I guess I’m saying:
And by the way, punishment isn’t the same to everyone. I feel punished when my rank is adjusted in a way that is known to be inaccurate. Maybe person X wants to be rated higher than he actually is, but I don’t.
The problem is, there is no practical way to “show” that. If anyone can think of a way, I would be interested to hear it.
I based my statement on my experience handling a number of serial-timeout cheating cases, in conjunction with GAJ, back in 2019, and on subsequent observations when I was watching a lot of games and filing a lot of reports. More than a year ago, I saw the worst case ever, in which someone had five long strings of timeouts in four months. I don’t believe someone had five incapacitating catastrophes in four months, so that case was either conscious abuse (cheating) or unconscious abuse (habitual escaping). The two mods I involved in this thought that the presence of one or two wins exculpated the player. If anything, that suggests conscious abuse, because the player was willing to sacrifice one or two wins in order to wipe many more losses off their record. Recently I saw a similar case on a slightly smaller scale, but didn’t report it—why bother?
That recent case, however, did inspire me to take a random tour of correspondence players, and I found that a large number of them habitually time out their losing games. As I alluded to earlier, because timeout is no longer a reportable offense, it has proliferated in the culture, which means a considerable increase in unconscious abuse of the rule. This leads to a philosophical question: if timeout is not a violation, how can serial timeout be a violation? Would it really be okay to time out every other game?
As for the examples you provide:
Your third example has no relevance to the discussion, because that person’s games were not annulled pursuant to the serial-timeout rule. They were annulled because they had fewer than six moves.
I don’t believe your second example is a life event because several of the later games were begun after the beginning of the string. This is an example where the log-on trigger would prevent the interpretation of this as a serial-timeout string.
Your first example looks most like a trollish, or just plain weird person, resembling what can be expected from a cancellation troll. It has four strings in four months (another string was just simply sub-six-move games).
It is conceivable that someone might have a chronic medical condition leading to periodic incapacitating hospitalizations. If that is the case, then they should take the responsible action and stop joining new games. I don’t think the OGS ranking system should be held hostage to such an edge case.
I think it would just be a scaled up version of the experiment I did.
Of course, there would need to be some agreement on what constitutes abuse. Like do we need to actually detect a “catastrophic life event”? Or does simply check whether their win rate indicates rank manipulation. (The latter is easier to automate)
Such an analysis may be too much work given OGS’s resources. But let’s at least be honest about what we’re addressing - “people don’t like having their games annulled” (evidence: frequently recurring forum posts) vs. “abuse of the serial timeout rule is widespread” (evidence: minimal)
Bingo! I for one, have too few years (or months) left to undertake such an effort.
My experience as a mod is that people don’t really mind. Whenever I have dealt with a report about this, I’ve explained how it works and the answer is always “ah, ok, thanks”
I guess those properly annoyed take to forums so this gives a very lopsided perspective of how annoying it is in reality.
I might also add that I’ve never felt that any of my opponents who’ve mass timed out have done so maliciously. I’m yet to be convinced that this is a significant problem.
I thought it was clear from what I wrote that unconscious abuse (not “maliciously”) is by far the biggest category, because a correspondence timeout is no longer reportable. I’m glad to have the opportunity to clarify that.
A helpful clarification.
At the risk of going yet further off topic, I think this then comes down to
I think if timing out is just “a thing” in correspondence and not “abuse” then there is no need to sanctions.
Indeed the rule is designed to protect the timer out from tanking their rank, with the additional benefit of limiting rating system drift and future sandbagging. So the reasoning behind the rule is that this kind of event is anyway not abusive. So it can’t be unconscious abuse, it’s just an unintended consequence.
[Edit: I’m open to the scenario where someone maliciously manipulates the rule for some purpose but then I think we are into maliciously timing out at this point and a different approach is needed. I remain to be persuaded that this is a significant problem]
To try and come back on topic, the question we are trying to answer is how can we protect the timer out without their opponents feeling “punished” where it seems likely those opponents would have won.
