Then maybe serial-timeouts should not count only when the timeouting player has not logged into OGS for some extended time (like two weeks) and count when he is normally playing other games.
why not let the winner decide if he wants the timeout - win or not?
Solution: Count Timeout wins as wins, Timeout loses as loses.
Now you have heard the solution Adam3141.
Note: If a player has serial Timeouts for a reason, then they should share it with Mods in those unique circumstances.
Backed by the fact that this works for a few million users on chess.com, Iâd say youâre onto something.
Yes, that is a solution. Now you are only missing the constructive and mathematically sound part.
And since I assume the most common reason for serial timeouts is that a player just stops logging in, this logic is kind of backwards. Canât tell mods if heâs not here.
Look I understand why it is bothering you, I too would prefer the simple way (as an individual), but you have to realise there is a reason why it is programmed this way for now and the programmer should care about the community as a whole. Just on the off chance that you are not just being salty and actually care about the issue, I will try to explain it to the best of my understanding:
- If a player escapes because he is losing and it is counted as loss that is of course great and all is well with the world.
- However there comes a time when somebody times out even though he was winning. If that happens the other player gets an undeserved win
- An undeserved win means an undeserved rank boost.
- An undeserved rank boost means another smaller undeserved rank boost to the next player that beats the original player.
- It is all very small, thatâs why you probably wonât care, but over time and over many games this effect accumulates and snowballs and suddenly 1k rank is actually a much weaker player then 1k ranks were before.
- This will lead to the fact that our ranks will be lower then on other servers which will be confusing/annoying. Which would then require a manual adjustment which will annoy many people and is hard to do and estimate.
If the data is missing (the game is annuled) it is OK. It is when the data are wrong (undeserved win) that it starts being an issue.
I am sure there some better solution and I am even quite confident that Anoek is thinking about it, but this constant âtimouts should be wins DUUHHHHâ is just not helping.
Not sure how it is on chess.com but I have heard only bad things about that place
And really itâs just a game. Get over it :D. yes it is rude, but some people are just more emotional and have trouble dealing with loss or mistakes. Does not mean they are bad people. Try to be forgiving and you can feel like the bigger person in exchange
What is not constructive here? It is simple and to the point. As for mathematical soundness: this is the same thing that is done not only at chess.com but also in professional Go (or do you find that not mathematically sound?). IMHO I think that you too should be required to give your argument a just as mathematically sound basis.
I think there is a problem for people playing many simultaneous correspondence games. If thatâs what they want to do, then they also must accept the fact that Go rules state that a loss by timeout in a correspondence game is a loss just as punishing as one in a live game.
Letâs clear this up and move on â not talk it to death and let the correspondence players play by really different rules concerning losses.
So what? That also happens in professional Go and chess games and should also apply here. What is the constructive and mathematically sound counter argument?
I do not agree with your next pointsâ conclusion that we will eventually need to manually adjust ranks because this (mass timeouts) really does not happen that frequently â and it will happen even less frequently when correspondence players start getting properly knocked down in rating when they loose by timeout.
Itâs just repeating what has been said countless times without trying to understand the issue. Thus not constructive.
In proffessional go it is quite different issue. Pros are already largely equal strength and ranks work quite differently there. They are more achievements you cannot lose and do not always reflect true strength.
Again, works for individuals, damages the whole community.
Thatâs unfortunately the issue. It does not have to happen often, the point is you are corrupting the data and over time this corruption accumulates. As stated before no data is just better than wrong data. Again no idea how it works/doesnât work on chess, so cant argue about that point.
Honestly I am no mathematician and I do not understand the issue deeply enough (or care) to keep debating over this. (not sure what more I could say anyway) this is just my understanding, take it or leave it. But if it was SO EASY donât you think Anoek would have done it already? I just find it weird that you are all so sure you know better than the man who spend countless hours implementing the whole system and getting it to work.
The reason that the debate continues about this is not incessant whinging, it is because the issue continues to raise its head.
Each time, it is a new person at the top of a thread who has observed timeout-escaping behaviour and comes to ask about it.
