Intentional Disconnections

[uday]

So I'm running this software that logs some data on tourney players to look into this Intentional Disconnect problem

 

here are some stats ( only 18 hours of data collection )

 

3613  (87%)     (registrants and subs) completed their events
535   (13% )     people (registrants and subs)  failed to complete their events

 

 

The vast majority of the people who bailed did so once, although about 1 in 7 of the bailers bailed more than once

 

 

What I'm planning to do with this data is implement the following rules.

 

if the bail-to-complete ratio over the last (?) days exceeds (?) then revoke the users ability to join tourneys for (?) days. At the end of that period allow the player back into tourneys. If the player has a repeat "offence" then suspend the player from BBO for (?) days.

 

 

I do not distinguish between people who log out, who sufer connection errors, who get subbed out. All of these are "bail events", and are black marks unless balanced by completion events ( logging back in and returning to your seat )

 

 

Opinions? What do you think are reasonable values for the question marks?

I expect the code to begin to trickle into production over the next few days, once i have data, and once the users have been told this is going to happen.

[hrothgar]

Hi Uday

 

I like the suggestion. With this said and done.

 

When presenting this to the membership at large, I think that it is important to "spin" the messaging. Make sure to emphasize that the disconnections counted include both accidental and "deliberate" disconnects. Regardless of motivation, both impose a substantial cost on tournaments.

 

I'm not sure whether we have enought information to draw a meaningful conclusion regarding the disconnect limit. It might be best to structure this dynamically. For example, if you are more than 2 standard deviations from the norm you get a warning. More than 3 standard deviations and ...

[JRG]

Since I don't generally play in tournaments, I suppose my opinion should not carry too much weight.

 

Since it is clearly very disruptive when people bail from a tournament, it makes sense to take measures; however, you are right to allow a little latitude as there will, I am sure, be unintentional disconnects by people who are generally well-behaved tournament players. Just what that flexibility should be seems to be a bit of a guess. Richard's approach is a sensible one, unless someone comes up with something more logical. One reason I'd go along with it is to keep things objective and avoid straying too far towards being arbitrary.

 

Question: Would this also apply to Team Matches, or just Tournament Events?

 

My guess is that Team Matches would tend to be more self-policing and people that bail do not get invited back.

 

What about applying a similar approach to such behaviour at the table? Is it even possible? I don't play pickup games very often, but when I do, it is intensely irritating when a player leaves his partner stranded and somewhat embarrassed. I've been fortunate. I believe it has only happened once to me once (i.e. MY partner bailed - cross my fingers and hope the incidence doesn't rise); however, I see it happen considerably too much by opponents.

 

It has to leave a sour taste in people's mouths. I hate to think we might actually lose some aspiring bridge beginners because of such behaviour (I personally know people who stopped playing face-to-face duplicate because of other (perceived) hard-to-prove unethical behaviour that simply spoit the experience regardless of whether they won or not).

[uday]

Just tourneys (not teams).

I want to see how this works for tourneys.

 

In the long run we have to have an equally automated approach for open table games. I dont know what that might be but i will look into this once the tourney bailers have been dealt with

[Dwingo]

I have observed this happening to me sometimes. I get disconnected (not intentionally ) from my ISP and by the time I get back ( usually in less than 2 minutes), I am already substituted from the tourney.

 

This may be due to the 2 reasons. 1) My partnership is usually an online pick up partner, whom I don't know, and he may not wait for me to return. He is probably assuming that I am bailing out. 2) The efficient subbing software programme makes it easier for Directors and players to make a substitution in no time.

 

Now my question is. I am not a finisher in this tourney even though I have returned. Am I counted as a bailer?

[uday]

Yes, you would be counted as a bailer.

 

The reason is that it doesnt matter to the TD or the opps *why* you left.

If the TD is forced to (chooses to) sub you out then you have bailed, as far as he is concerned.

