Join Bridge Winners
All comments by Helene Thygesen
You are ignoring the author of this comment. Click to temporarily show the comment.
I prefer 2. because it uses the information from the hands where he didn't use the five finger signal as well. His probability of holding a 5-card suit may differ from 44.34%, for example becasue holding a 5-card suit would make him less likely to defend. As said before, I don't think the Fisher test is the best test, but that is a relatively minor issue.
Oct. 20, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
I don't understand what you are getting at, Nicolas. Obviously you must enter the data in whatever format your software expects. All software I know of requires what you call a,b,c,d, but if it expects for example a+b+c+d, a+b, a+c, a, then you just enter it that way.
Oct. 20, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Nicolas writes:

Q: The “some (subjective) judgement” is probably going to bother some people.

A: Well, there is no way around it. Suppose you encounter an isolated tribe and find that of 10 random newborn children in that tribe, 9 had brown eyes. You also found that 9 were girls. Would you conclude that almost certainly more than half of the population has brown eyes? What about gender? Your answers to the two questions will be different because of what you a priori deem plausible. So you just cannot say objectively (i.e. purely on the basis of data) that the probability of something is x%.

Q: Is it possible that you can translate “p-value of 18%” into something that non-statisticians could understand? 1 in X? Or XX.YY%?

A: A p-value of 18% means that if we used this as a threshold then 18% of all innocent bridge players would be convicted. So obviously this is too high a thresshold. Maybe a thresshold of 0.01 or 0.005 or such would be ok if applied only to players that are already strongly suspected for different reasons, but if used for screening probably a thresshold of 0.00000001 or such would be reasonable

Q: For the Fisher test, you should be able to change the arguments - there are only 4.

A: No, the Fisher test only relies on the data, there is nothing you can change. The question is whether the Fisher test is appropriate. Personally I don't like it except in the contrived situations where the margin total (i.e. the number of deals in the sample on which five card suits held, and the number on which the signal were sent) is known to be fixed for the experimental design.
Oct. 20, 2015
Helene Thygesen edited this comment Oct. 20, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
The p-value is the probability, given that the signal is random i.e. not related to the holding of a 5-card suit, that we would observe such as a strong correlation as we actually found.

Note that a p-value of 18% (as in this case) does not mean that there is 18% chance that BZ are innocent. That probability can't be calculated from the data as it requires some (subjective) judgment.

The issue being discussed here is how to calculate the p-value. Most statistical packages offer you the choice between a Fisher test (which is an exact method but relies on the dubious assumption that the total number of hands on which the signal would be given would always be 8), and the chisquare test which is based on more reasonable assumptions but is only approximately correct.

Bill suggests that the “coin” package can do an exact calculation based on chi-square like assumptions but I am afraid that that isn't true. The code
chisq_test(as.table(matrix(c(14,9,2,6),2,2)),distribution=“ex”)
just gives the Fisher test.

It is not so difficult to implement the exact chi=square test, it took me about half an hour. But it would be nice to have an implementation that has been peer-reviewed so that people can trust the calculations to be correct.
Oct. 19, 2015
Helene Thygesen edited this comment Oct. 19, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
I don't think it matters what club players think about. What matters if what prospective club players think about, for example those who played a bit of Whist or similar in their youth and now consider taking up bridge when they retire.
Oct. 19, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
It's a lot of work to learn a foreign language. I have colleagues who have been living in the Netherlands for years who still don't speak the language. So I certainly wouldn't expect most players to learn the language just for a total of a few hundred days of Dutch bridge tournaments. Expecting someone to learn the language just for the purpose of playing bridge for a week every year is absurd.

When playing in Italy, I learn the numbers and the suits in Italian. When this does not suffice, some opponents can enough English or French for the communication to go through, but otherwise we usually manage using some improvised sign language or scribling symbols. We can ask a kibber to translate as a last resort but it is rarely necesarry.

Making system agreements with a partner with whom I didn't have a common language was a bit more challenging, but we managed. It is the charm of bridge that although it is a highly communicatie game you can play with people who speak different languages.

This is really a non-issue.
Oct. 19, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Strange. R can only do the Fisher test which gives p = 0.1134, and approximative chisquare which gives 0.1809. Reading the SAS manual and seeing how they implemented the exact chisquare test, and doing the same in R, I get 0.149283. But if I run your code, it just gives me the fisher test.

So you can do the same with http://www.socscistatistics.com/tests/fisher/default2.aspx and also in spreadsheets like openoffice calc, maybe even Excel. But SAS should somehow be able to give you the correct p-value.

This confirms that R>>>SAS. At least R has the guts to admit that they use the wrong formula :)
Oct. 19, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
While a BBO team match may not be the Cavendish, there are people on BBO who make substantial amounts of money by cheating. At least one player was caught doing so.
Oct. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Depends on system. If 3 on the previous round would have been forcing then this is ERKCB. But maybe this is the only way to show a slam invite in hearts with 3-card support. OTOH maybe it would have been better to bid 5 on the previous round to avoid confusion.
Oct. 5, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
I can understand that the TD was afraid of losing a pair but it is a very short-sighted solution to adjust the score. From now on everyone will be asking for adjustments whenever opps do something slightly unusual (bidding a slam with two missing aces, four-card overcall, preempt with a bad suit, underlead an ace against a suit contract etc.

