupted data.

third, i personally find topazg's practice pretty unusual, with its biggest flaw that it encourages escapers to escape more instead of punishing them. but don't really mind, if he likes it

fourth, i think KGS admins actually derank someone for intentional losing pretty rarely. from this discussion one could think that wms with KGS admins pick every day hundred of players who have been intentionally losing and derank them on the top of the pyramid to celebrate the god of Perfect Ranking System, which i guess is not the case

fifth, i myself have played hundreds of games at KGS, going from 18k to 1d and met only few escapers, i would say less than ten. no big deal. but from this number of escapers, i am pretty sure i didn't actually get more than one win by forfeit. for this reason i think that KGS policy is ineffective. it seems that more people share similar experience (at least Kirby said it explicitly and i believe there are others)

_________________
Spilling gasoline feels good.

I might be wrong, but probably not.

by the same logic... resigning those unfinished games is still "virtually meaningless for ranking calculations"

By making your opponent sit there for perhaps 10 minutes, it's still rude, but you avoid the possibility of being deranked and avoid having a game counted as an escape by you. Interesting idea.

Anyway, I don't really like Bartelby's NoEscape idea, because I can see one of three things happening:

1) The community is splintered, and you can only find games against half the number of people looking for games

2) Or, most people switch over to the NoEscape option, and not choosing NoEscape becomes a sort of "Beginner's trap" where people new to the server do not select that option and only face people likely to escape.

3) Nobody uses the NoEscape option (due to being too difficult to find a game), and the escaping problem is as bad as ever.

Some of these issues might be alleviated if NoEscape was an option the host of a game could set up, which the challenger agrees to (much like time settings are handled). Still, I prefer a simple 5-10 minute grace period before an escape turns into a resign. If someone's connection issues are so bad that they consistently cannot get back on KGS within 10 minutes to resume a game, then they'll probably end up being branded as an escaper by the current system anyway (as they won't be able to resume all the games that they get kicked from).

_________________
We don't know who we are; we don't know where we are.
Each of us woke up one moment and here we were in the darkness.
We're nameless things with no memory; no knowledge of what went before,
No understanding of what is now, no knowledge of what will be.

I'm sorry, my labeling got a little mixed up when I was multitasking....allow me to restate.

You brought up two issues "Not getting an escaper when, while people are getting stuck ranks" and "not getting awarded wins from people who play very few games, escape three, and never return"

In response to the first one -- There appears to be an incongruence with the problem and proposed solution. The problem of stuck ranks it that (in theory) an excessively large number of wins is required to promote, and the proposed solution is including a disproportionately small number of wins to the sample. My suggestion was that a presently available solution would be to simply play ranked games handicapped at what you feel your present strength is. If you are correct in your assessment you will reach the correct rank much more quickly (and what's more it will help improve the rank accuracy of all those you have played against, and even potentially reducing their own similar "stuck rank" problems.)

The second issue was that if a player is never playing enough games to get a confident rank from the server (i.e. they escape a few games then make a new account), then a victory against them will not carry very much weight with respect to the ranking system. What instead would happen would be once the losses were awarded, your rank would be used as a benchmark for scaling the "solid" point on their rank. Adding the losses would leave your rank virtually unchanged, and their rank significantly lowered (which if they never play on the account again anyway, doesn't really matter).

This exists, you can censor your opponent once the game has started, then you will be unable to see their chat. Likewise you can disable it once the game has ended.

I'm not quite sure which quote you are referring to for "Not getting an escaper when, while people are getting stuck ranks". You can get stuck at a rank, yes. It's possible that you are stuck at an inaccurate rank due to escapers.

I agree that playing more games will increase the probability of having an accurate rank, but this is true of most ranking systems. More data typically helps in such an analysis.



Again, the example I provided was an extreme one. You can also imagine players that play on a regular basis, but simply escape periodically, not enough to trigger a forfeit for their escapes.

