BBO hands generator

[skorchev]

If there is such discussion anywhere just give me a link.

 

I'm spending a lot of time playing bridge both live and in BBO for the last several years. I have played handly dealt boards, pre-dealt boards from live tourneys and boards dealt by the BBO generator.

Some experienced friends have told me that this is not the same bridge and the hands are different. I can say the same. Handly dealt boards in average are more balanced. In BBO you feel like they are constructed for action, we play more unbalanced hands here (imho). At the other side those of you who have participated or watched on VuGraph hands from World and European Championships (maybe Aussie, Asian etc. too) maybe has determined that there the boards are stimulating good players, I mean if you play good (theoretically) bridge you will have better results. Also there the boards for MP (pairs) scoring and IMP (teams) scoring are different.

However, the aim of this post is to hear your opinions whether hands from BBO are different, then the handly dealt ones. What do you think?

[gwnn]

I have never experienced anything like you described. Sometimes there are jokers who deal 8-5's and 10-111's, so BBO strikes me as a little bit more balanced than Real Life. Then again, I haven't played that much, so my testimony is at most incidental.

[helene_t]

As for computationally dealt hands: you can be pretty sure they are in accordance with the probability theory. Anyway, to verify any deviation you would have to collect millions of hands and do a formal statistical analysis. Gut feelings based on a few thousands hands will tell you nothing.

 

As for manually dealt hands: some rumors maintain that they are more balanced than they should be. I find this hard to believe (except if you play rubber bridge) but that's just me.

[hrothgar]

If there is such discussion anywhere just give me a link.

 

I'm spending a lot of time playing bridge both live and in BBO for the last several years. I have played handly dealt boards, pre-dealt boards from live tourneys and boards dealt by the BBO generator.

Some experienced friends have told me that this is not the same bridge and the hands are different. I can say the same. Handly dealt boards in average are more balanced. In BBO you feel like they are constructed for action, we play more unbalanced hands here (imho). At the other side those of you who have participated or watched on VuGraph hands from World and European Championships (maybe Aussie, Asian etc. too) maybe has determined that there the boards are stimulating good players, I mean if you play good (theoretically) bridge you will have better results. Also there the boards for MP (pairs) scoring and IMP (teams) scoring are different.

However, the aim of this post is to hear your opinions whether hands from BBO are different, then the handly dealt ones. What do you think?

Some variant of this thread seems to crop up every few monthes. Sometimes charges are leveled against the "Computer Dealt Hands" that are used at major events. On other occasions people start kvetching about particular online sights. either way, the basic complaint is essentially) the same:

 

Hands dealt by computers seem to be more shapely than hands dealt by people.

 

There is a fairly standard response to these sorts of claims

 

1. There are a number of standard statistical tests that can be used to analyze the output from hand generator programs. Most folks who produce hand generator programs use these tests to validate their programs. Fred has specifically stated that he has tested BBO's hand generators and that the hand generators pass these tests.

 

2. It is fairly well established that manual hand generator can produce artificially flat hands. After a board has been played cards tend to be arranged in patterns. You'll find a sequence of four spades played in a row or eight hearts or some such. Combine this trend with lazy individuals who don't shuffle enough and you might very well bias hand generation towards flat patterns.

 

To the extend that there is any bias in the system, it gets introduced by players who are dealing by hand.

 

Unfortunately, there really isn't any solution to the supposed problem.

 

1. Computer hand generation is far to useful to abandon. Online bridge would be impossible without it. The ACBL might like sorting cards by hand, but anyone with half a brain has switched over to using dealing machines for major tournaments.

 

2. In theory, you might be able biased a hand generator program that would attempt to replicate the flaws introduced by manual dealing. I don't think that this would be remotely feasible. First and foremost, this would require collecting massive amounts of data to attempt to determine just how flat the boards should be. Equally significant, the political fights surrounding any such decision would be remarkable in scope...

[P_Marlowe]

Hi,

 

if you deal manually, than you can be pretty sure,

that the shuffling is not sufficient for all hands.

 

There was a mathematical paper, which said,

that you need to riffle shuffle a hand at least

7 times to have a good result, ... additionnal

shuffeling wont improve the randomness

significantly.