 

But i don't think single incidents will count so much, as long as you have other tourneys where you complete the event. I will be checking something like the last 7 days, seeing if you (say) finished more tourneys than you started or some such.

 

 

 

Current stats ( 1 or 2 days )

 

Total 8,700 players entered some tourney or the other

1,100 did not finish

[luis]

Idea:

When a player at a tournament or at the table decides to leave the table either by standing up or trying to exit the application send a message to the tournament host or the table host "foo is trying to leave the table. Ok?" If says yes then he is out, if the host says "no" then display to the player "The host doesn't allow you to leave this table/tournament, leave anyway ?" If he leaves without permission then log the incident.

This will educate players to say "bye" or indicate a problem and leave without incidents.

For tournaments it can be used when a player has an emergency or the tournament has extended beyond planned and he must go, he apologizes to the host telling the problem and can leave and be subbed without a warning.

[uday]

I have not given Main Bridge Club bailers much thought yet. I dont like the idea (currently) of asking the host.

 

But, again, it doesnt matter why a player bails. it only matters that he only completes 50% of tourneys he enters or whatever. When this number is too low, he shouldnt be allowed to enter Ts. One trusts that "emergencies" are rare enough that they wont affect a players overall ratios.

 

The education will begin shortly, once we get around to updating the lobby news. I might also send reminders to people who bail, via bbo_mail, and might mail them again as their bail-ratios worsen and come close to getting them banned from Ts.

[square]

This might be a silly comment but the figure you quote of 1,100 out of 8,700 seems high. You are allowing for the survivor tourneys where not all players complete them, by definition ? :(

[JRG]

Dwingo: I suspect you are counted as a bail-out.

 

Try and catch me when I'm online. I'd be willing to play the odd tournament with you. Also, I prefer to wait a couple of minutes (or even more), unless I am positive partner jumped ship (semi-mixed metaphor?).

 

What I'd be willing to do if the opponents were antsy, is to call the director and suggest he give us an average minus (and opponents average plus). Seems a reasonable punishment without being unduly harsh and it would keep the event moving.

 

Come to think of it, that might be a reasonable approach for TDs to take if a pair is holding up the game (i.e. give them an average minus and force them onto the next board).

[ehhh]

I like the concept and generally think the standard deviation to flagging a way of handling the discipline steps suggested.

I am concerned though that using a time frame (for example 1 week) for measuring the BO (bailout ratio) somewhat unfare to those who may only play a few tournies a week.

Now if in that week they have a legit disconnect problem their BO would be comparitably high and would likely be subject to discipline - even sitting out the next time frame.

Could the BO factor be just a function of the number of games entered and the tracking doesn't kick in til after a set number of tournies have been entered (xx)?

[uday]

I havent done the code that actually boots yet, so sure, i agree that time alone is not sufficient, maybe we need a min # of plays as well.

 

I dont know anything about how to compute/use the std deviation, sorry.

I'm working in a mysql context, if anyone wants to offer more.

[aisha759]

Hi Uday;

I think it would be a very bad idea to consider a player who just leaves a main bridge club table and logs off, as a bail out... how many hands would he have to play in one session, to consider him in the "completed" category.. i know you haven't done anything about this yet, but your posts suggests you are thinking about it.. in tournaments, i couldnt agree with you more; as part time TD and semi full time tournament player, it's a nuisance ( the intentional ones of course).... since TD'scant really tell the difference , unless the player is a notorious "crashee" measures need to be taken and you could take a poll as to how many warnings, days, tourney, bans etc... if you allow for everyone to give a number, it will be never ending; you might end up with a thousand different opinions as to what the "?" should be..

 

P.S. i thought yellows could tell the difference between intentional and bad connection, and if they can differentiate, why punish both types of log offs in the same manner?

I personally had to log off turney yesterday, reboot, and come back; i was in luck, that i had time left in round, made up for it next round, and my partner knew my problem .( i previously mentioned i had this bug in the 3.6.7 version where i freeze/get stuck and cant see lobby nor use chat bar while kibbing or playing) should i be punished for this? it is happening more often than not lately.. that would put me in the "no no" category soon :() Ctr+alt+delete was the only way i could log off by the way..