Once again, I think that this could be resolved by:


It's been conceded that the current escaper policy allows for a different result than if the escapes were counted as losses. If an escape is not mutually agreed upon, why should this be the case? The only explanation I can think of is disconnection issues which:
a.) Can be resolved by making a loss after a set period of time (eg. 5 or 10 minutes).
b.) Are becoming less and less common with today's technology.

_________________
be immersed

There's a few points that some people are being careless about.

Be sure to distinguish "you can resign at any time for any reason" and "you can resign anytime your judgment indicates that you are losing." Those are completely different norms. Even the latter isn't the actual norm, since Go etiquette mildly discourages resigning during counting. But it's at least close to reality.

Similarly, if the rule is that you cannot intentionally resign a won game, then it's no argument to point out that you might not know. There are clear instances of losing a won game. If a player alternates between playing at dan level and playing his first twenty stones in one corner, he is a sandbagger. A natural thought is that if you favor a rule against intentional losses, you would only enforce it in those clear cases.

Lastly, it certainly seems to me as if escaping a game you are losing is against the KGS rules. It just isn't directly punished. Certainly it is not "fine by the rules."

_________________
Occupy Babel!

Perhaps. I guess the idea of being able to resign at "any time for any reason" is a separate argument from being able to resign in the case of escapers. I feel that, if escapers were more directly punished, people would not feel inclined to resign to get closure on the game.

In the case of escapers, I feel that the one that should be at fault is the escaper - not somebody that wants to finish the game.

I do personally feel that resigning should be allowed at "any time for any reason". The goal of the game is to win. You can have any strategy that you want. You might want to try a wild fuseki, or some mind trick, or something totally different. I feel that resigning should always be an option (though, again, this is a slightly different argument than the idea that resigners from escaped games should be punished).



Sure. So I propose a more direct punishment.

_________________
be immersed

At KGS, you are deranked when an admin removes your ability to play rated games and at the same time removes your rating information from the system.

I don't derank many people at all, and I hardly ever run into escapers. If I meet one, I can't say that it particularly bothers me at all. It always surprises me that some people get very emotional about this issue.

For me, it's this kind of statement, from a rather intelligent person, that surprises me. I too have no particular problem with escapers, but I understand very well that they can annoy others greatly.

Trying to understand others persons and not watching always their own bellybutton is a good way to improve, not only in Go

Funny to read this about deranking, when messing up the rating system was the concern to begin with. It's like hunting down the lonely honker in the night by helicopter. The neighbours will appreciate it.

Kirby: I'm not really sure if this is a necessary clarification or not. However, I did not intend for my comments to be an argument that you must keep the current KGS system. Everything I wrote is compatible with a stronger punishment for escapers.

I do disagree that you can resign for any reason, though.

_________________
Occupy Babel!

Sorry again, that should have read "Not getting escaper wins..." The point is that the stated problem is that an excessive number of wins will not cause one's rank to properly adjust, and the proposed solution is that a disproportionately small number of additional wins will make a difference compared to the excessive number of already existing wins.

My solution was not "just play more games" it was "play games handicapped at the strength you think you should be ranked." That is, if you are ranked 10k, but think you are really 7k, play 7k's even and give 10k's H3. This will do much more to remedy your issue (and will further alleviate the problem for any of your past opponents suffer from the same).




The persistent escaper problem has been discussed in other threads, and as I've mentioned before (and alluded to in this thread) will result in an error of less than 1/3 of a stone.






This has been discussed time and time again (dating back to RGG), and in fact has even been answered in this thread. The principle of "No one should be forced to continue a game they have no desire to finish."

Let me just end my participation in this discussion by saying, "???".

If I have to explain my confusion with this rationale, I don't know if it's worth arguing anymore.

_________________
be immersed

I don't meet many serial killers, and they don't bother me much at all either. They're still a bad thing, and it would be good to
address the issue. Some would say they should be locked up after killing only one person. And as for your personal experience, you are an admin, and also you don't play automatch which is where many more escapers are since you cannot screen your opponents.


If only there were an option to make a game end that they don't want to finish. Maybe something similar to the resign button...