 

Now, ask yourself, how often do you riffle shuffel

the cards for a give board until you deal the cards,

... and I dont mean only one board, sometimes you

have to shuffle more than one board.

 

And if the hands are not completly randomized,

than at least for me, it is fairly obvious, that you will

have clusters of colors, which means that the

cards of a color will be evenly distributed,

... cards get dealt one by one.

 

With kind regards

Marlowe

[inquiry]

The answer is yes and no.

 

If you play in ghoulash tournaments, where the director loads intentionally distributional hands (wildly distributional), then yes, hands on BBO are not like 'handy' dealt ones.

 

A second reason they may not be consistent is that ';handy' dealt hands frequently suffer from inadequate shuffling, and so are not truely randomly distributed. In fact, humans don't shuffle well enough, so the handy dealt hands are more "balanced" than theory should require. This is not a problem with computer dealt hands.

 

With these two exception, however, the distributional patterns, the hcp stregnth, etc are consistent with "normal" bridge hands patterns. (one exception, intentional distortion by some directors who load hands, the other inappropriate shuffling by real world players of cards).

 

For more on this, see articles on line, like this one... Disorder in the Deck

[Trumpace]

If there is such discussion anywhere just give me a link.

 

I'm spending a lot of time playing bridge both live and in BBO for the last several years. I have played handly dealt boards, pre-dealt boards from live tourneys and boards dealt by the BBO generator.

Some experienced friends have told me that this is not the same bridge and the hands are different. I can say the same. Handly dealt boards in average are more balanced. In BBO you feel like they are constructed for action, we play more unbalanced hands here (imho). At the other side those of you who have participated or watched on VuGraph hands from World and European Championships (maybe Aussie, Asian etc. too) maybe has determined that there the boards are stimulating good players, I mean if you play good (theoretically) bridge you will have better results. Also there the boards for MP (pairs) scoring and IMP (teams) scoring are different.

However, the aim of this post is to hear your opinions whether hands from BBO are different, then the handly dealt ones. What do you think?

I would expect the BBO dealt hands to be statistically different from the ones dealt by hand. If not, then _that_ is an issue to take up with BBO, and not the other way round.

 

Also, perhaps you and your friends have been afflicted by what is called the Observer Bias, where you tend to exaggerate what you expect to find (in this case unbalanced hands) and ignore what you don't.

 

And as Helene said, your sample size is probably just too small to draw any kind of reasonable conclusions.

 

Perhaps BridgeBrowser can come to the rescue here.

[skorchev]

OK. Thanks to everyone for his/her opinion.

[matmat]

hehe

 

when you deal boards by hand, you gotta make sure to shuffle well :P

[Tola18]

This is a classical dispute between machine predealt cards - like in barometers, and hand shuffled and dealt hand, like in common club-pairs.

 

It is true, the distributions arent exactly same taken statistically as a whole.

 

Why?

a:

it is like others said (feks Marlowe): hand shufflers dont shuffle well enough.

 

If a hand shuffler shuffles 4 times, shehe is unusual thorough. It should be 7....

 

b:

it is a myth a randomly distribution is even. feks throwing a penny.

I had this experiment done several times in school classes:

 

Pairs of pupils throwing a penny, ten pairs, 10 times each if in hurry. Otherwise, 100 is still better. (one throwing, one writing down the result).

Did they fell one this, next the other? NOT AT ALL!

There were PLENTY of series of 5-6, even 7, of same in a row!

 

But the whole class, taken together, did gave a practically 50-50 count.

 

Thus: With correct shuffle / predeal, there should be quite many rather wild distributions. More then someone unprepared would expect.

[jkljkl]

Hello,

 

Ok now we know that hand dealt hands are not what they should be from a statistical point of view. But the original question, does bbo tweak the hands to make them more exciting, hasn't got an answer or did I miss something?

 

To take it to an extreme:

 

I say that every 100th flat hand is suppressed by the software to make the game more exciting.

 

I can't find anything in the answers I have read that would reassure me that the statement above is not true.

