Version 1.48f - please post feedback and suggestions here

[diana_eva]

A new version of BBO was released today. Version 1.48f. Partner compatibility score is the main change in this version.

 

For details, see:

http://www.bridgebas...-compatibility/

 

Please post your suggestions/feedback here!

[hrothgar]

A new version of BBO was released today. Version 1.48f. Partner compatibility score is the main change in this version.

 

For details, see:

http://www.bridgebas...-compatibility/

 

Please post your suggestions/feedback here!

 

I'd be interested in understanding what types of factors this takes into account.

 

I just checked and I have a wide variety of different recommendations, ranging from 1/2 star to five stars.

I have no idea how any of this was generated, nor do I understand the distribution of stars.

(I'd expect a normal distribution, but it looks pretty uniform)

[uday]

atm, we're just experimenting, but we factor in ( unordered) profile-country, real location, language, masterpoints, avg adjusted points earned per hand, #people who mark u as a friend vs enemy , award symbol, starriness

 

I eventually plan to include ( for tourneys , anyway ) times played together , # friends in common

 

for the MBC ( main bridge club ) we're going to ultimately need a real/rich profile, i suspect.

[scarletv]

I like the idea but I guess it is difficult to make a good ranking. You would need more information in the profile concerning preferred system, importance of carding and that stuff. I am not so glad with high weighting of nationality, language or being on the friends list. One of the things I like very much on BBO is the variety of nationalities that you can meet here. On my friends list there are a lot of players I rarely play with but kib their play sometimes as I know they are good players. I just checked a couple of players I am playing with regularly and the compability ranking does not fit with what I think it should.

[Wayne_LV]

I think it is yet another attempt to measure an apple with a micrometer. I think the profile is already too cluttered. I give up on Skill Level fiasco and now just show "Private" and refer others to MyHands if they they are interested in actual results rather than my inflated perception of my skill level. The profile is not a convention card either and I just refer others to my Favorite CC or the convention card posted. Just my opinion and I predict it will not be a popular one.

[FM75]

Looks like a great idea. I can imagine that Singular Value Decomposition might work quite well. After all you have some objective measures in the mix as well as known "friends" and enemies. You could also likely use "facebook" style networks with the friends component. I hope it works, and expect that eventually you will have it working quite nicely.

[dlks]

I know the compatibility rating is a very new feature and some people could find it useful, but I actually dislike the entire concept for several reasons.

1) Partnership compatibility is too subjective a concept for any sort of statistical data analysis to be useful. IMO, such ratings fall into the third category of lies: lies, damned lies, and statistics. (Forgive the slight profanity, that's an old quotation.)

2) "Compatibility" is widely recognized as a positive attribute; conversely, a lack of any compatibility rating or a low compatibility rating in a given player's profile could easily be seen as something negative about that individual. We have enough negative opinions about individual players floating around BBO already.

3) The factors listed as being taken into consideration have little or nothing to do with how I form my personal opinions of compatibility with a given partner. I don't care about a potential partner's "profile-country, real location, language, masterpoints, avg adjusted points earned per hand, #people who mark u as a friend vs enemy , award symbol," or "starriness". Some of these things are already available in player's profiles if they are of interest.

4) The most important factors for me in determining partnership compatibility are whether we can agree on the same approach to bidding/carding and whether my partner recognizes that bridge is actually a partnership game wherein we should respect and trust each other and just generally be civil and courteous. It's also nice if our levels of expertise are compatible, but we find out all of these things after just a few hands.

So I'm agin' it. I am FOR creating a richer profile format (system/carding/languages spoken) but I do understand that would be a major undertaking.

[fred]

I know the compatibility rating is a very new feature and some people could find it useful, but I actually dislike the entire concept for several reasons.

1) Partnership compatibility is too subjective a concept for any sort of statistical data analysis to be useful. IMO, such ratings fall into the third category of lies: lies, damned lies, and statistics. (Forgive the slight profanity, that's an old quotation.)

