Yes, I too think that is still the case.
If you only consider âan undeserved winâ a distortion but a âdeserved win not creditedâ not a distortion because âyou do not gain anything, opponent does not lose anythingâ, then there is no way discussion could continue.
Again using an example, but instead of Ke Jie some newbie comes to play. He declares a rank of 13k but instead he is 23k. He is losing the games and he would have dropped to 26k had he kept playing but he times out and remains 13k. If you donât call that a distortion, then I have nothing further to say.
This topic of timeout, if people still have not yet realized, is essentially an ECONOMIC PROBLEM. I recommend reading the book Freakonomic or any other subjects on Pollution. Rank distortion is sort of like pollution, itâs bad for the whole world, but the person pumping out the pollutant is gaining something (he is saving his rank). So the only way to stop him from doing that is to impose costs on him (if he times out, he would lose his game), not his opponent.
Youâre right, it is a cost-risk question. The proposal in my previous post would impose a cost on the cheaters, whether you count the games as losses or annul them.
Letâs consider this cost-risk paradigm further. What is the more likely event: a cheater times out of multiple bad games deliberately or someone times out due to catastrophe? I suspect the cheater is far more common. Therefore the system should aim to thwart the cheater, while delivering justice to others. Perhaps the person with the catastrophe could appeal to the mods in the way that @GreenAsJade proposes, if that is technically possible.
Another idea to consider is to limit the number of simultaneous correspondence games one can play. This would reduce the distortion to the ranking system (whether counting multiple timeouts as losses or as unranked) and would have the added advantage of reducing what must be a huge burden on anoekâs servers.
This discussion reminds me of a recent topic on LittleGolem, where every timeout is counted as a loss :
https://littlegolem.net/jsp/forum/topic2.jsp?forum=1&topic=3139
Well, then say nothing further I still maintain that you cannot lie if you do not say anything. You might be hiding something⌠But you are not lying. (it is not perfect, and I am not saying it is and it may of course lead to slight distortions, but the distinction is important, and the other way around the distortion would be MUCH worse. Because while you would not be hiding the correct data, you would be feeding completely WRONG data with them). And as I tried to explain in my understanding uncertanity is better then nonsense.
Friends, letâs end this. Either I am going crazy and am talking nosnese (which is always the possibility tbh) or you are not even trying to consider what I am saying. Either way this is leading nowhere. I thought I had enough logical arguments to at least trigger your suspension of disbelieve and lessen your anguish over a rule that seems bad by realizing there might in fact be reason for such a rule. Aparently I do not. Worse things have happened. Maybe you think my explenations are wrong, but still, try to at least entertain the possibility that our dev is not keeping the rule to spite you all, but he actually has an understanding about the system we may be lacking.
And keep reporting the cheaters I think we might be able to agree on that.
One problem is that reporting cheaters is really quite difficult in this type of cheating.
It takes a fair amount of watching, waiting, and deducing, because the record of who did what and when is sketchy.
And so full disclosure: I havenât tried reporting one of these yet thought I have seen a few.
I guess I have to resolve to start reporting them, since I care so much about it
OK. If anyone has instances, letâs take a look at them and report them.
GaJ
⌠and straight away I am reminded how hard this is, going to look at the first suspected one I know of.
Iâve thought before that we could do with a âlast seenâ line on a personâs profile: the time at which they were last seen on the site. That would be a massive leap forwards in tracking this sort of thing down.
But we also could do with the time showing when the last move was played in each game (indeed the time of each move might be cool!?)
Indeed I would stop arguing. The opinions are so divided that I think no further discussion could resolve the difference.
Therefore I call for a vote to settle the matter once and for all.
Let all registered member, or paid site supporters if the admin thinks fit, vote on this matter and let the people decide.
That might be the right thing, though actually this isnât a democracy. Actually, itâs a dictatorship. Anoke decides whatâs best for the future of the site, based on his considertion of input and his analysis of the issues.
A vote is a way of showing Anoek what people think, but Iâm not convinced itâs the best sort of input to make a decision like this. The problem with democracy is that it assumes people are educated.
In fact, even us present in this debate are not yet fully educated on the topic: Iâm not even ready to vote, and Iâm paying as much attention as anyone. Some of the reasons âwhy it is the way it isâ absolutely need to be addressed by any proposal for change. I havenât seen that yet.
What I have been intending to do is summarize the arguments so far. I might do that on the weekend - or someone else could free to give it a shot.
Not trying to be mean or anything, but if he once was, he is no longer, once he started asking for and accepting donations.