Those who care about it point out where the answer is (as Tongue did) and what to do about it (as bugcat did) and it could have sat there as question and answer if it werenât for the people who think it should stay the same leaping back in and fervently telling us all yet again why we should just suck it up.
Of course, this situation might be improved if it were written officially somewhere what the timeout rule is (so people donât have to ask), but I doubt it, because new people encountering it would still come and ask âwhy is this so?â. Because the problem is so obvious, and the reason why it is the way it is is so subtle.
And for better or for worse, timeout-escaping-looking-behaviour does happen quite frequently (so any concerns about rank distortion do need to be taken into account).
⌠actually, I wonder about this whole argument. What percentage of games would have to be timeout-escaped to even noticeably affect the rank pool at all? Wouldnât it be lost in the noise unless itâs happening âall the timeâ? And if itâs happening that much, isnât that too much cheating to tolerate?
⌠and if we changed the rule so you lost rank when you timeout-escape, wouldnât this cut the occurence of it to once in a blue moon? Arguably, any appreciable frequency of it at the moment is due to cheating - the number of times âreal life intervenes and I couldnât get to the internetâ is surely negligible?
On the contrary, I find that those who wish to end this cheating (loss of a game but not loss of ranking points) understand the issue quite well and that is the reason we continue repeating our position because it seems that you donât understand it (as you even stated at the end of your post), or perhaps you just donât care (as you also stated near the end of your posting) and prefer the current situation.
And professional Go should be our standard (unless you wish to invent a new variant). Ranks need to be established, of course, but not by letting timeouters remain unpunished (at least as far as ranking goes). Do you really believe that professional Go is so different that only they are the ones who should give timeouters rank reductions? Ridiculous point of view: youâre just inviting cheaters (with the current system).
What is that supposed to mean? This is just another fog-over to divert attention from the problem and retain the status quo. What is your mathematical basis for this?
I think that Anoek needs to reply to that. But, as a former professional programmer, I can tell you that counting timeout losses as losses (rather than ignoring them) is not a problem to program. In fact, it is already done in live games!
This discussion has become absurd. I find it amazing that anyone who thinks about this issue even plays correspondence games⌠Perhaps we should start discussing why timeouts in live games count as against ratings for the timeouters and award some of the winners (who were actually behind in the game) to advance their rating? I mean: what is the difference? Be consistent.
I am sorry, I donât think I will be able to explain my understanding of the issue any better, than I already have. If you do not agree with my explanation we will just have to agree to disagree. Which is ok too And just to be super sure, maybe I am wrong, but from your passionate response I got the feeling that you might think my opinion on the matter somehow matters. It does not. I have no say in this and no gain from keeping the status quo or changing it. And you are right in thinking that I do not really care. I just wanted to offer my point of view on why I think the solution might in fact not be as simple as you seem to think.
sorry for the misunderstanding, I am not reffering to the difficulty of programming, but to the seeming simplicity of the solution, which might in fact be misleading.
just for the record, I donât think they do. As far as I know pro can never lose a rank he/she has once gained.
Maybe that should change too!
(slowly backs away while avoiding eye contact)
Here is an idea I havenât seen before. It may be a bad idea (please donât throw things), but I will be bold. As I understand it, a person who times out of a correspondence tournament is booted out of the tournament. By analogy, what if a person who times out of any correspondence game has the rest of his/her games annulled or automatically resigned (depending on which way you want to go with this rule). This would inhibit most people from gaming the system because they would lose (or have annulled) their good games as well as their bad games. Admittedly a few people with most or all bad games might slip through. If this puts too much of a burden on short correspondence games, then it could be applied only to the longer ones. If a person is legitimately incapacitated for a week, for example, he or she will likely timeout of all their games anyway.
I should add that I have a horror of timing out of a correspondence game due to incapacity (a real possibility at my age). So if it ever happens, youâll know Iâm either in the hospital or dead. In all other cases, I will resign my games.
