[suresh3]

We, me and my wife, have been playing bridge on BBO for a few months now and it has been a blessing during this pandemic. We want to thank very much to BBO organizers for providing this such a great website to so many players. We play intermediate, casual and relax bridge most of the time. We have experienced that the pair who starts the table many times get most points, sometimes 10 to 15 hands in a raw making the other side play boring. We think cards are distributed randomly without any favors to one side. We would like to bring this situation BBO software experts and find out if there is anything can be done. Thanks again.

[johnu]

I've gotten good results by rotating my chair 90 degrees. This makes BBO think I'm somebody else and equity is restored.

[Gerardo]

If you are playing duplicated hands (MP/IMP as opposed to Total Point), please use Myhands, do actual calculations, come back with results. .

[pilowsky]

Hi Suresh - Welcome to the Forum. We are very lucky to have so many sites to play different mind games. BBO is the biggest of the Bridge sites. John is, of course, being silly. Your question has come up before, and it is just a matter of luck.

Gerardo is one of the administrators of BBO (he looks after the site. The site (Myhands) that he has directed you to holds the records of all the hands that you play on BBO. This will allow you to check for yourself what has been happening. You can also use it to find out how other people played the hands that you were dealt.

In fact, the hands are dealt quite randomly with respect to distribution and points. Here is a link to a chart that shows a distribution of what sort of shape and high card points you can expect to get with a deal. It can be quite surprising.
Good luck with your games.

It is possible to create specific hands and import them into BBO for practice, but that is another story. These types of hands can be very unusual. Really strange hands in BBO are called "Goulash hands".

[barmar]

suresh3, on 2020-July-19, 12:01, said:

We, me and my wife, have been playing bridge on BBO for a few months now and it has been a blessing during this pandemic. We want to thank very much to BBO organizers for providing this such a great website to so many players. We play intermediate, casual and relax bridge most of the time. We have experienced that the pair who starts the table many times get most points, sometimes 10 to 15 hands in a raw making the other side play boring. We think cards are distributed randomly without any favors to one side. We would like to bring this situation BBO software experts and find out if there is anything can be done. Thanks again.

I'm a BBO programmer, and I guarantee that there's nothing that biases the hands like this.

In fact, what you're describing isn't even possible. The same hands are played at 16 tables. Each of them may have the player who started the table in a different seat.

There have been accusations that the hands are biased towards NS or EW. While they aren't, at least that would be more possible than what you suggest.

We will admit that our dealing software is not as good as something like BigDeal. But it is categorically not biased. And we've had the author of BigDeal perform a statistical analysis of our hands, and he couldn't find any problem with them.

Someone accuses our dealing software of being biased every few weeks, we get tired of having to refute it every time. Unless you can back it up with data, it's more likely that you just have selective memory of the bad hands.

[StevenG]

barmar, on 2020-July-19, 20:48, said:

We will admit that our dealing software is not as good as something like BigDeal.

What does that mean? Either the deals are perfectly random or they're not.

[pilowsky]

The quality of a random number generator is an extremely important question for codebreakers and amongst computer scientists. It is not something that you will be able to detect much difference in when looking at a few deals.
Mathematicians argue for hours about random numbers and the best way to generate them (https://stackoverflo...pts-math-random)
The link that I provided originally shows that two sites can get very slight differences in their curves depending on where they start. This will only affect a tiny number of deals AT THE MARGIN.
But, this may mean that one population of random numbers contains the unique deal with the AKQJT98765432 spades North. and a particular variation of cards in the other seats.
If you want a discussion of what 'perfection' is, then I think you need to start a topic in the Water Cooler.
Interestingly, nobody ever writes to complain that they had a string of hands with a lot of HCP and that this caused marital discord. Suresh and his wife could simply swap seats randomly to solve their problem if that were the case.

[smerriman]

StevenG, on 2020-July-20, 01:54, said:

What does that mean? Either the deals are perfectly random or they're not.

It's to with predictability. If you look at an individual deal, or set of deals, they will satisfy all statistical tests of randomness (and all conjectures like the one that started this thread are completely rubbish).

However, with a lower quality pseudorandom generator, after you've seen sufficient data, you will be able to predict the next random number that will be generated (and indeed every one from then on).

This is completely irrelevant on BBO as the random number generator is used for so many aspects throughout the software that you'd never be able to gather or use the data you need to make such a prediction. Unlike the old ACBL generator, which was cracked a few years back.

[suresh3]

barmar, on 2020-July-19, 20:48, said:

I'm a BBO programmer, and I guarantee that there's nothing that biases the hands like this.

In fact, what you're describing isn't even possible. The same hands are played at 16 tables. Each of them may have the player who started the table in a different seat.

There have been accusations that the hands are biased towards NS or EW. While they aren't, at least that would be more possible than what you suggest.

We will admit that our dealing software is not as good as something like BigDeal. But it is categorically not biased. And we've had the author of BigDeal perform a statistical analysis of our hands, and he couldn't find any problem with them.