 

Thank you;

 

Aisha

[uday]

I dont want to get into Main Bridge Club issues just yet.

 

A non-survivor in a survivor T is *not* regarded as a bailer.

 

yes, the numbers seem high, but this is why the TDs are squawking.

Right now there is no penalty for this bad behaviour..

[uday]

yellows can tell the difference (between Bail and Disc) with a little work.

Not worth the effort. if player can complete only 10% of his tourneys, who cares why? He shouldnt be allowed into tourneys, or only allowed into :Only bad connections allowed" tourneys :(

[hrothgar]

I havent done the code that actually boots yet, so sure, i agree that time alone is not sufficient, maybe we need a min # of plays as well.

 

I dont know anything about how to compute/use the std deviation, sorry.

I'm working in a mysql context, if anyone wants to offer more.

Hi Uday

 

Regretfully, I don't knowmuch about MySQL. With this said and done, here is a basic description regarding what you want to do. Please don't ascribe any specific meaning to words like "list' or any other computerese...

 

Step 1: Create a list of all BBO members, along with the total number of times that they have bailed from a tournament.

 

Step 2: Calculate the average bail rate expressed as a percentage of the total number of tournaments played.

 

Step 3: Calculate the variance: This link [http://davidmlane.com/hyperstat/A16252.html] will give you the basic formula along with an example.

 

Step 4: The standard deviation is the positive square root of the variance.

 

Lets assume for the moment, that the "average" player disconnects from 10% of all tournaments. Furthermore, lets assume that the standard deviation is .0333

 

I advocate that anyone who is more three standard deviations away from the mean should be locked out of tournaments. In this case,

 

Anyone who has bailed from more than 16.666% of torunaments would receive a warning.

 

Anyone who has bailed from more than 20% of the tournaments would be locked out.

 

I'd also suggest adding a minimum number of disconnects to avoid problems that might arise if someone was accidentially disconnected from their first torunament.

[Free]

I would also ban them for a period of time, depending on how many tournaments they play and how many dc's they have. If you play 10 tourneys each day, and you disconnect about 20% of the time, it's a whole other story than if you play 1 each day and disconnect the same amount of times. I'd suggest the more they play, the more chance they get to dc, so the banning period would be shorter. With repeated bans, I think the banned time should be expanded...

[Mirjam_3]

I believe the TD can do some work too. If i have time i send message to the player that left. Well, i dont succeed there that well, but all bits might help. And most players are kind enough to leave me message about what happened.

I cannot manage this behavior when connections are bad, of course, and i certainly hope that NO TD me ever is going to ask why i left...

 

I have a cat, u know, and this cat likes to run over the bed where i am with my laptop and a very insecure connection to the net... so the cat rans and i boot... how can i explain it??

 

Lucky for me the players in my tournaments dont have such cats or they have much better connections!

[blackest_k]

just a thought is it possible to break down the disconnects data grouping tournys by length. Are short tourneys less prone to this than long ones.

 

perhaps some people enter tourneys not fully aware of how long they have committed themself for if its clearly stated this is a 2hour tourney maybe people would be less likely to enter realising they havent the time available or the bladder control to remain sat at table I can't be the only person praying my partner will get in a contract so i can get away from the table for a minute.

 

warm regards

john

[Free]

I dont think we should protect these people (maybe a bit cruel).

 

Everybody SHOULD read the tournament description and rules BEFORE entering it. If they don't and dc later, then you get double price with the banning policy: you catch+ban deliberate dc'ers AND you catch+ban people who don't read the tourney rules. If they would get banned after a while, they'll start to read them in the future I think...

 

But in long tourneys the chance of a dc is a lot bigger, so there I'd also suggest some adjustment. Perhaps with a % of boards they didn't play. So the sooner they leave, the bigger punishment will be. ;)