[barmar]

timouthy, on 2014-May-22, 12:10, said:

The same way you assembled all of the current info. to make up your "compatibility" rating. The only thing I want you to determine is a player's post-hand predilection to discuss hand results. Your terse tone/answer suggests your lack of imagination how to do this, so heres a blueprint: Ask the question and have players pick from these possible responses: yes! very much, No not really, sometimes, or other with explanation. Thats not so hard now is it?

The way we assembled the current info is with information we already have in profiles, friend/enemy lists, and playing results. We didn't create any new profile fields or require people to respond to a survey. We'll want to get some experience tweaking the current formula before we consider adding new profile fields specifically designed for it. And if we do go that direction, I think more people are interested in something related to system compatibility than post-mortem preferences.

[timouthy]

barmar, on 2014-May-23, 10:11, said:

The skill levels have been discussed ad nauseum for years. Everyone knows that lots of people do not set their skill levels accurately. The criteria are very subjective.

There you go again. As if previous discussion and "concensus" by self-selected contributors to this site is your proof. I maintain that the vast majority of players on bbo make a good faith estimate of their skill level. Yes I know that means some don't. So you and other bbo administrators have decided your skill rating policy based on the actions of a minority of players? Utterly ridiculous! If you are so convinced that skill level is irrelevant, why do you even have it at all? I'll tell you why, because you and every other bbo participant, knows in your bones that your site would be chaos. You seem so tone deaf and entrenched against this idea, that you have lost perspective about what is good and desirable about playing bridge on bbo. Normally I don't waste anywhere near this much time in a discussion with someone. I was naively under the impression that a "forum" would have a minimum level of sophistication of discourse to it, but I am done trying.

adv exp

This post has been edited by diana_eva: 2014-May-24, 06:14
Reason for edit: fixed quotes

[diana_eva]

The problem is not that BBO refuses to discuss this issue, but that is has been discussed over and over, and you can run a search for "skill level" to see for yourself just how many times there were lengthy, meaningful discussions about it. Here's the most recent one:

http://www.bridgebas...21-skill-level/

The conclusion has been so far that lacking a better solution it's gonna remain as is.

Your suggestion was to add more levels to the existing system. It's a reasonable suggestion, and for what is worth it has been suggested before.

[Antrax]

timouthy, if you have something to add to the skill level discussion, just bump the largest thread you can find with the new data or suggestion. So far you haven't said anything that wasn't said many times before.

[1eyedjack]

barmar, on 2014-May-16, 08:55, said:

This is mainly oriented towards helping you select among people you don't already know. If you have an established partnership, you don't need us to tell you whether you're compatible.

The benefit of looking at compatibility ratings of players already familiar to you is that it enables you to assess how much confidence you have in the algorithm being used to assess your compatibility with unfamiliar potential partners. If the system tells me that I am consistently compatible with those that I know otherwise (and vice versa) I know not to place any confidence in the scores provided for unknown potential partners. So far, that exercise leaves me with little confidence of its accuracy. For as long as the algorithm remains opaque and without any user influence I don't hold out a lot of hope for it. Still, it is early days I suppose.

[hotShot]

If you really want to help me to decide which partner to pick, then you could provide me with 2 numbers and a filter:

How many boards did we play together and what was our average score at this form of scoring.
Filter out those without a tcr better than a number I set globally.

Together with the friend/enemy colors this would be good enough for me.

Since I play a lot of Indies there are a lot of players I have at least played 1 board with.

[hotShot]

Minor language bug:

The tooltip for indies in german says it a pair tourney. (Noticed at the Express Automated ...)

[barmar]

hotShot, on 2014-May-24, 04:05, said:

If you really want to help me to decide which partner to pick, then you could provide me with 2 numbers and a filter:

How many boards did we play together and what was our average score at this form of scoring.
Filter out those without a tcr better than a number I set globally.

Together with the friend/enemy colors this would be good enough for me.

Since I play a lot of Indies there are a lot of players I have at least played 1 board with.

The primary purpose of this is for estimating compatibility of people you don't know and haven't played with before. If you've played with someone, you can use your own judgement to decide whether you want to play with them again. It's for people going to the Partnership Desk because they don't have a partner they know.

Folks have complained that we use geographic location. But the idea there is that you'll be able to discuss conventions because you speak the same language. You're also likely to be starting from the same base -- if you're in the US or Canada, you probably know SA and/or 2/1; if you're in England, you can probably use Acol as a starting point.

[scarletv]

I played with a number of proposed players but as long as there are no more hits than one out of ten I can just throw a coin. I wonder if others tried and had more luck.