Someone accuses our dealing software of being biased every few weeks, we get tired of having to refute it every time. Unless you can back it up with data, it's more likely that you just have selective memory of the bad hands.

We absolutely have no intention to accuse BBO to be bias in any way, our only intention was to bring our experience to BBO software expert’s attention and suggest to improve the software if possible. But is the fact that on several occasions we have felt bored playing because the other side, who set the table, were getting most points, sometimes 10 to 13 times in a row. But like Pilowsky suggested, maybe it is just a matter of luck or a coincident. Once again, we are thankful to BBO for providing such a great site.

[FelicityR]

Welcome to the forum. There's nothing boring with defending hands. Actually far too much emphasis is made of bidding and declarer play. But I do realise how frustrating it must feel having a series of hands where 'No bid' turns up a bit more often than usual. Always when I am playing rubber bridge against competent opposition, it seems

There's a flip side to this, too. Experienced and seasoned bridge players would not let the (apparent) lack of high card points get in their way. Just because the cards seem to be going the opposition's way, it doesn't mean you should lose concentration yourself.

[hrothgar]

smerriman, on 2020-July-20, 03:40, said:

It's to with predictability. If you look at an individual deal, or set of deals, they will satisfy all statistical tests of randomness (and all conjectures like the one that started this thread are completely rubbish).

However, with a lower quality pseudorandom generator, after you've seen sufficient data, you will be able to predict the next random number that will be generated (and indeed every one from then on).

This is completely irrelevant on BBO as the random number generator is used for so many aspects throughout the software that you'd never be able to gather or use the data you need to make such a prediction. Unlike the old ACBL generator, which was cracked a few years back.

In addition to issues around predictability, some hand generation algorithms are unable to generate all possible hands.

Big Deal is able to generate each and every possible bridge hand (and does so in an unbiased manner)
The generator that BBO uses is unbiased, but can only generate (a large) subset of the available hands.

[mycroft]

And barmar's argument is that if they can only generate enough hands that you'll not see the same one again for 15 lifetimes instead of 20, that's good enough. Not as good as "theoretically perfect", but good enough.

Even if we could generate a book of hands that can't be generated by BBO, I can't imagine it being of anything but theoretical assistance.

Agreeing with hrothgar, as (almost) always where statistics are concerned.

[Vampyr]

It might be interesting, though, to learn what types of hands BBO cannot generate.

[hrothgar]

Vampyr, on 2020-July-20, 13:43, said:

It might be interesting, though, to learn what types of hands BBO cannot generate.

The issue here is not that BBO can't generate a given type of hand

The hand generator is unbiased
This means that it should be able to generate any / all shapes and do so with the correct proportions

The issue here is NOT that the dealer can't generate a given distribution around the table, but rather that some specific deals are impossible

In theory, I suppose that one could get an edge if you knew that you only had to deal with the set of hands that the dealer can generate.
In practice, this is beyond the ability of any human to make practical use of.

[pescetom]

hrothgar, on 2020-July-20, 14:18, said:

The issue here is not that BBO can't generate a given type of hand

The hand generator is unbiased
This means that it should be able to generate any / all shapes and do so with the correct proportions

The issue here is NOT that the dealer can't generate a given distribution around the table, but rather that some specific deals are impossible

Sorry but I don't follow you.
Which specific deals are impossible and why?
How does this differ from BigDeal?

[hrothgar]

pescetom, on 2020-July-20, 14:34, said:

Sorry but I don't follow you.
Which specific deals are impossible, in what sense and why?

Here's a hand that I played yesterday

https://tinyurl.com/y4dncrv2

I know with certainty that BBO's dealer can generate this hand

Hypothetically, lets tweek the hand such that the Two of Spades is now in the North hand and the Three of Spades is in the South hand. It is conceivable that the BBO dealer is UNABLE to generate this specific deal. (I have no idea whether or not this is true, but it might be the case)

Note that most hand generators are UNABLE to generate all possible deals. (In fact, there are a whole bunch of deals that the majority of hand generators can't generate. The precise set of deals that can't be generated are specific to a given implementation)

[hrothgar]

As to why this is the case, it boils down to

1. Design choices

Many of the dealing programs that have been used were designed back in the weird old days when memory and processing power was much more expensive. As such, the folks who implemented this algorithms chose implementations than we efficient and fair but could not generate all possible hands. (This is a relatively common choice in software. if you're interested in it, you might want to look at the distinction between space filling designs and pseudo random number generators)

2. Lack of entropy

Even if your implementation for dealing cards is flawless, if you don't have enough entropy you might be SoL.

[smerriman]

pescetom, on 2020-July-20, 14:34, said:

Sorry but I don't follow you.
Which specific deals are impossible and why?
How does this differ from BigDeal?

You might like to read this.

As mentioned there, most PRNG use a 32 bit seed, and there are more than 2^32 possible bridge hands.