[sallyally]

I am wondering about all this completion rate thing. When I first noticed it my rates were something like 98% and 99%. Now I see I am down to 83% and 91%. I don't understand this since I don't leave boards unplayed nor do I leave tournaments.

I have had a troublesome router for some time and I would go in and out of tournaments usually immediately back in. Occasionally I would not be allowed back in as the tournament didn't allow original players to return. One recent tournament I played in I played 11 of 12 boards and got kicked out and could not get back in. Does that show I didn't complete a tournament? Fortunately I have a new router and hopefully that problem will be solved. I expect this is much ado about nothing but I find it all annoying - as if I had failed at something.

[Bbradley62]

sallyally, on 2011-March-24, 09:59, said:

One recent tournament I played in I played 11 of 12 boards and got kicked out and could not get back in. Does that show I didn't complete a tournament? Fortunately I have a new router and hopefully that problem will be solved. I expect this is much ado about nothing but I find it all annoying - as if I had failed at something.

Yes, the system will accurately show that you did fail to complete the tournament.

[barmar]

The system doesn't distinguish between players who bail intentionally and those who get disconnected due to network problems outside their control. And there's a reasonable argument that it shouldn't. It doesn't matter why you disappear, it's just as disruptive to the game.

This may discriminate against players coming from locations with poor Internet connectivity. TD's who want to be friendly to them don't have to use the completion rate filter.

[onoway]

For hand completion rates; does this apply when playing with 3 GIBS? Sometimes I will bail from a hand that the GIBS and I have apparently been in different universes. But if the Robot tourneys aren't counted, are the robot table hands? Is this ONLY in Main bridge club or anywhere that the hands are recorded?

Also, when leaving a table..the new deals come in very quickly after the last card played so should we all make sure to get out instantly or how does that work? I think the idea is great, esp for tourneys! But if I am going to get zinged if I leave a robot table I may reconsider using them as much as I do.

[j_with_a_B]

I have two simple quick questions...

Firstly, as for how a player's completion rate is calculated: Does the system treat a player that leaves in the middle of a hand (whether from a true connection failure or who logs off by intent to avoid a bad score on a hand) get treated the same as a player that freezes and quits playing or responding at all and so gets replaced with a sub by a TD?


Secondly, as far as an active TD in free tournaments: If I do not use the minimum completion rate, do the runners from my tourneys still have their behaviour in that regard recorded and used by those that do?

I am assuming that yes is the answer to both, but I often get asked both, (especially the first) by players and other TD's. I'd feel better knowing for sure and not having to qualify my answer with "I think...".


I think the whole idea is a very positive step even though I have only used it a few times. Since I host a few tourneys most every day, and also usually at the same times, I have gotten to know the regular daily players as well as the occasional casual ones and the ones whose names are completely unfamiliar to me (whether from being truly new to the site or from a using a freshly minted identity, for whatever reason, but usually from being banned under the last one).

The main reason I have not used it regularly is that some players in my tourneys are ones I have known to play in them for years, and they have always had connection problems. I know I could add "Include friend's list of the host" to other conditions, but the first few times I tried the MCR I had to keep raising it to a point where they quit asking why they were excluded and were able to play but saw that I was allowing many other notorious runners who were then allowed to enter.

I keep good notes on players that have qualified for my enemies list, but that includes players that I merely what to watch more closely due to either an allegation of cheating or rudeness, although not confirmed to my satisfaction, to ones that have been so rude or obscene that I will remove them on sight without hesitation and I try to mention to their (usually)unsuspecting partners that asking or accepting an invite from that player is pointless since I will not allow them to play again under any condition.

Most of the players on my enemies list that I have no notes for are, by default, runners that left an obviously tragic hand or players that have obviously made a spite bid to punish their partners (eg: bid 7NT as an opening bid with 2-3 points, redoubled it and left to punish their partner for a poor bid or misplay in a previous board). Those latter ones I will make a manual list of to add to my enemies list and edit the online BBO notes to reflect that, even if I need to hunt them down in the browser version to do so...

My way of not using inclusion or exclusion lists is simply my preferred way of doing it. I may cause myself a little more work and may need to rely more on notes or my memory to handle things, but that is my choice. As I mentioned, I am quite familiar with the wonderful regular players, the recent newcomers, the occasional and casual players, the "problem children" and the completely unfamiliar ones, after doing each of my regular tourneys hundreds of times.

