[diana_eva]

3. To score a hand every board is compared between 16 pairs, this result means nothing especially the tops and bottoms have an disproportianate influence on the average. On OKB 52 pairs are compared, and if I am not mistaken the top 2 and bottom 2 do not count for the average. On BBO you have so many more players, can't you compare 100 boards and delete tops and bottoms for a more realistic result?

[diana_eva]

I agree here, a result like 7d xx -6 should simply be removed from the list, otherwise it weights too much in the general result

[babalu1997]

why is it the the maximum number of baords is 16?

would be nice to have hands scored for something like 50 or 30 boards.

Only once it happened to me to play the same deal, i was declarer as south and then joined another table as west.

but i did not realize it till the deal was over.

long time ago, did not report to anyone

[jjbrr]

the results of your scores in the main bridge club are completely irrelevant as of right now, you realize?

[fred]

babalu1997, on Feb 5 2010, 03:13 AM, said:

why is it the the maximum number of baords is 16?

Because 16 lines fit really nicely into the first version of the BBO Windows client othat contained a "Movie window". No scrollbar was required

At that time there were so few people on BBO that it typically took hours or even days for a given hand to be played even 16 times.

Unfortunately I implemented this clever decision with some notably poor design that has left us more or less endplayed in this area to this day.

Look on the bright side - now that there are a lot of people on BBO, it takes only a few minutes for a board to be played 16 times and then retired - sharing information between 2 tables that play the same board is basically impossible

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[mrdct]

I think 16 boards is more than enough results to compile a datum, but a few outliers should definately be excluded; probably top 2 and bottom 2 and then calculate the average against the average of the middle 12 results rounded to the nearest 10.

I don't have a proper mathematical argument for it, but I prefer imping against a fixed datum rather than cross-imping.

[helene_t]

IMO it's fine as it is.

You don't delete extreme results in XIMPs. I suppose OKB used butler. XIMP is more robust so deleting outliers is not necessary.

Personally I prefer not to delete outliers. I can see a case for doing it but since the rest of the World doesn't, I think BBO should just stay to normal XIMP scoring.

Those concerned about the issue might prefer to play matchpoints.

[hotShot]

I think in tourneys there are more than 16 results / board.

And Helene I guess you agree that in statistics averaging over more counts is usually better, although at some points the improvement is minimal.

[helene_t]

hotShot, on Feb 5 2010, 12:07 PM, said:

I think in tourneys there are more than 16 results / board.

And Helene I guess you agree that in statistics averaging over more counts is usually better, although at some points the improvement is minimal.

I think 16 is adequate, but a cycle of 16 means that the average number of tables that have played the board at the time you finish it will be 8. I think that's more of a concern. Sometimes you get a silly score because it has only been played once before and that happened to be one of those 7NTxx-8 or such. That will always happen but the shorter the cycle the more often it happens. Of course, a few minutes later the score will be more normal so it isn't really a big issue.

[barmar]

It's been several years since I used OKB, but I think they use cross-IMPs just like BBO. I also don't recall them throwing out the extremes, but I think it was a frequent suggestion, just like here.

[Gerardo]

As I said in the parent post, IMO Butler is a crutch from pen and paper days, to be able to get a result doing less math.

Now computers do the math, and the crutch is not needed.

There is no need to eliminate top and bottoms, as the only purpose of it is to make Butler introduced datum fairer. No Butler => no datum => no need.

[NOosAnCE]

mrdct, on Feb 5 2010, 03:08 AM, said:

I think 16 boards is more than enough results to compile a datum, but a few outliers should definately be excluded; probably top 2 and bottom 2 and then calculate the average against the average of the middle 12 results rounded to the nearest 10.

I don't have a proper mathematical argument for it, but I prefer imping against a fixed datum rather than cross-imping.

16 boards is by far not enough, and 12 is nothing, only large tourneys will have normal results, using 12 gives random outcomes, if you do not believe me, ask others

[NOosAnCE]

barmar, on Feb 5 2010, 08:35 AM, said:

It's been several years since I used OKB, but I think they use cross-IMPs just like BBO. I also don't recall them throwing out the extremes, but I think it was a frequent suggestion, just like here.

OKB as I said compares 52 pairs and does not count the 2 tops and the 2 bottoms for the average

[Gerardo]

What average? There is no average.