And yet (to the time point of this quoted post) your comments account for 9 of the 32 posts. Hmm âŚ
But OK, we obviously disagree about how timeouters should be handled.
Concerning professional Go players â I think too that you are correct about never dropping in rank. I forgot about that. +1 for @Adam3141
But now seriously: if this system applies to correspondence games, why does it then not apply to live games? Or does it, and I have again forgotten how things work?
The way i see it, the main problems are:
Escapers, who doesnât want a rank reduction (timeout does not count as loss)
Players, who win too often because the other players time out in correspondence games (timeout counts as loss)
If someone timeouts, ask the other player if he wants to count the win by a simple YES/NO Question.
If you feel you do not deserve the win: just press NO
If you feel the other is escaping from a loss: just press YES
If the other player timeouts in the first move, just annul the game.
what will happen:
Escaping is senseless, because an escaper can be sure the other player will press YES
Time-Outs, which seems real timeouts, where you feel you are losing you can just press NO
In the case, where you are losing badly and the other still timeouts and you press YES - also that is possible, and it should be - this is like losing on time in a live/blitz game.
I do not understand the other arguments about the rank effects for the whole pool at all. Ranks are dynamic, not some fixed values.
The main part of rankings is to find players similar of your own strength - if that works, the system works fine.
This argument, if you donât mind, is total nonsense.
You worry about âAn undeserved win means an undeserved rank boostâ. But with the current rule a deserved win which is not credited also means the rank is suppressed. To be fair, unless a game is played to the end either way you do it the rank will be distorted.
BUT, and it is a big BUT, if losses counts people will only do it when they canât help it, and if losses donât count people will do it much more often.
So now you have it, when time outs donât count the distortion will be more than when it counts.
as far as my understanding goes you are not correct. If you annul the game it is like nothing happened, you do not gain anything, opponent does not lose anything, there is no distortion, just data missing. There is data missing, but it is not wrong. Missing data canât distort the truth.
in a far fetched metaphor letâs say Ke Jie played 10 correspondence games on OGS and he timed out all of them, because he had to attend WGC. Which statement is more true:
- 10 people on OGS have beaten Ke Jie
- We do not know whether someone here could beat Ke
Yes, neither is perfect (Maybe one of us couldâŚ), but uncertainty is better than potential nonsense.
Unfortunately, that is true only partially, we also want our ranks to stay relevant in the whole go community, not just for purposes of matchmanking on our server. Otherwise it would indeed be just fine. But yes, your solution seems somewhat better. I think similar idea has been proposed and discussed elsewhere but canât find it now.
Donât be angry with me I am basically on your side I too would like it if the good guy would get the win. I am just trying to show you that from what I understand it is not that simple as you seem to think, and more variables than oneâs sense of justice has to be taken into account.
If thereâs one goal in discussing a potential improvement in this area, itâs not foremost âthat the good guy should get the winâ.
I think it is âthat the bad guy should not be able to escape the lossâ.
What would be wrong with a system where the moderators an annul games in the case of genuine disaster (like being called to WGC )
In fact, they could restore the darn things: simply have the power to re-instate timed out games.
Then, in the very rare occasion where someone has a case to make, they flag a mod and tell their sob story. The mod reactivates their games - if they deem it worthy.
This means that the cheaters can only do it once before the Mods say âdude, the dog can only eat you homework onceâ.
Iâd sign up to be on the list of people manning that duty. I donât think it will get called on that often.
GaJ
I am still baffeled that mods cannot change the result of the game now (but letâs leave that discussion for another thread )
but the main problem (again only as I understand it) is the people who just stop logging in and timeout a bunch of games (Ke Jie just left and did not even let us know ). And appereantly it is not uncommon for people to juggle like 50 corr games at once. Thus there will be no indication of undeserved victories and thus corruption of the data.
The way I understood things, the first correspondence timeout is ranked as a loss, and all subsequent consecutive timeouts are unranked. So my perception was that if you finish a game by win or loss, rather than timeout, the unranked timeout condition reverts to normal and the next timeout is ranked. Tell me if Iâm wrong about this.