I think now_and_zen has a good point though. There are some clever programmers who could probably find a way to flag some of the problem causers which would at least reduce the problem. Chess.com is halfway between the wild west and a civilized chess site. Remember yahoo chess? That was hilarious, anything went on yahoo chess.
Why can't you figure out how to thwart malicious disconnectors?
Sort:
Oh, no! Bali is muted!
@leiph18 wrote: "With 1400 minutes in a day, you'd need to dedicate resources equivalent to 24 average computers to this job."
Interesting post. Some nice data and specifics. Except, I suspect you're more focused on finding ways not to solve the problem than to solve it.
For example, you wouldn't have to crunch every game, just the ones involving disconnect or abandonment with some high % of the clock remaining, and only for players with an above average disconnect/abandonment rate.
Then, I'm sure there are optimizations. Maybe caching a reasonable amount of common board positions to be able to determine advantage, including weighing material without "crunching". To make it faster, you wouldn't fully processes any of a player's games maybe until you'd achieved a fair number of 'good guesses' for a suspected intentional disconnector.
This is what I quickly considered is in just a couple of minutes of thinking about it. I'm sure good engineers determined to resolve this problem efficiently could come up with some scheme, maybe using fuzzy logic, that got crisper as probability of an abuser increased. Where there's a will there's a way.
And, as for your post about new players and false positives - similarly, since you couldn't crunch one game and reach a conclusion, accounting for various outliers and corner cases and noobs, etc... some sort of reasonably accurate conclusion could be algorithmically or heuristically determined, I've no doubt about it at all. And I'd be willing to bet the effort to figure out a solution like that would be easier than developing a good chess engine by a circular mile.
A practical advice: There almost arent any disconneters in 5 minute blitz.
I've never seen any in games with increments either.
I think the disconncters just play 10 minute chess, because they are people who are very casual with it much and just use the setting offered to them by the website.
I got almost completely rid of them by leaving 10 minute blitz.
Agree they are a pain in the *** and the site owners should do more to get rid of them, for whetever reason they just wont do it. I dont understand myself either why they dont care since its their own site, their own product, that gets damaged by the disconnecters. But well I guess they are making enough money on the site as it is and dont care if it sucks a bit here and there and there are some unsolved problems.
@Raspberry_Yoghurt - And it is true, as someone pointed out earlier, at the "higher" ratings (e.g. 1250+, which I realize isn't that high), disconnectors become rare. Nevertheless, for whatever reason, on a bad night I can put up with, I'm guessing, 5-8 disconnects or game abandonments, and it gets frustrating, because I want to keep going.
Anyway, whatever the rating, or reason, we both agree (as do many many other blitz players, obviously), intentional disconntectors are a royal PITA. So it really seems to be nothing but a plus for chess.com to invest effort into making the problem virtually non existant.
louie_grenouille wrote:
I think now_and_zen has a good point though. There are some clever programmers who could probably find a way to flag some of the problem causers which would at least reduce the problem. Chess.com is halfway between the wild west and a civilized chess site. Remember yahoo chess? That was hilarious, anything went on yahoo chess.
They could use a system like reddit where you can downvote or upvote things. I sugest:
After a game has been playes, both players have a time window of 2 minutes to upvote or downvote the other player.
Players then get paired up against players with equal amout of down/upvotes withint brackets.
For instance if your score was -13, you'd get paird up with players in the -10 to -20 bracked.
Over time disconnecters and trolls would mainly get to play against each other. And well behaving people would get to play against each other. VOiLA everyone is happy. The people that like a yelly, insulting, disconnecting
Should be super easy to make. It's standard stuff for MUCH smaller websites really. And there are many other ways to do it. Websites have developed anti trolling/bad behaviour systems for 20 years lol.
@Raspberry_Yoghurt - I wonder if that's too subjective for this particular sport? It's chess, not a popularity contest per se.
I don't want it to not work, I was just curious was some simple math would show, and then posted it.
Yes, there would be a lot of chances to optimize it. So then you're weighing cost vs how much value you'd get e.g. what % of blatant abusers would we catch?
One discouraging thought is with a system that quickly punishes abusers people may delete their account and make a new one. Or perhaps if the penalty is something like a 24 hour ban, then they cycle between 5 separate accounts.