2) "Compatibility" is widely recognized as a positive attribute; conversely, a lack of any compatibility rating or a low compatibility rating in a given player's profile could easily be seen as something negative about that individual. We have enough negative opinions about individual players floating around BBO already.

3) The factors listed as being taken into consideration have little or nothing to do with how I form my personal opinions of compatibility with a given partner. I don't care about a potential partner's "profile-country, real location, language, masterpoints, avg adjusted points earned per hand, #people who mark u as a friend vs enemy , award symbol," or "starriness". Some of these things are already available in player's profiles if they are of interest.

4) The most important factors for me in determining partnership compatibility are whether we can agree on the same approach to bidding/carding and whether my partner recognizes that bridge is actually a partnership game wherein we should respect and trust each other and just generally be civil and courteous. It's also nice if our levels of expertise are compatible, but we find out all of these things after just a few hands.

So I'm agin' it. I am FOR creating a richer profile format (system/carding/languages spoken) but I do understand that would be a major undertaking.

 

Thanks for your feedback.

 

We are already trying to measure some of the things you care about. For example we see the friend/enemy ratio as a reasonable way to guess at how "generally civil and courteous" a person rates to be. Similarly, "levels of expertise are compatible" can be estimated by looking at the average adjusted points earned per hand.

 

Don't be surprised if at some point it is possible for users to specify some (or all) of the information you would like to see in a richer profile. If/when that happens, such information will almost certainly be factored into the compatibility formula (assuming we stick with the concept - as Uday suggested it is really just an experiment right now).

 

Fred Gitelman

Bridge Base Inc.

www.bridgebase.com

[Qiana]

I cannot see this as being a very useful tool. I would much prefer more room for profile. Players can make up their own minds as to who may be a compatible partner and who isn't. That is what the friend's list is for. Different strokes for different folks. Please concentrate on more profile room. thank you..

[Mbodell]

I found it funny that my regular partner I signed on to tonight is only 3.5 stars while nearly all my friends and acquaintances and people who play in our regular weekly league game are nearly all 5 star with me.

[Antrax]

Great idea.

 

One thing to keep in mind is that people might mark other people as friends to be able to track them - for instance expert players whose games you want to kibitz. That might cause incompatible people to have many friends in common.

To work around that, you might want to add a category besides friends and enemies, or to discount common friends whom many other people are friends with.

[Khrystyna]

Garbage in, garbage out. I use the "enemies" list not for enemies per se but for players with different styles and systems with whom I'd rather not partner. I delete the list occasionally and start over. I would hate to think that some excellent player is damaged just because we do not play the same system.

[uday]

I don't know how useful the compatibility will be in the main bridge club. We originally came up with as one of the steps to ease the process by which we find partners in the partnership desk for a specific tournament, and decided to toss it into the profile.

 

In a tournament, you don't have a lot of time, typically. a large % of pairs are formed from the partnership desk in the last few minutes before a tourney fires.

 

In the main bridge club, assuming u have 'perm required to play' enabled at your table (most deals are played at tables that are candidates for 'Help me find a game', which is not too picky at the moment) there is very little info to go on; we're trying to make that a little easier.

 

I'm sure that we have more refining ahead of us on both fronts.

[eleanorsf]

Can you make this feature an option, that could then be turned on or off as the player prefers?

[Fluffy]

I found it funny that my regular partner I signed on to tonight is only 3.5 stars while nearly all my friends and acquaintances and people who play in our regular weekly league game are nearly all 5 star with me.

 

Wow, I got 3.5 stars the most (from star players only that speak english). While most people I get 0 or 0.5 stars, and some people who happen to have a star or are from my same country give me 2.

 

 

I think a good meassure for global compatibility is connections/enemies (marked by me, or me marked by others).

 

I don't think you should meassure at all the country flag, it is somethign that is on plain sight, so I now know that whenever I see someone from my ocuntry he has 2 stars for granted just for being from there.