My answer is that we cannot. Therefore either we just make the rule work as well as possible (close loopholes etc) or ditch the rule.
I think you have the question backwards. The purpose of this thread is how to protect the opponent from having wins stolen from him/her because of numerous habitual escapers and some cheats, while preserving a substantial part of the rule through the annulment of indeterminant games. This is what I called the “balance.” I think it lies somewhere between the 60-90% win rate, plus some of the other criteria that you mentioned.
Aside from that, I see no reason not to implement the log-in trigger as soon as possible.
No, I think @GreenAsJade was quite clear about the question
Nothing about protecting from cheaters/escapers all of that. Just a question about mitigating the undesirable impact of the rule operating as it does.
Whether or not the rule is abused a lot or not is a separate thing.
What exactly does that mean? It sounds like an ambiguous way of saying exactly what I said. What is the “undesirable impact” if it is not to stop the unjust loss of wins?
Someone is hit by a bus. Games are annulled. Opponent was pretty much guaranteed to win one of them and feels sad.
That is undesirable impact. Nothing to do with cheating or escaping.
And to be clear, I’m using a specific example of hit by a bus. The reason for the timeout system triggering doesn’t matter. It might just be someone gets busy at work or has a baby or just can’t play for any of a million reasons. None of that matters to the question GaJ asked.
So actually we are on the same page. The connection to escaping and and cheating is that these are the main causes.
This was certainly not the case when I was moderating. The idea was to accommodate extraordinary incapacitating events. Too busy? What a pitiful excuse for not taking five minutes and resigning the games.
The main causes of what? Serial timeout? I think we’ll have to agree to disagree.
We wouldn’t encourage this kind of rank tanking behaviour. Better to let them all timeout and be annulled!
Huh? All the games subject to the serial timeout rule ended with an escape. when bad behavior is permitted, one gets more bad behavior.
I think that the premise of the serial timeout mechanism is that the behaviour it is there to deal with is not bad behaviour. It’s not bad behaviour when you time out all your games because of an unavoidable life event … this is the premise.
Serial timeouts can be caused by
My guess is that reason #2 is more frequent that #1 (especially for people who have hundreds of ongoing correspondence games) but of course there is no way to check. Anyway, if we want to limit intentional timeouts (#3), solutions have been proposed:
Nice! This seems to be a mitigation we could put in place that supplements anything else.
(Of course, it’s only a solution for as long as vacation lasts, but it seems like it’d have to help)
This hits the nail on the head I think. We can argue if 2 is acceptable or unacceptable but the point is that it happens and we need to deal with the consequences somehow.
I’m also pretty sure that this is a suggestion that has been made a while ago.
Thinking further it also has the benefit of “punishing” the timer out but in a way that probably doesn’t matter for the people in situation 1, that is helpful for the people in 2 and annoying for the people in 3!
Ooh… It seems a super helpful feature.
As a player with a lot of ongoing correspondence games, this would have also helped with several disappointing accidental timeouts of interesting winning or undecided/even games which I’d really hoped to finish & put a lot of time/effort into over a long period of time too, so extra bonus for that.
(I’ve timed out games by only a few minutes a few times when I couldn’t get back to internet to activate vacation time, and I’ve encountered that happening occasionally with opponents of mine who weren’t logged in and timed out accidentally/due to unexpected events/life busy too, only to notice some hours later, or the next or following day, with many days of vacation left they could have activated.
It was always a disappointment, often for both players, to have the game end like that.)
So I’m sure it’d benefit players in general.
Perhaps we’d want to notify the player of the game about to time out and that vacation mode had activated, too ?
A downside of auto holiday is it means someone who has really left the site (due to death, illness, forgetting about OGS or whatever) will then make that game last for a lot longer (what is max holidays for subscriber? 60?) which would prolong tournaments waiting on that game.