Although this is a simple work around, I do suspect most abusers wouldn't go to this much trouble.
now_and_zen wrote:
@Raspberry_Yoghurt - My playing skill varies a lot, based on how tired I am, mood, responsibiltiies, distractions, etc... On a good week I can play 5 minute chess above 1300. On a bad day I can go on a losing streak that gets me down toward 1100 and I have to fight my way back again.
And it is true, as someone pointed out earlier, at the "higher" ratings (e.g. 1250+, which I realize isn't that high), disconnectors become rare. Nevertheless, for whatever reason, on a bad night I can put up with, I'm guessing, 5-8 disconnects or game abandonments, and it gets frustrating, because I want to keep going. It's bad enough that I have to waste too much time on blitz, but to have to waste time wasting time on disconnectors while wasting time on blitz chess is over the top of over the top.
But, that notwithstanding, we both agree (as do many many other blitz players, obviously), intentional disconntectors are a royal PITA. So it really seems to be nothing but a plus for chess.com to invest effort into making the problem virtually non existant.
Yeah I know there are ridicioulous amounts of them on lower rated 10 minute blitz.
For me it worked to quit the standard 10 minute blitz. The site owners dont care, so I suggest you play with the settings and find another playing modes that suits you. They site owners prefer 10 minute blitz to be useless and it's their site *shrugs*
Online chess works almost the same if you find someone online to play against.
now_and_zen wrote:
@Raspberry_Yoghurt - I wonder if that's too subjective for this particular sport? It's chess, not a popularity contest per se.
I think the random downvotes from people that voted down because they dont like your avatar or because they feel like it for no reason whatever would be so few that on the whole, obnoxious players would have low rating and well behaving high. Reddit works very well that way so I dont see why it shouldnt work here also.
@Raspberry_Yoghurt - The site owners don't care? Are you sure? I wonder if we could get a statement from them. If the attitude is like Comcast where, due to lack of competition they have a captive audience and good income, and don't care what customers who pose an insignificant threat of attrition think, then that would make me inclined to find a competitor site and vote with my $. But I wouldn't want to prematurely conclude that.
Apr 14, 2015
0
#73
now_and_zen wrote:
Bottom line is it is against chess.com's policy to abandon games or disconnect, and it is an annoyance. Further chess.com posts stern warnings about it frequently. So far, the people replying say, to hell with doing anything about it (even if something can be done), let's let thousands or millions of live chess players all repeatedly endure disconnect abusers pointlessly because that's how tournament chess works or something. Meanwhile over and over again people get annoyed by it, leave nasty notes in violator's profiles, and come here to complain.
If I ran this site and saw so much displeasure at a problem I could solve I would probably try to do something to fix it. That's just good business.
I'm not sure if this has been mentioned but the site does do things about it. First, accounts get restricted for abandoning games. Second, players that are habitual disconnectors/abandoners get put into their own playing pools. I would have to search out the specific places where this is mentioned but the purpose is that auto-paired seeks will tend to pair disconnectors with other disconnectors.
It isn't 100% perfect and new accounts and people that haven't crossed the threshold still might get paired with "fair" players. I don't play a ton of live, so I don't know how well it works in practice.
Why would you want to
@Martin_Stahl - not sure if you read the OP of this thread, but the thing that pissed me off and got me to post here was noticing that the guy who abandoned the game leaving me with approx 2 min or more on the clock has almost two years of complaints backlogged in his notes. And I get disconnected on or abandoned, some nights several times. I think there's room for improvement, and I've suggested some possible solution approaches in this thread, as have others. If the site was interested in resolving this, it would be great if they could have a moderated thread where people proposted technical solutions or management approaches, or maybe if the NIH sentiment was strong enough, just incent their engineers to figure something out. I'm sure this can be improved substantially, making a lot of members a lot happier in the process. It really is my only gripe with this otherwise great site. Oh, and I think the phone app should have a setting that lets 'advanced' users disable the resign confirmation requirement and have a quick-resign icon to avoid accidental abandonment/disconnect.
Martin_Stahl wrote:
now_and_zen wrote:
Bottom line is it is against chess.com's policy to abandon games or disconnect, and it is an annoyance. Further chess.com posts stern warnings about it frequently. So far, the people replying say, to hell with doing anything about it (even if something can be done), let's let thousands or millions of live chess players all repeatedly endure disconnect abusers pointlessly because that's how tournament chess works or something. Meanwhile over and over again people get annoyed by it, leave nasty notes in violator's profiles, and come here to complain.