And I donât want to get into politics but when the government starts saying that the people is incompetent and shouldnât be allowed to vote, âŚ
never mind.
You are entirely mistaken about that.
Your donation is just that: a donation. It does not entitle you to anything, other than the ability to withdraw it if you donât like what it is going towards.
Unless you are donating millions to the campaigns of US Congress people. Then your âdonationsâ entitle you to all sorts of stuff.
One simple step to move forward would be to alter the codex now so that all Timeout wins and loses are ranked, and see how it plays out doing it that way. I assume, but do not know, that this would involve removing the code that scores these situations differently (as opposed to rewriting new code).
Letâs see how it plays out when OGS counts all wins and losses as ranked wins and losses for a couple months and go from there. If global ratings, timeout challenges âfor causeâ, Mod requests, OGS Forum complaints reveal an issue, re-introduce the codex to not rank serial Timeout wins again. If not, leave it as all games are ranked.
Just to drop in my .02$ into another controversial topic. Iâm a simplistic person and I do like simplistic solutions. So in that sense I completely donât get the purpose of special timeout rule. If a person has âlife circumstancesâ that make him/her abandon all their games, why would they care about the outcomes? Iâm stating this not as an argument, but to make it clear where I stand in the discussion. Youâll find the actual .02$ in the next paragraph.
We need more scientific approach to the whole problem. Exchanging incidental examples and what-ifs leads to argumentation stalemate.
Step 1. Make the rules clear (admins).
If we donât know what the rules are, we are just wasting everybodyâs time on the forum. This is the title of the thread, and many people have agreed with this point. Are the rules documented somewhere already? Admins, can we have your assistance here, please?
Example: Every subsequent timeout in corr games is voided.
Step 2. Present the spirit of rules (devs)
Here some input from people in the know is useful. Rules however silly they look, may protect important idea. It would be good to have that idea phrased as well, so that it could be questioned.
Example: Correspondence games are statistically vulnerable to timeouts and without the house rule, rating system is unreliable.
Step 3. Discuss effectiveness of the rule (community)
Only with above 2 bits of information at hand, you can actually evaluate if the house rules (Step 1.) do any good and if they address actual problem (Step 2.).
Example: Current rule promotes abuse and leads to gaming the system. While the statistical impact of timeout on rating system is relevant, to address the root of the problem, rules should encourage people to NOT timeout from their games, rather than the opposite.
Step 4. Improve the rules. (admins, devs)
Based on the discussion it should be easy enough to look for constructive ideas that improve the rules, while maintain their spirit.
Example: Timeouts are voided retrospectively only if player times out of ALL his correspondence games in limited time span and presents no other activity in that time, i.e. does not start new games or play any moves in other games
yes, i mean if nothing else, at least have the thing written out in black and white. For all that matters, we arenât even sure what we are complaining now is real or not.
I especially like your example in step 4. This seems like a very good idea.
Ok, let me try respond as good as I can, but be patient with me.
Borrowing your quote: Every subsequent timeout in corr games is voided.
Yes. This rule is still in effect.
To the extend of my understanding. Especially in correspondence games where one player can time out tenths of games at once it can deliver a severe blow to the ranking precision, since suddenly several players can get an undeserved boost. We all agree that this solution is not ideal, but at the moment it is still on, because the simple fix of counting every timeout as loss could have a very bad impact on the ranking inflation/deflation.
To me this sounds reasonable. In time, I will try to talk more with anoek on this, but being the new guy, I donât want to ask too many qustions too quickly and have other questions I need cleared first
on a related note:
The current documentation is here https://ogs.readme.io/ but we do not even link to it, since it is VERY outdated. Good documention is important in my opinion, but it is a LOT of work on something that is changing quickly, so realistically speaking it will probably never be done by one person. The page can be community driven (similar to wikipedia) so if there is enough people willing to help work on it, shoot me a PM and given enough help and some management we could maybe make it happen
Several players getting an undeserved boost?
a) Not counting Timeout wins gives the players an undeserved ding! Not counting serial Timeouts is just as detrimental to the rating system as counting them.
b) Not counting serial Timeouts leads to deliberate cheating which is detrimental to the rating system; counting Timeouts eliminates deliberate cheating which improves the rating system.
It is not. At least as far as I understand. I tried to explain it somewhere above.
yeah, we know, thatâs why we are discussing it
Statistically/mathematically, it is equally detrimental in the opposite direction to the rating system. As least as I understand.