Take a serious discussion about whether Canape systems should be allowed at this club. Probably you will decide to allow it because it would be impractical for a small club to have its own convention charter.

But by all means, make SAYC-only evenings if there is a market for it. Those would have to be individuals, probably, since otherwise you have the issue with pairs turning up having all kind of weird agreements, thinking that they are part of SAYC.
Oct. 2, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
If I won an event with cheating teammates I would resign even if I was 100% sure that I could have won it without the cheating I pair. I mean, otherwise we may as well award the gold medal to the team that a priori is most likely to win. That way we wouldn't have to waste time playing cards.

Easy for me to say as I am not in that situation. Still I think that is the right thing to do, unless it is widely agreed that F-N didn't cheat in that particular tournament.
Sept. 23, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Allan, is it really that simple that you always know if you have two LA's? On the bridge law forum, maybe the most common theme of discussions is which alternative is suggested by UI in a particular case. Then there are discussions about whether there is UI at all, and whether the non-suggested alternative really was a logical alternative at all. These discussions almost always generate opposing views, both from the players' and from the AC's point of view. So I think getting rid of UI would be a huge advantage.

OK, what is lost is “not easy to describe”. I really have no clue what it is about. I can understand that something is lost when playing online with four players in four rooms, and I can understand that something is lost when playing with screens as opposed to without screens. What is lost in this proposal I just don't understand. But maybe if I tried it I would feel that something was lost. Who knows. Taking clues from the temperature of the bidding tray to gauge my non-screenmate opps' stress level, maybe?
Sept. 23, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
I still don't understand the opposition to this proposal. You still feel your screenmate's vibrations and if you insist on playing with paper cards it can easily be implemented as I suggested. So what do you lose? You don't feel the vibrations of your non-screenmate opp, that is the only thing you lose. I think the advantage of not feeling partner's vibrations must surely outweigh this. Those who disagree with this should be opposed to playing with screens also.

Maybe we should not “let the cheaters win” and continue playing as if nobody would ever cheat. But even then, it would be a huge advantage not to have to worry about whether you are allowed to follow your intuition when you are not sure if it is partially based on partner's vibrations.

And then there are all the other advantages not related to UI.
Sept. 23, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
This reminds me of an article in the Dutch BF's magazine by Ed Hoogekamp about his experience from online bridge. Playing a slam with a clueless partner as dummy, he decided to go for a Vienna Coup instead of the simple finesse in the trump suit, because if the coup failed he wanted to give his p the opportunity to point out that “you should have taken the finesse”. So he starts by onblocking dummy's trump ace. When the king fell under it (offside), Ed's partner just said “claim, p”. So that was the title of Ed's article: “Claim, p”.
Sept. 23, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
A car key is a communication device so obviously it must be forbidden. But pacemakers we should condone.
Sept. 22, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
hi Robert,
I haven't heard about it but there are shuffling machines that use standard cards without barcordes, so it must be possible to buy the code that recognizes the cards. Otherwise the bar code reader functionality of the iPad could be used. I think it will consistute a very small part of total costs of building all the software we are talking about in this thread.
Sept. 21, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
To those who are afraid they will be missing holding the physical cards: would you be ok with an iPad solution in which you hold paper cards in your hand and play them by presenting them to the iPad's camera? That would mean you only needed one iPad in each room so it would not necesarilly be more expensive than the pure iPad solution.
Sept. 21, 2015
Helene Thygesen edited this comment Sept. 21, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
This may be somewhat offtopic but I have experienced a different kind of psychological stimulus to cheat. I was in a partnership in which my partner would (in my perception) haras me if I made wrong choices or if we got a bad score for whatever reason. Of course I would avoid eyecontact but if I heard an angry sigh when I led an honour that held, I simply didn't dare to continue the suit. Of course that kind of psychological stress gave much more bad boards than good ones, so typically we ended up with about 10% less matchpoints than either of us would normally get in other partnerships. But this is by no means an excuse since the particular pair who was victim of my unethical switch would be damaged.

Maybe more to the point: on the very few occasions that I have told my p that I had to make the wrong choice because of some UI I received from him, I had to mobilize a lot of courage. It is certainly not something I would be able to say in any partnership. If it is a genuine boarderline situation then OK, I can make the ethical move and then afterwards just say that I flipped a mental coin. But sometimes the ethical thing to do is not what I would normally do. In such a situation, it is easier to cheat than to be honest. Unless you have an unusally high level of partnership trust.

This is all from the perspective of an average club player. I appreciate that things are different at the top level.
Sept. 21, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Paul: next in line (after Denmark) is France.
Sept. 17, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
I would assume support doubles to apply if playing Walsh. If not playing Walsh I would not have a clue and would not make this double.
Sept. 11, 2015
.

Bottom Home Top