 

ciao stefan

germany

[inquiry]

Hello,

 

Ok now we know that hand dealt hands are not what they should be from a statistical point of view. But the original question, does bbo tweak the hands to make them more exciting, hasn't got an answer or did I miss something?

 

To take it to an extreme:

 

I say that every 100th flat hand is suppressed by the software to make the game more exciting.

 

I can't find anything in the answers I have read that would reassure me that the statement above is not true.

 

ciao stefan

germany

The quick answer is no, the BBO does not tweak the hands to make them more interesting.

 

Fred Gitelman wrote the following

The hand generator is certainly intended to be

completely random and the extensive statistical

tests that I have done indicate that it really is!

 

Fred Gitelman

Bridge Base Inc.

www.bridgebase.com

 

To see the whole thread, see BBO hand generator

 

Having said that, i will repeat that bbo allows TD to upload hands if they like (And team games) those can be anything.

[hrothgar]

Hello,

 

Ok now we know that hand dealt hands are not what they should be from a statistical point of view. But the original question, does bbo tweak the hands to make them more exciting, hasn't got an answer or did I miss something?

 

To take it to an extreme:

 

I say that every 100th flat hand is suppressed by the software to make the game more exciting.

 

I can't find anything in the answers I have read that would reassure me that the statement above is not true.

I think that your are (significantly) misinterpreting the earlier comments,

 

I strongly suggest that you go back and carefully read what has already been posted.

 

1. The posters state that they don't believe that BBO's hand generators are biased.

 

2. The posters state that perceptions of bias reflect flaws in manual shuffling methods and not BBO's hand generators.

 

3. The posters are stating directly that BBO does not tweak hands (with the obvious exception of ghoulash tournaments)

 

You are welcome to say whatever you want. However, don't expect anyone to take you particularly seriously when you can't read...

[matmat]

You are welcome to say whatever you want. However, don't expect anyone to take you particularly seriously when you can't read...

but Richard, how do you *really* feel?

[mycroft]

A wise man once said:

There have been lots of freakish sessions over the years of bridge. They tend to be divided into two types:

 

    * Hand dealt ones, where people say "What a freaky set!"

    * Computer dealt ones, where people say "Those computer dealt hands are completely unfair!"

 

There have also been lots of flattish sessions over the years of bridge. They tend to be divided into two types:

 

    * Hand dealt ones, where people say nothing

    * Computer dealt ones, where people say nothing

[jkljkl]

I think that you

I strongly suggest that you go back and carefully read what has already been posted.

 

1.  The posters state that they don't believe that BBO's hand generators are biased.

 

2.  The posters state that perceptions of bias reflect flaws in manual shuffling methods and not BBO's hand generators.

 

3.  The posters are stating directly that BBO does not tweak hands (with the obvious exception of ghoulash tournaments)

 

You are welcome to say whatever you want.  However, don't expect anyone to take you particularly seriously when you can't read...

Since I can't count it does not baffle me that I can't read.

 

So as you say we have the believes of posters.

 

We have the perceptions of the posters.

 

Wow that's hard data :-) to refuse an hypothesis

 

I have no reason to doubt that the deals are random, I do not care if they are not are not random, even: give me pls an option at the table to get wild deals.

 

To not say that it totally suffices to me that Fred says that it is so.

 

I just wanted to point to the fact that it is a matter of faith in the makers of this software, but not more.

[MickyB]

A wise man once said:

There have been lots of freakish sessions over the years of bridge. They tend to be divided into two types:

 

    * Hand dealt ones, where people say "What a freaky set!"

    * Computer dealt ones, where people say "Those computer dealt hands are completely unfair!"

 

There have also been lots of flattish sessions over the years of bridge. They tend to be divided into two types:

 

    * Hand dealt ones, where people say nothing

    * Computer dealt ones, where people say nothing

I once had an opp blame a 3-3 break on computer-dealt hands.

[Cascade]

A wise man once said:

There have been lots of freakish sessions over the years of bridge. They tend to be divided into two types:

 

    * Hand dealt ones, where people say "What a freaky set!"

    * Computer dealt ones, where people say "Those computer dealt hands are completely unfair!"