If I ran this site and saw so much displeasure at a problem I could solve I would probably try to do something to fix it. That's just good business.
I'm not sure if this has been mentioned but the site does do things about it. First, accounts get restricted for abandoning games. Second, players that are habitual disconnectors/abandoners get put into their own playing pools. I would have to search out the specific places where this is mentioned but the purpose is that auto-paired seeks will tend to pair disconnectors with other disconnectors.
It isn't 100% perfect and new accounts and people that haven't crossed the threshold still might get paired with "fair" players. I don't play a ton of live, so I don't know how well it works in practice.
Hehe I dont believe that ANYTHING is done because
a) You come across accounts that have 10-15 pages with complaints on their profile page about them disconnecting that spans over years. Quite obviously the site allows them to play (and disconnect) against other players. If they were somehow constrained by a system, they shoudlnt be able to do it for 4 years. They are, thus i conclude there is no system.
b) That there are so many disconnecters, I'd say 10%-20% for 10 minute blitz rougly 800-950 rating. I have tried playing 4 games in a row where playing time was like 8 minutes total, and wating time 32 minutes because all of them disconnect after a few minutes. YAY for a site thet claims to offer chess but instead offers sitting looking at a clock and an abandoned chess board. If something was done there quite obviously wouldnt be so many of them.
now_and_zen wrote:
@Raspberry_Yoghurt - The site owners don't care? Are you sure? I wonder if we could get a statement from them. If the attitude is like Comcast where, due to lack of competition they have a captive audience and good income, and don't care what customers who pose an insignificant threat of attrition think, then that would make me inclined to find a competitor site and vote with my $. But I wouldn't want to prematurely conclude that.
Well my logic is this that:
If they cared they would do something.
If they didnt care they do nothing.
In fact, they do nothing, so they must not care.
Third options is of course they want to do thing but are too stupid to know that there is a gazillion different solutions to "bad player behaviour" developed in all sorts of games. I dont see that as very likely though lol. Everyone today must know there are thousauds of games with thousands of millions of players on the internet.
now_and_zen wrote:
@Martin_Stahl - not sure if you read the OP of this thread, but the thing that pissed me off and got me to post here was noticing that the guy who abandoned the game leaving me with approx 2 min or more on the clock has almost two years of complaints backlogged in his notes. And I get disconnected on or abandoned, some nights several times. I think there's room for improvement, and I've suggested some possible solution approaches in this thread, as have others. If the site was interested in resolving this, it would be great if they could have a moderated thread where people proposted technical solutions or management approaches, or maybe if the NIH sentiment was strong enough, just incent their engineers to figure something out. I'm sure this can be improved substantially, making a lot of members a lot happier in the process. It really is my only gripe with this otherwise great site. Oh, and I think the phone app should have a setting that lets 'advanced' users disable the resign confirmation requirement and have a quick-resign icon to avoid accidental abandonment/disconnect.
For all the little bugs they still have (plus V3 is something like a year late now) I have to wonder how many capable people they have on staff.
Sure Erik could hire more (and/or more qualified people) but then he'd have to take a pay cut right?
I guess my POV is a bit pessimistic.
@leiph18 - I wondered about the status of V3, since they had requested community input a long time ago. But in this industry (sw engineering), it is very challenging to keep up with bugs and new features. Even major companies with huge budgets and IT departments struggle. Consider Apple - one of the richest companies in the world - lots of bugs on all platforms, but superb engineers and technologies overall. And I'm sure Google has the same problems. So a site like chess.com has to make due. If the effort into bug fixing is going into V3 and we can expect improvement on the disconnect issue and they're working on it, great! Otherwise, this is one problem that I think should be fairly high on the priority list to resolve, just because it elicits so much frustration in the community and is an ongoing "evening of mosquitos" scenario.
Also I'm not sure how you'd set up the flagging. Most people use extra time after an error. And not all abusers immediately stop making moves (some continue to play a lost position and stop 1 move away from mate for example). What's the threshold for abuse concerning eval drop and time used?
Another difficulty for false positives would be new players. Maybe they see mate coming, but they try to figure a way out of it and lose on time. Many weak players aren't good enough to recognize when it's unavoidable. Sure you could only flag higher ratings for abuse, but as you've pointed out it's more prevalent at lower ratings.