AFK Concede

@naissun said in AFK Concede:

@vejeta said in AFK Concede:

I thought it was said before that going AFK is not against any rules so therefore would not be punishable.

That would be interesting. Where did you see that?

I saw that from this topic here:

https://forums.focus-home.com/topic/46/ps4-champion-ladder-rules-season-9

"To be absolutely clear, fouling is allowed at any time you are allowed to do so. So is stalling (i.e. clock management). These are not exploits, or considered unsportsmanlike in this league"

Edit: Its just occurred to me that maybe this is talking about choosing when to score when it talks about stalling and clock management, so maybe I misinterpreted it if that is the case.

last edited by vejeta

Yep, that's specific to choosing to not score rather than deliberately wasting time. Thing is with AFKing that it's very difficult to show it's happening. In the case specific to this thread the chatlog showed the guy had intent to do it.

@naissun said in AFK Concede:

@vejeta said in AFK Concede:

I thought it was said before that going AFK is not against any rules so therefore would not be punishable.

That would be interesting. Where did you see that?

The admins of CCL said (in 2016), that they won't moderate for it and that this is not a problematic behavior. The background of it was, that there is no reliable, easy and fast way to identify games in which a player goes afk (for concessions it is just checking a column in the data base). A longer discussion about it is here.

I think, the main question for the implementation of a punishment system is still the same: How do you identify games in which a person went afk?

last edited by Arne

This is a very important problem and i wanted to write somethings too 🙂

This is a game and there should be freedom to go afk, got angry, cry or wine ... imo even insults should be ok.

@coldy said in AFK Concede:

This is a very important problem and i wanted to write somethings too 🙂

This is a game and there should be freedom to go afk, got angry, cry or wine ... imo even insults should be ok.

I disagree, It is a game but also a moderated competition.

Rules that are implemented are choosen with care to ensure most people may have fun. But yes, when enforcing it, without being lenient, give the other one the chance to give his version and issue warning before a permanent ban are a minimum basic wisdom to my opinion.

last edited by JRCO

@naissun said in AFK Concede:

trust me.
No disrespect but why should I?

No disrespect but I do not really care whether you trust me or not. This sentence ("Trust Me") was what we called a "Figure of speech", I am a little bit amazed you seems to take it personnally.

If one day you receive one, you will understand what I am saying.

Do we know each other?

Nope, but why are you suddenly becoming agressive and ungentle ? I would like to understand what I wrote that did angry you. How a genral statement can be felt as an offense to you in particlar ?! That would be very kind of you to explain (But yes why should you be kind ?....etc...LOL).

If you want to discuss which of a warning or a ban is more appropriate in our case, I'm glad to do so but your assumptions about my person are not a proper argument.

Where in the hell have I made assumption about you ?! My answer was a general statement and didn't point to you in particular. There ain't any reason here for you to become emotional.

last edited by JRCO

The rules for CCL are clearly posted for anyone willing to look. I feel like that's the warning.

Why would anyone play in a league that they didn't understand the rules? I don't think it ask a lot of a player to understand the rules of the league they are playing it. That's my 2 cents.

@arne said in AFK Concede:

I think, the main question for the implementation of a punishment system is still the same: How do you identify games in which a person went afk?

I'd suggest the same thing I suggested back in 2016: you wait for someone to report it, have them submit the replay file, and then see how much activity the accused person did or did not engage in on a turn-by-turn basis. You don't need to watch the replay file, you can stuff it into onesandskulls and look at the roll log to see if someone is actually playing or not... and that's just one way to go about it if nobody is willing to further automate the method of investigation.

Now, as then, it has never been an issue of can't, it is purely an issue of won't.

@voodoomike said in AFK Concede:

@arne said in AFK Concede:

I think, the main question for the implementation of a punishment system is still the same: How do you identify games in which a person went afk?

I'd suggest the same thing I suggested back in 2016: you wait for someone to report it, have them submit the replay file, and then see how much activity the accused person did or did not engage in on a turn-by-turn basis. You don't need to watch the replay file, you can stuff it into onesandskulls and look at the roll log to see if someone is actually playing or not... and that's just one way to go about it if nobody is willing to further automate the method of investigation.

Now, as then, it has never been an issue of can't, it is purely an issue of won't.