[NOosAnCE]

Gerardo, on Feb 6 2010, 02:22 PM, said:

What average? There is no average.

sigh, you are very stubborn, if you dont want to solve problems, then don't respond please, you know perfectly well what I mean, if I don't use the right lingo does not mean you should insist to discuss averages. The point is 2 pairs score an odd result this affects the rest of the scores and it should not, also to make a score from 16 pairs in the largest field on the planet is absolutely ridiculous, I expect this at the local club, not in a place where as many as 17000 players are online at the same time. On OKB they USE (see, I do not use the word AVERAGE) the scores of 52 pairs to make results for everyone else, and the top 2 and bottom 2 are not counted for the calculation of the results of the other 48 pairs, but they are for the 4 extremes, see I didnt use AVERAGE again.

[helene_t]

OK it seems you were aware that BBO uses XIMP and what the difference is between XIMP and Butler and why extremes are not removed at XIMPs while they are at Butler. But your use of the word "average" suggested that you weren't aware of that, so no need to take offense.

[fred]

Martin,

Sorry if I did not state this clearly, but it doesn't really matter whether or not you are right that increasing the number of comparisons and throwing out extreme scores would be a good thing - you are not going to see this change happen in the foreseeable future.

That is because, due to poor early design by me and due to the fact that the Windows client continues to exist, it would be a complete nightmare for us to make any changes in this area. By this I mean it would take a lot of difficult work to implement the kind of changes you want, these changes would be dangerous (ie screwups by us would result in severe system instability), and everyone who uses the Windows client would be forced download a new version (a nightmare in itself both for us and for many 1000s of our members).

Meanwhile, there are plenty of other areas of the software that have a lot of room for improvement and require our attention (which is a limited resource). The bottom line is that the scoring changes you would like to see, if they ever happens at all, are not going to happen any time soon. I realize such changes are important to you - sorry to be a disappointment.

Of course feel free to continue to argue about the merits of various ways of keeping score, but even if you convince everyone it won't make any difference.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[NOosAnCE]

Ok Fred thats a pity, but a fair answer. I was just irritated by the responses of someone talking about XIMP or butler, this was not the issue.

[Gerardo]

NOosAnCE, on Feb 8 2010, 11:28 AM, said:

Gerardo, on Feb 6 2010, 02:22 PM, said:

What average? There is no average.

sigh, you are very stubborn, if you dont want to solve problems, then don't respond please, you know perfectly well what I mean, if I don't use the right lingo does not mean you should insist to discuss averages. The point is 2 pairs score an odd result this affects the rest of the scores and it should not, also to make a score from 16 pairs in the largest field on the planet is absolutely ridiculous, I expect this at the local club, not in a place where as many as 17000 players are online at the same time. On OKB they USE (see, I do not use the word AVERAGE) the scores of 52 pairs to make results for everyone else, and the top 2 and bottom 2 are not counted for the calculation of the results of the other 48 pairs, but they are for the 4 extremes, see I didnt use AVERAGE again.

So, according to you, OKB uses a way to calculate the score for a set of tables(the extremes), and other way to calculate the score for everybody else?

Are you authoritative on this, or can provide documentation?

[Sadie3]

I've never seen a Regional or National event that did not cut the tables down to a maximum of 18 tables per section, so I do not think that the complaint is a valid complaint. Granted, overall results use all the entries, but a 65% game in one section could be a 58% result in another section depending on how tough the section pairs actually were.

As for a fix... I understand that there are restraints due to original system design, but does that mean that the web version ( aka, the new version) could not design an alternate scoring system option perhaps at a small cost to the users that wish to look at their results differently from the rest of us? It has implemented new convention cards and many other new features, (at no additional cost I might add) and I think that this could be a new option for those that think they would prefer to look at the results in a different light. I think maybe a poll would show just how wanted this wish list item really is.

Since BBO does not have a system rating attached to individual players ( which I totally like), I do not see why this poster is making such a big deal out of it. Playing in the open rooms is a place to work on your game for serious players and a place to kick back and socialize for others, and checking to see what other pairs made compared to what you made is quite enough for me in that venue. I have no problem eliminating the 7NT x and xx hands from my comparisons. I do not care that they skewed the results because I care more about the pairs that made one more trick than I did and how they bid it.

I also encourage many of my face to face player friends to try online bridge and I have no interest in having them look at some complicated rating system and get discouraged before they even have a chance to get used to the software.