Whatever the method used by any of the individual TD's, the important thing is that we have effective ways for keeping the truly undesirable players away from the vast majority of the rest. BBO's efforts to make that so is one of the things that has earned my highest respect for itself as a site that is well run and so open to suggestions for improvements.

[blackshoe]

Perhaps there should be three lists: friends, enemies, and suspects.

[j_with_a_B]

blackshoe, on 2011-June-19, 19:52, said:

Perhaps there should be three lists: friends, enemies, and suspects.

I so completely agree! The bulk of the players in my enemies list are merely "suspects". Fortunately the others that are marked ROS (remove on sight) are very few and whatever they have done to earn that distinction is usually vividly memorable enough for me to rarely have to confirm that status.

[0 carbon]

I'd suggest that a list of restrictions should be processed in order, inclusions, then "inclusion only if", then exclusions.

If a player is included then they play.
If they are excluded, then they are not allowed
If they satisfy an "Only", then they are processed further

This would imply changing the list of restrictions to read:

• Include friends of Host
  
• Include custom list(s)
  
• Only language of Host
  
• Only country of Host
  
• Only player level X+ or X- (ie, all >= or <= a particular level)
  
• Only BBO Masterpoints X+ or X-
  
• Only ACBL Masterpoints X+ or X-
  
• No enemies of Host(s)
  
• Minimum logins N
  
• None on these list(s): ...
  
• Minimum Completion rate %

tOM

[hallway]

I need to get to understand this correctly in order to ensure that every single one of my members who want to play in the Audrey Grant tourney next week before attending the review in the BIL can do so.

How many tourneys they have played in is immaterial ( I am hoping that many will do so for the first time )

How many tourneys they have been disconnected from for whatever reason is immaterial

How long ago they played the last tourney is immaterial ( in fact they are ones I am trying to encourage to give tourneys another go)

So will my putting 0% ensure that every single one who goes to register will get registered ?

Is there some other codes I need to add to the description to ensure that they can register , that they can talk to the tables they are at etc. etc.

I will of course be using the "Include Members of my Private Club" restriction

[FM75]

The "aging" algorithm seems peculiar to me.
Maybe that is just because you wanted really simple SQL for
the selection process and it was easy to select all greater
than today - 30.

It would make much more sense (IMO) to take the last N tournaments
by a participant. Use select ... order by date desc limit N. By using the
"last N", you also avoid biasing against someone who has not yet reached N,
or does not have N recently.



Firstly, it is not biased against infrequent, but reliable, players.
Secondly it does not suffer from the defect of eliminating a
reliable sample size.

Example 1:
Player played in 1 tournament per day in Feb. On the last day,
there was a communication failure. The other 28 days were
completed tournaments. The player goes on vacation for a few
weeks. Then the TCR declines monotonically daily to zero.

Example 2: This player also played once each day. He left
tournaments intentionally in each of the first two weeks. At
the end of the month he has a 51.7% TCR. He takes a vacation.
While on vacation, his TCR improves monotonically to 100%!

I suspect that neither of those are desirable properties of
the algorithm

[diana_eva]

hallway, on 2013-February-14, 14:19, said:

I need to get to understand this correctly in order to ensure that every single one of my members who want to play in the Audrey Grant tourney next week before attending the review in the BIL can do so.

How many tourneys they have played in is immaterial ( I am hoping that many will do so for the first time )

How many tourneys they have been disconnected from for whatever reason is immaterial

How long ago they played the last tourney is immaterial ( in fact they are ones I am trying to encourage to give tourneys another go)

So will my putting 0% ensure that every single one who goes to register will get registered ?

Is there some other codes I need to add to the description to ensure that they can register , that they can talk to the tables they are at etc. etc.

I will of course be using the "Include Members of my Private Club" restriction

You should not use the completion rate filter at all if you want to make sure all non-regular tourney players can join.

[barmar]

FM75, on 2013-February-14, 14:46, said:

It would make much more sense (IMO) to take the last N tournaments
by a participant. Use select ... order by date desc limit N. By using the
"last N", you also avoid biasing against someone who has not yet reached N,
or does not have N recently.

The LIMIT clause limits the number of results of the entire query, not per player. MySQL doesn't have a built-in way to do this per-player (other databases have something called "window functions" for it). I've found some web sites with kludgey solutions, but I suspect the performance may be too slow for the volume of data we're dealing with (more than 80,000 players and records going back to Oct 2011).