[diana_eva]

scarletv, on 2014-May-25, 13:33, said:

I played with a number of proposed players but as long as there are no more hits than one out of ten I can just throw a coin. I wonder if others tried and had more luck.

I did try this, I was curious how far off the formula was. It went reasonably well, scored over 60% with both random but compatible partners. What I noticed in both cases was a level of a priori trust that helped a lot in avoiding misunderstandings. Most of the times I play with an unknown partner I am hesitant to make the most appropriate bid on convoluted sequences for fear they won't understand it same way as I mean it. This time I just went ahead and bid like I thought was right and pd did get it right (probably trusting me the same way I trusted them). FWIW I am aware that compatibility score has nothing to do with system or bidding. But it helped to built trust, and that made things a lot easier.

[u4eni4ka]

OK-but how can I feel this kind of pre-trust, when one of the two hiest rated players (3 1/2 or 4 stars) is marked in black and I've already added acomment like "<45% average for >1000 boards!!!!!" or "absolute idiot" ... Yes, sometimes I add that type of comments for my enemies... And yes, I thied more than 20 times playing with the predicted compatible player, and it's usually worse then playing with a GIB.

[nickdk]

This is the first time I've posted to the forums.
I wanted to thank you guys for the compatibility ratings. Really changed how I'm using BBO.
I used to mostly play robot tournaments because it was hard to assess partners.
I now play mostly with human partners using the compatibility ratings. they don't always work well. that's true.
But since starting to use them, I've actually found 2 or 3 partners with whom I play quite well.
It's the most useful thing I've seen for trying to find tournament partners.

I have a lot of friends and enemies. I use those fields for many things. but this is actually helping me gauge how well we might play together. Thanks!

[lycier]

Beginner's aspirations : Fans Rank

I don't think the Partner compatibility score is useful.
I have found many strange players with same flag have five star in my eyes,but I never had played with/ against them,so I know even that five star is unuseful to me in fact.
We are beginners -- most players at BBO,love this game online,we prefer win to failure,this is a our fun style,and we don't envy professional players !
If BBO think the Compatibility is a good try,I suggest the Fans Rank is more useful to most of people at BBO.
We know it is impossible for us to like Fred,being bridge championship in the future!
For BBO,no beginner,no BBO ! Just a game,a fun life style !
Here I should say fairly that many of real experts rarely play at BBO,even they often log into BBO,their expert skill or world class skill are useful to any?
Who is the most important contributor ? expert? beginner?
Who know our beginner's aspirations?
Though we are beginners,we honestly play ths game ,year in year out,so please BBO respect our beginners :
BBO should encourage us,and grant us some honour symbols ,this is Fans Rank !
I mean that BBO can grant every BBOers a honor star ---- as long as we played hands over 10,000 boards,and we log into BBO over 1,000 times!
And BBO grant 100 Gold medal members with a Gold medal symbol in the profile every year.
This is just a preliminary idea,however better than compatibility score.

[arnprince]

I agree with what others have said regarding compatibility qualifiers - I only care about how we will work as a partnership. I have actually found the system has been fairly good for me so far, but HATE the idea that social aspects of a person's experience on the site has anything to do with compatibility as bridge partners. I can mark my own friends and enemies and get recommendations from them - or waste time on Facebook if I want to know who knows whom! The system records all hands a member plays, yes? Does it also log in the bidding? Surely an algorithm can be developed that can measure compatibility based on the way members bid and play using tracked data - and I would suggest it be based on the previous six months of play. In this way, as players improve, their compatibility would be adjusted appropriately. In fact, the right algorithm would not only be able to match players based on their established record - it would also be able to assign them the proper ability level as well - no more lying. If dating sites can use self-entered, often fabricated data as the basis for matching people, and they are still going strong - surely someone can write code for a program that will match people based on established and factual records. And please do something about those who cannot be bothered to enter anything on the profile to inform prospective partners about the way they approach the game. Thank you.

[broglet]

What's the point in issuing compatibility star ratings which are unexplained and apparently of no practical or other use?

[FM75]

broglet, on 2014-June-02, 00:58, said:

What's the point in issuing compatibility star ratings which are unexplained and apparently of no practical or other use?

It has been explained several times, in this thread. No, BBO did not publish their "algorithm" which they have described as "under development".

That said, what is the point in describing what medicine you should take based upon your individual genetic makeup, and the likelihood of whether it will work and for how long? You would not understand the science behind it, even if you were given the volumes of details about it, or even the abstract describing the technique.