Problem with this approach is, that it is quite common, that a coach just spams the end-turn-button in the last turns, if he doesn't want to play anymore. This would look like an afk in the replay

Didn't you suggest this activity score the last time?

last edited by Arne

@arne said in AFK Concede:

Problem with this approach is, that it is quite common, that a coach just spams the end-turn-button in the last turns, if he doesn't want to play anymore. This would look like an afk in the replay

Replay files also include timing information so it's possible to tell how long they took before the turn ended. That's probably not available on onesandskulls, certainly, but someone spamming end-turn rather than waiting will result in a shorter match length which we can see in its record on goblinSpy.

Similarly, just using onesandskulls we can see if they only took no actions at the very tail end of the match or if it was constant throughout. Someone would have to decide how much AFK is too much AFK beforehand, and they'd use that as their method for differentiating between acceptable and unacceptable away-ness.

@arne said in AFK Concede:

Didn't you suggest this activity score the last time?

I suggested the creation of an activity score for investigating the prevalence of AFKing, yes... since we have fields in each match record that cover how long the match lasted and various actions that each team can take. A calculated index based on dividing prorated action values by the duration of the match could give us a means of finding the most overt examples of AFKing, and could feasibly be used as another means of verification.... It's not guaranteed to find every example, or to never have false positives, but combined with someone reporting it as AFKing is probably more than reliable enough.

Even testing for deadly, communicable diseases has false positives and false negatives... but we don't use our inability to be 100% accurate all of the time as an excuse for doing nothing; when we do, we're really just trying to excuse our own laziness.

@voodoomike said in AFK Concede:

@arne said in AFK Concede:

Problem with this approach is, that it is quite common, that a coach just spams the end-turn-button in the last turns, if he doesn't want to play anymore. This would look like an afk in the replay

Replay files also include timing information so it's possible to tell how long they took before the turn ended. That's probably not available on onesandskulls, certainly, but someone spamming end-turn rather than waiting will result in a shorter match length which we can see in its record on goblinSpy.

What kind of timing information does it include?
Would it be possible to write a programm, which gets all replay files and checks for afking in them?

That sounds far more reliable than just going by the columns of the data base, but I am not certain that this can be done so easily, I have no experience with the replay files, I don't even know where you can download them

last edited by Arne

@arne said in AFK Concede:

What kind of timing information does it include?
Would it be possible to write a programm, which gets all replay files and checks for afking in them?

Replay files store individual actions taken during the game, so... every action you take that would alter the game state in any way (which would show the other person you doing anything) is recorded, and part of the recording is the number of seconds from the start of the match at which the action occurred, as well as which turn it was on.

It'd be possible to make a problem that took a replay file and looked for someone taking no actions during their turn and just letting the timer run out, yes. It's not so easy to just scrape all replay files off Cyanide's servers, though, and they only keep replay files for 7-14 days before deleting them to conserve space.

The obvious medium would be to let people submit complaints about facing an opponent doing AFK griefing, and submit the match replay file with it. An automated system could theoretically verify that AFKing had occurred according to the content of the replay file, and then signal a moderator with the relevant information in order to make a judgement call.

Why a judgement call? Well, we're talking about an XML file... one that people could conceivably edit. There needs to be some checks and balances in place to avoid that possibility (though, given the general technical expertise we see in this community I'm not convinced its a serious danger... I err on the side of safety).

@arne said in AFK Concede:

That sounds far more reliable than just going by the columns of the data base, but I am not certain that this can be done so easily, I have no experience with the replay files, I don't even know where you can download them

I have never advocated using statistical analysis of match data exclusively to track down AFKers, simply as a means of investigation. While some people farcically believe in absolute certainty, the realistic path to handling these things is obviously going to be bayesian in nature; we take multiple things that suggest a phenomenon and let them cumulatively bolster our confidence in its existence.

@voodoomike said in AFK Concede:

[...] and part of the recording is the number of seconds from the start of the match at which the action occurred, as well as which turn it was on.

It sounds like that it would be possible to identify afking with replay files then - only the pause function or connection issues may cause some problems.

@voodoomike said in AFK Concede:

It'd be possible to make a problem that took a replay file and looked for someone taking no actions during their turn and just letting the timer run out, yes. It's not so easy to just scrape all replay files off Cyanide's servers, though, and they only keep replay files for 7-14 days before deleting them to conserve space.

Why is it not so easy to scrape all replay files? Where is the limit? (Like I said: I don't know the procedure for getting replay files). If it is not possible to download all, would it be possible to get a bunch? If it would be possible to get a bunch of them and check them for afking, it would at least be possible to do some statistics regarding the occurrence rates of afking. Furthermore, such a data could be used to develop a classifier for the matchmaking data - it would even give a estimation for the sensitivity and specificity of the classifier.

last edited by Arne

@arne said in AFK Concede:

It sounds like that it would be possible to identify afking with replay files then - only the pause function or connection issues may cause some problems.

