Pump up the rank (the wrong way): win by timeout


#1

Hi,

I do play almost exclusively correspondence games, and
I do like that a game doesn’t last forever, so
I try to play “fast correspondence” games.

I subscribed to “Fast correspondence” group
I participate in “Fast” and also “Really fast” tournaments, also
I created my own games with fast timesetting, and eventually…

… I won a lot of games by timeout. :slightly_frowning_face:

In the last two months I won 9 games (out of 42) by timeout, against opponents whose rank was between 13k and 1d.
Some of them I would win anyway, but some others I definitely wouldn’t.
I dont’ know how to check if those games actually gave me a rank improvement. I believe most of them did.

So I would revamp a request already done by others: could we have a sort of popup asking the winner (of a correspondence game by timeout): “would you validate the result or would you annul the game?”.

Otherwise, I think I should call a moderator each time and ask him to annul that single game. What would the OGS Team like best?

Similar requests:





#2

I think that calling a moderator each time is the right answer, because it supplies data about how many people actually care and would chose to annul if the button were there.

GsJ


#3

That could be a good provisional solution.

Could you help me also on this?


#4

Isn’t that a case of checking whether the record says “annulled” or not?


#5

It seems it’s not.
As far as I know, if a player times out multiple correspondence games, only the first of that series if rated. Looking at the profile of someone who has a timeout series, none of which are flagged with annulled.
Therefore I think, if the game is not annulled, one has to look up if the player has an active timeout series.

You have to go to your opponents profile and look in her “Game History” the last correspondence game, which was finished before yours ended.

  • If this ended by timeout of your opponent, your game is not rated. (since your opponent has timed out recently)
  • If this ended by normal means (no timeout, not annulled) your game is probably rated normal. (since your opponent stated it timeout “series” with your game)

Note I’m not sure about the details, how a timeout series ends. Since the probably above.
As fare as I can tell it ends if a correspondence game ends by normal means. I read somewhere in the Forum (don’t find the post now) that this game has to last more than 24h or 48h to end a timeout series.


#6

Ranking up on timeouts reminds me of a Jean Ralpheo quote from Parks and Rec:

“I made my money the old fashioned way - Got run over by a LEXUSSSSSS!”


#7

A useful reminder we really should have a visible indication that mass-timeout games were not ranked.

Is there any reason why they should not simply be marked as “annulled”?


#8

Just guessing: An annulled game has no result I would assume. A mass-timeout game still counts as a win for one side, if ranked or not. At least I think it should, especially if part of a tournament of course. (Although since a single timeout leads to disqualification it’s not that important. Hmmm, seems I can’t find a hard reason. Still my feeling is there should be a difference between timeout and annulment.)


#9

It shouldn’t count for one side, since that would result in rank inflation. The same number of points needs to be added / extracted on both sides to maintain an average of 1500 on the whole site.


#10

I think you just argued the opposite case :wink:


#11

I think there’s a way to check users rank history with evidence of how much each game contributed to it. It should be done using API and someone once gave me a link, but I can’t find it now.


#12

https://online-go.com/termination-api/player/308797/rating-history


#13

Suh-weet!


#14

To maintain the 1.5k points average we need to change points of the two players, if we the winner doesnt win points the loser cant lose points… so it would create a wave of players that doesnt want to have the rank decreased to let the game end by time-out


#15

Exactly what I’m saying, right? It shouldn’t count for one side.


#16

What’s the difference between “termination-api” and the “api/v1” I used to know?


#17

I found the url for the “termination-api” by reading the code. It is used to produce the ratings charts.

Unlike the “api/v1”, I don’t believe that there is any documentation available.


#18

I think that’s the opposite of what Joao was pointing out. It has to count for both sides or neither, otherwise the ranking system gets skewed.

That’s the root of the whole slew of problems in this area.

GaJ


#19

I believe V is saying “it shouldn’t count [JUST] for one side.” :wink:


#20

Damn communication. I thought he was saying “the result of the timeout shouldn’t count for one side” :smiley: :smiley:

I overlooked his earlier post where it’s quite clear that he is saying the opposite (correctly :slight_smile: )