Moneybridge vs. GIB

[IdiotVig]

It's not about the math...

 

Quasi-seriously, we can run an experiment. We'll get some kind of RNG to simulate coin flips. After some arbitrary number of consecutive flips that come up the same (tails = 6, heads = 10, whatever), I'll take 55-45 that you can't name the next one. Whatever stakes you like, minimum 100 trials.

[Free]

It's not about the math...

 

Quasi-seriously, we can run an experiment. We'll get some kind of RNG to simulate coin flips. After some arbitrary number of consecutive flips that come up the same (tails = 6, heads = 10, whatever), I'll take 55-45 that you can't name the next one. Whatever stakes you like, minimum 100 trials.

Go to http://www.random.org/integers/?num=100&mi...at=html&rnd=new

 

Just too bad there aren't any quick statistics...

 

1 = I win, 2 = you lose ;)

[gnasher]

On the topic of hcps and seats, it probably depends a lot on how BBO is programmed? Say it is designed such that throughout a sessions N, S, E and W all have an average of 10 hcps. Then changing seats randomly should also increase the variance on your average hcp range in the session. If the hands are randomly dealt however it makes no difference whatsoever where you are sitting.

I think all of this discussion has assumed that BBO deals each hand randomly. Obviously any systemic bias in the dealing process might cause the sort of effect that Free is suggesting.

 

If BBO did do that, though, what if offered wouldn't be "Bridge", since it would breach Law 6E4, which requires that on each deal the result of dealing (other than hand-dealing) be "wholly random". Also, it would mean that Fred (or one of his employees) is an idiot.

[P_Marlowe]

Simply put: If you are on the loosing side, the probability that

you will stay on this side is higher and growing over time.

The probability that the score will be +-0 will go too zero, if time goes on, if I remember it correct even with exponenatial speed.

Say person A lost 6 coin flips in a row against a neutral person. Person B lost 10 coin flips in a row against another neutral person. You're telling me that person B has more chance of losing again than person A? ;)

<snip>

No, but the chance that B will be in the minus after the series stopps

is higher, compared with the chance that A will end up in the minus.

 

In the end I was talking about "random walk", a series of random steps

and the question where do you will end up after the random walk ends.

 

So to translate this to the topic:

If you start counting, which side has more HCP in a specific tournemanent,

and you discover at one point in time, that one side got more than av.,

the probability that this will be as well at the end of the tournament is higher

than the probabilitiy that you will catch up and have the same amount of

hands with more HCP your way.

Of course at the start of the tournament it is 50/50 on which side you will

end up.

With kind regards

Marlowe

[helene_t]

Dunno if this is what the thread is about, but:

- I suspect at total points non-duplicate there is a small advantage of being vulnerable as the higher game and slam bonuses are not offset by the higher down trick penalties. Also I would expect it to be an advantage to be dealer (or one's p to be dealer!)

- Therefore it is "unfair" if one player is vulnerable/dealer more often than the other.

 

In the main room where boards are filtered by not having been played before by any of the four players, I can understand that I can be vulnerable more (or less) than 50% of boards, by accident. Maybe also in robot race where if the seats rather than the hands are cycled to give the human the most HCPs (is that actually what happens?). I would expect money bridge to use the usual regular pattern of vulnerability and dealer, though. It surprises me that (or if?) this is not the case.

[hotShot]

...

In the main room where boards are filtered by not having been played before by any of the four players,

....

They are also filtered by vul.

Or to be precise, there are 16 pools of boards for any vul. situation and the next board for your table is picked from the pool that has the required vul.

[awm]

Which direction I sit and how many points I get are independent events.

 

In other words:

 

Pr[i have x high card points] = Pr[i have x high card points given I am East]

 

(for any x)

 

So which direction I sit will not effect either the expectation or the variance of the number of points I receive. Even in a short event, switching seats does not make a difference. It doesn't make things any more random as far as points are concerned.

 

Of course, my results on the boards are not solely a function of the cards dealt. It also clearly depends on vulnerability (I get a different score for making or failing in the same contract) and on position (the auction will go differently depending on who is dealer).

[uday]

I would expect money bridge to use the usual regular pattern of vulnerability and deale

 

I can't --- people might find it advantageous to bail after a desirable vul/positional hand, before the other guy had his.

 

In MB, the board # is randomized each deal, also sits S and who sits W

[Free]

I would expect money bridge to use the usual regular pattern of vulnerability and deale

 

I can't --- people might find it advantageous to bail after a desirable vul/positional hand, before the other guy had his.

 

In MB, the board # is randomized each deal, also sits S and who sits W

That's an interesting and valid point you have there.

 

Questions for Uday:

- would it be possible to obligate people to play per 4 boards instead of 1?

- (out of interest) how are the boards randomized? With a true RNG or with a pseudo RNG?

 

And a general question: would it be a big loss if the vulnerability in MB would always be the same?

[Lobowolf]

And a general question: would it be a big loss if the vulnerability in MB would always be the same?

Yes.

[jdonn]

Just brainstorming out loud. What if some sort of collatoral amount relative to the stakes being played were forced to be put down if you played a board with favorable vul before one with unfavorable vul, and then when you play an unfavorable board it is returned to you?

 

Alternatively (and perhaps simpler), what if, should you play a w/r board before a r/w board and get a positive score, you wouldn't be paid for it unless you stay through the next r/w board?

[uday]

would it be possible to obligate people to play per 4 boards instead of 1?

 

I can't force them. What if they quit after , say, 1 ?