I'm sure the pause function and reconnection stuff are listed in the replays somewhere since they'd register as visible actions on both people's screens... though I've never used the pause system. Even if it did not we could assume that legitimate use of those two systems wouldn't be a "no action every turn" thing.

@arne said in AFK Concede:

Why is it not so easy to scrape all replay files? Where is the limit?

As I understand it from what has been said by the people who download them directly, Cyanide doesn't want their servers bogged down with tons of downloading. Relay files are at least half a meg in size for completed games (at least... often more) so taking any serious number of them would be quite the download, even assuming they're available in some batch-downloadable way.

Mordrek's site doesn't download them except when someone requests it - goblinSpy generally only works from match data with the option to snag recent replay files for certain leagues.

@arne said in AFK Concede:

If it is not possible to download all, would it be possible to get a bunch? If it would be possible to get a bunch of them and check them for afking, it would at least be possible to do some statistics regarding the occurrence rates of afking

Seems like a lot of downloading and work for what... trivia? What does the rate of AFKing matter in the large scale? If we consider it a problem then we want to address it regardless of how widespread it is. You're welcome to hit up Cyanide for batch replay downloading, but you're still stuck with having to make a program to automate determination of AFKing in order to create a dataset to then investigate for what you feel to be "reliable" occurrence information.

@voodoomike said in AFK Concede:

I think maybe a warning the first time an then eviction from the current season might be a better way to go. Yes, it's a douchey thing to do, but it's always best to let people know that they're doing something wrong on the chance they don't realize it (they had to put warning on dish soap that it wasn't for drinking, after all). By "wrong" I mean "against the rules" not "being a dick" because really, the latter is just modern society.

There is an opinion that modern society is like that exactly because people started to care about some feelings too much. It's "action-feedback" sort of issue, some people just do not learn how it works as their parents apparently too modern to explain it them in a more... straightforward way, when needed. Some intermittent mischief inflicted out of passion can be taken lightly, for the first time, but this is not the case. Here we have a person who clearly understood what he was doing and was in perfect control of his actions, as he was sitting there doing it for more than an hour. I even dare to say that league would be better without a person like this. People like that rarely are of any good to anybody.

@Arne @VoodooMike - we tried the thing with the replay files. I enabled the downloading of batches of replay files from CCL and mordrek wrote a program which output the length of each turn and the number of actions taken by a coach as a line in a csv file, giving us an output table for actions a turn lengths for each game. From there I wrote a program to determine mean and median turn lengths and number of actions for each turn, giving me the ability to pick out any anomalous matches.
The problem was an awful lot of tweaking was required due to inconsistencies within the files meaning that particularly the last turns (the ones we were most interested in) appeared corrupted. These inconsistencies only really became apparent when comparing large numbers of files for their output. After a few weeks of trying by mordrek we stopped as it was a lot of effort for little reward.

It'd be practical to build an AI to identify suspect behaviour.

Just don't let Cyanide design it...

@dode74
Ah, thank you for the information Dode

last edited by Arne

@mori-mori said in AFK Concede:

There is an opinion that modern society is like that exactly because people started to care about some feelings too much.

There is an opinion that vanilla is the best flavour of ice cream... and one that the world is flat. People have been dicks since time immemorial, and widespread concern about people's feelings is a modern thing. Being dicks doesn't particularly mesh with high concern for other people's feelings, but it does with the inherent selfishness of people that we see in every society since we began recording history.

@dode74 said in AFK Concede:

The problem was an awful lot of tweaking was required due to inconsistencies within the files meaning that particularly the last turns (the ones we were most interested in) appeared corrupted.

If the replay files manage to accurately reproduce the match then the replay files are not corrupted or inconsistent. What it means is that the parsing of the files was imperfect and that led to producing inconsistent output.

@voodoomike said in AFK Concede:

People have been dicks since time immemorial, and widespread concern about people's feelings is a modern thing. Being dicks doesn't particularly mesh with high concern for other people's feelings, but it does with the inherent selfishness of people that we see in every society since we began recording history.

That's why people have been doing their best to fend off those who tend to be dicks from their communities, inventing all those sophisticated codes of conduct and rules, with appropriate punishments. This worked pretty much fine until somebody thought that feelings of those who deliberately chooses to be a dick should be regarded at the same value as feelings of those who can control their dickness - probably because it sounds fair and full of justice, or some else equally retarded reason. Clearly not because it's practical and helps a community to maintain some standards in what it tries to achieve.