 

I speak 2 languages, and I think you are taking that into account from my initial tests, that is actually working.

[savage1]

Can we please review the term "enemy"- and find another term that is not so harsh . The players whom I have marked "enemy" are not that at all - just not compatible.

[Oceanss]

My friends/pards do not show nearly as compatible with me as it shows for other people i do not really know (great players, stars, Jec & Co, etc.. - people I only admire watching). Wish they'd call me to play since we are made for each other :P

[Cromlyn]

The concept may have a lot of merit but measuring compatibility on anything other than bridge systems, signalling agreements and expertise is unhelpful.

 

How does this system cope with those of us who enjoy playing different systems with different partners?

 

 

It would be better to give more useful information on the profile, including a wider range of ability. I know some people feel they must upgrade their skill level because otherwise they have difficulty finding someone to play with. So I think this new gimmick will increase their difficulties.

 

If it is just to help tournaments why not just have it show up at the tournament desk?

[patilan40]

Why I can't see my compatibility score?

[42krunner]

This seems like another case of the tail wagging the dog. Why should enemies be included in the star calculation? I thought this tool was to predict.

 

Obviously newer players, like me, will be scared we won't be compatible with most, and people will shy away. I feel like I am dating again. :P

[scarletv]

I wonder that a lot of players show up with 2 stars that do not seem to be compatible at all. Could the reason be they use German as language in BBO?

[hrothgar]

Few quick comments:

 

1. Khrystyna made an extremely important comment:

 

While I am quite sure that many folks use the "friends" and "enemies" tags, I doubt that they are using them to describe player compatibility. I think this would inject a lot of noise into the system

 

2. I very much support they end goal that you are attempting. If I were pursuing this type of project, I'd be following very different approaches

 

(A) Use a quiz like the one dating sites use

 

Set up a system by which a large number of people can bid a corpus of hands

Divide the population into clusters based on bidding preferences

Determine which bidding choices have the best predictive power and turn this into your quiz

 

(B) Follow an approach similar to Google's PageRank

 

Players substitute for web sites

Number of boards weighted by results substitutes for incoming links

 

I suspect that either of these approaches would yield much better results.

[Fluffy]

I wonder that a lot of players show up with 2 stars that do not seem to be compatible at all. Could the reason be they use German as language in BBO?

 

Same happens to me with spannish flags, I have zero stars for most except spannish who have 2.

[uday]

I have two primary goals here

 

 

1. In a tourney context, make it easier for players to find partners in the PD

 

For this, system-type compatibility is important, but, imo, other things are even more important

 

 

2. In the MBC context, make it easier for people to know whether they want to join a table, or whether they want to accept a request to sit from a stranger

 

I think that here, system type compatibility is not so important unless the person is playing with the table host

 

 

I think we can't avoid a full blown quiz at some point, but we have to cater to people who won't bother.

 

Consider someone who is a total jerk. For whatever reason, for every 10 users who mark him as a 'Friend', there are 100 users who mark him as an Enemy.

I don't *know* this but I think i can safely assume that most of the F/E marks are from people who like/dislike this person. Sure, there will be some noise from people who use F/E for other purposes.

 

 

I think we have to be careful about who we claim is compatible with this user. It isn't all about bridge things, i think. This "niceness" thing is pretty important.

 

is Jerk1 compatible with Jerk2 ?

Is Sweetheart1 compatible with Sweetheart2 ?

 

 

All else being equal, I know that *I* would much much much rather play with someone who was not at my skill level but who was polite, than someone who was my clone except that no one liked him.

 

 

The 'quiz' will definitely help a great deal. before we jump into that, i want to learn if this approach has any merit at all, before working on fixing it.

 

 

What I'd love to discuss, perhaps in a different thread, is 'what goes into determining whether A is compatible with B' ? Who should decide? Can this be made simple enough to be usable?

[uday]

Note: we use country of IP address, not country in profile