Would that make it of "apparently no practical use"? Well perhaps to a sceptic like you, it would not be of any practical use. You could criticize it as much as you like. But a doctor treating you, with knowledge of the methodology - absent the details - and the understanding of how well it worked, would be crazy not to take it into consideration. He might even be found culpable of malpractice if he ignored it.

Note: I am not suggesting that BBO's algorithm has the same level of predictive ability, but analyzing the predictive value of their model should/could be their next step.

[1eyedjack]

FM75, on 2014-June-02, 20:09, said:

It has been explained several times, in this thread. No, BBO did not publish their "algorithm" which they have described as "under development".

That said, what is the point in describing what medicine you should take based upon your individual genetic makeup, and the likelihood of whether it will work and for how long? You would not understand the science behind it, even if you were given the volumes of details about it, or even the abstract describing the technique.

Would that make it of "apparently no practical use"? Well perhaps to a sceptic like you, it would not be of any practical use. You could criticize it as much as you like. But a doctor treating you, with knowledge of the methodology - absent the details - and the understanding of how well it worked, would be crazy not to take it into consideration. He might even be found culpable of malpractice if he ignored it.

Note: I am not suggesting that BBO's algorithm has the same level of predictive ability, but analyzing the predictive value of their model should/could be their next step.

This sounded rather arrogant to me. I can't get too excited about it because I think that the whole exercise is a blind alley.

In other areas of IT, I place a fair amount of trust in open source programs. Personally I have no hope of understanding the code, but I take comfort that there are others who do. On the other hand I sympathise with companies' desires to protect their intellectual property. That said, publication of some of the principles, short of releasing the code, could add to public confidence without compromising security of IP. Whether we lowly minions would be capable of understanding it is a question that can never be resolved for as long as it withheld, but the assumption, based on currently provided info, that we cannot understand it is unwarranted. Unless FM75 knows something that the rest of us do not.

I also can't get excited about it because it is an evolving feature. Criticism today of a feature that is likely to be different tomorrow anyway sounds like a bit of a waste of time.

I suppose that it had to be unleashed on the public in an unfinished state. How else can it be properly tested? But we are currently in a position where (in my perception) there is a fairly low level of public trust in its recommendation, driven by large disparities between computed compatibility ratings and known compatibility ratings of familiar members. Heck, I am only 3.5 stars compatible with myself.

Furthermore, just as it is arrogance to assume that we users are incapable of understanding the basic principles applied, so I think it is arrogance on the part of BBO to assume that the same factors, with the same weighting, are appropriate across the entire population of members in determining the result. There are a lot of subjective factors in compatibility. One member may regard a common country of origin to have a higher relevance than another. Earlier in the thread I suggested that this will only gain acceptance if users have some control over the weighting applied to the factors, such as by a slider bar. Maybe that will come in due course. I hope not, because I would rather that they devoted their resources in other areas. But for as long as this rating is here to stay I can't see that issue disappearing until that is addressed.

[Antrax]

1eyedjack, on 2014-June-03, 00:16, said:

this will only gain acceptance if users have some control over the weighting applied to the factors

I disagree. If it works, in the sense you find you consistently have good games when playing with randoms with a high comparability rating, then people will use it without regard for how it works or the desire for control.

[diana_eva]

FM75, on 2014-June-02, 20:09, said:

It has been explained several times, in this thread. No, BBO did not publish their "algorithm" which they have described as "under development".

That said, what is the point in describing what medicine you should take based upon your individual genetic makeup, and the likelihood of whether it will work and for how long? You would not understand the science behind it, even if you were given the volumes of details about it, or even the abstract describing the technique.

Would that make it of "apparently no practical use"? Well perhaps to a sceptic like you, it would not be of any practical use. You could criticize it as much as you like. But a doctor treating you, with knowledge of the methodology - absent the details - and the understanding of how well it worked, would be crazy not to take it into consideration. He might even be found culpable of malpractice if he ignored it.

Note: I am not suggesting that BBO's algorithm has the same level of predictive ability, but analyzing the predictive value of their model should/could be their next step.

This was pretty harsh. I think this poster was merely signalling that there is no documentation on BBO itself explaining what compatibility stars are and how they are useful. I did not read his post as a request of disclosing the complete algorithm.

[Antrax]

Really? I sort of agree with him. Chess grandmasters don't consciously know the weights to assign to different piece configurations, which is why nobody hires them to tell them which weights to use, and instead learning algorithms are used. This seems like a very similar scenario. People don't really know whether the country is 0.7 or 0.4 important to them, so suggesting to give them control just seems like Luddism more than anything.