[Free]

would it be possible to obligate people to play per 4 boards instead of 1?

 

I can't force them. What if they quit after , say, 1 ?

Loss of 100 points per quit game or something similar, the current game is finished by GIB like you do now when someone leaves in the middle of a hand. You might make this an option that the creator of the table can select, so the current formula is still available.

 

I like Josh' idea as well, but it's probably difficult to implement. With random vulnerability it's possible that you've won thousands of points but no money. It would definitely reduce people leaving after 1 board, but may need some practical adjustments to make people stay longer than 3 or 4 games.

 

Biggest problem with all this would probably be to communicate this properly to the players. Another issue is that you'll need more funds to join a table.

 

I also had another idea (brainstorming): why not implement some kind of Chicago scoring for MB? I don't have any experience with how accurate Chicago scoring is, but this could actually reduce the luck factor (which is imo important for 1 on 1 money games). Switching seats or vulnerability wouldn't be a big issue, even getting no points from the dealer won't be a disaster. Implementing this is, at first sight, only introducing a new scoring method.

[jdonn]

Having come back later, I don't like my ideas that much, randomizing must be the way to go.

[H_KARLUK]

"Claim" button would be nice! Is it really hard to add?

Good day gentlemen.

[barmar]

"Claim" button would be nice! Is it really hard to add?

Good day gentlemen.

How would you explain your line of play to robots?

[jdonn]

"Claim" button would be nice! Is it really hard to add?

Good day gentlemen.

How would you explain your line of play to robots?

Bridge computer games have a claim button, I had an old one called micro bridge companion that (correctly) allowed or rejected claims that were far from obvious.

 

Perhaps the logical first step is a button that lets you claim all the tricks if there is no way they can be lost on any order of play.

[Free]

Perhaps the logical first step is a button that lets you claim all the tricks if there is no way they can be lost on any order of play.

That would be pretty useless since it's too rare. For example QJx vs AKxxxx cannot be picked up according to your rule unless the suit splits 2-2. If there's still a trump out, you can't claim either. Etc.

[jdonn]

Perhaps the logical first step is a button that lets you claim all the tricks if there is no way they can be lost on any order of play.

That would be pretty useless since it's too rare. For example QJx vs AKxxxx cannot be picked up according to your rule unless the suit splits 2-2. If there's still a trump out, you can't claim either. Etc.

So play 3 rounds and claim the last three. You don't think that is better than nothing?

[Mbodell]

Perhaps a useful button might be let GIB see all my cards. That way when it is clear to you that you are claiming GIB will not take a lot of time analyzing the various possible lines since he'll know the layout and know what to do.

 

Not quite as useful as a claim, but this way GIB will play very fast for you and there is no real problem of someone making false claims, as they'll just get perfect defense as a result.

[TimG]

Could there be some sort of claim coup with GIB? Suppose you have AKJxxx a opposite xxx in trumps and claim. GIB will only reject the claim if trumps don't split 22 (or there is a singleton Queen) so that playing them from the top is not good enough. Once you know that line is not working, you can take the finesse.

 

This sort of thing won't work against humans (usually), but GIB won't recognize the nefarious nature of such claims.

[Mbodell]

Could there be some sort of claim coup with GIB? Suppose you have AKJxxx a opposite xxx in trumps and claim. GIB will only reject the claim if trumps don't split 22 (or there is a singleton Queen) so that playing them from the top is not good enough. Once you know that line is not working, you can take the finesse.

 

This sort of thing won't work against humans (usually), but GIB won't recognize the nefarious nature of such claims.

Yes, that can be a problem with a claim button that results in "yes you win the rest of the tricks" vs. "no you don't definitely so play on". But my earlier suggestion of letting GIB see your cards would be safe from this sort of problem. Because GIB doesn't give you feed back, it just plays super fast.

[barmar]

Could there be some sort of claim coup with GIB?  Suppose you have AKJxxx a opposite xxx in trumps and claim.  GIB will only reject the claim if trumps don't split 22 (or there is a singleton Queen) so that playing them from the top is not good enough.  Once you know that line is not working, you can take the finesse.

 

This sort of thing won't work against humans (usually), but GIB won't recognize the nefarious nature of such claims.

Yes, that can be a problem with a claim button that results in "yes you win the rest of the tricks" vs. "no you don't definitely so play on". But my earlier suggestion of letting GIB see your cards would be safe from this sort of problem. Because GIB doesn't give you feed back, it just plays super fast.

Except for claims early in the hand, this is probably not going to buy much. On the last few tricks, GIB generally plays pretty quickly, since it has a pretty good count of the hand.

[basy]

Oh, cmon, nothing, nothing at all comes near to taking out as much random factor as switching declarers would. argh

[junyi_zhu]

would it be possible to obligate people to play per 4 boards instead of 1?

 

I can't force them. What if they quit after , say, 1 ?

You can implement a feature to enforce both players to deposit a certain amount of money before each hand (or every 4 hands ) (or each rubber if you want to implement a rubber bridge). So when one player quits in the middle of the hand (or rubber), gib would finish that hand for the guy who leaves and the other player and bbo would share the deposit. This also solves the problem of frequent quitting when strong opp declares a game or a slam, cause strong opp would declare better than gib does. Also, this is really a simple method to implement a fair rubber bridge, which would attract more players and penalize those frequent quitters and offers bbo profit from them. Of course, the amount of the deposit should be tested to have a effect to penalize the quitters, but not to penalize too heavily against those unlucky players who occasionally disconnect from internet.