 

There have also been lots of flattish sessions over the years of bridge. They tend to be divided into two types:

 

    * Hand dealt ones, where people say nothing

    * Computer dealt ones, where people say nothing

I once had an opp blame a 3-3 break on computer-dealt hands.

If they were computer dealt then it was the computer's fault if not he doesn't know what he was talking about.

[mrdct]

I'm finding it hard to believe that any remotely serious bridge is played with hand-dealt boards in this day and age.

 

I'm quite sure that I haven't played a manually dealt hand of bridge since the mid 90s; be that club duplicate, tournament or (obviously) online. I can't imagine a bridge club staying in business in Australia if they didn't provide hand records with makable-contract analysis at the end of each session.

[Gerben42]

A few years ago I have written a computer program that simulates hand-dealt boards. It was pretty obvious from the results that this hand-dealing wasn't effective. I think I did 4 rifle shuffles or so (this seems to be average) and the result was less extreme distributions. If there is interest I can dig up the results.

 

On my homepage you will find a script for HvS's Dealer that will generate hands of a pre-set wildness, check out:

 

http://www.geocities.ws/gerben42/funbridge.html#goulash - Goulash

 

Another point: It is trivially easy to make a dealing program that deals according to the statistics. The only non-trivial part is to generate random numbers to feed the program.

[Guest Jlall]

lol@ this thread.

 

Anyways, I'll give the ACBL haters another thing to hate on: the national swiss teams we play here are shuffle and deal with the exception of the final day, later rounds, for the top few matches which are preduplicated.

[helene_t]

For clumsy people like me who can't riffle, a lot can be gained by dealing in a better way. The traditional

S-W-N-E-

S-W-N-E-

S-W-N-E-

etc

is almost the worst possible way of dealing (arguably SSSSSSSSSSSSS-WWWWWWWWWWWWW-NN.. is worse).

 

I did some simulations similarly to what Gerben describes. In 20000 hands, I got 2148 4333-hands by honest shuffling. Using the traditional dealing method and starting with SHDCSHDCSHDC... sorted cards, I got, for (1,2,3... riffles)

313 738 1162 1462 1731 1760 1900 1937 1988 2070

4333-hand. Using the 5-pile-method, I got

1592 1640 1682 1744 1884 1940 1942 1989 2048 2001

 

If instead I started with cards sorted by suit, I got

10772 6765 4960 4234 3463 3187 2920 2700 2735 2500

and

8088 5112 4021 3506 3088 2839 2742 2517 2478 2458

 

The 5-pile method has a serious flaw, namely that the convergence rate differs between the four hands. There are some other dealing methods that give even better convergence and which don't have that problem.

 

Of course, in real life the statistics are much less skewed than these numbers suggest because one doesn't start sorted cards.

[barmar]

I'm finding it hard to believe that any remotely serious bridge is played with hand-dealt boards in this day and age.

Believe it. In the US, most club games are hand-dealt. And team games at tournaments are hand-dealt, except for the late rounds of national events.

 

How large national team games in Australia? At our regional and national tournaments, there can easily be 100-200 tables in team games. Do they pre-duplicate boards for all of them?

[Cascade]

I played in the Summer Festival of Bridge in January in Canberra, Australia.

 

There were 14 rounds of 20 boards in the Swiss qualifying with around 200 teams playing at two different venues.

 

All boards were predealt.

 

In the past there have been more teams. The Gold Coast Congress next week may well be bigger in terms of number of teams but is a shorter event. They will have predealt boards.

 

It takes me about 10 minutes to deal a set of 24-26 boards with a dealing machine. Maybe the full time dealers for these big events deal even more quickly.

 

At the Summer Festival they share 20 boards between six tables. But it is possible to share twelve boards between six tables - I played an event at the weekend that did exactly that.

 

At these rates there doesnt seem to be a good reason why predealt cards are not always available at any decent event.

[helene_t]

We use hand dealt cards at the club. We do have bridgemates though so that we get detailed, personalized score sheets instantly.

 

In the Netherlands, most pairs tournaments use computer dealt cards, even more social things like the "kroegetochten" (pub races?) where you play four boards in each of six-seven pubs.