Join Bridge Winners
All comments by Alan Frank
1 2 3 4 5 6 7 8 9 ... 261 262 263 264
You are ignoring the author of this comment. Click to temporarily show the comment.
It also avoids the problem of a kind person transcribing the auction to make it clearer, but getting one of the calls wrong.
20 hours ago
You are ignoring the author of this comment. Click to temporarily show the comment.
Or the bidder could give his/her screenmate either the bidding card, to be placed in the tray by the screenmate, or the alert card, to be returned to the bidder.
July 18
You are ignoring the author of this comment. Click to temporarily show the comment.
It is not clear to me that the learning techniques which work well for games such as chess and go are the right way to go for bridge, with one exception. I believe that the greatest gains in the play can be made by attempting to model the other players' views of the hand (with defensive signalling being an important specific example).

For bidding, I'd like to see a bot do better on early rounds than blindly making bids based on shape and point count, and on later rounds, the current strategy of simulating a bunch of hands and deciding the optimal contract seems good, except that the simulation needs to be single-dummy, and this is an area where machine learning may do well.

If resources are available, I'd like to see teams working on multiple approaches and see what is most successful. It seems like the NukkAI group is well on the way with the AI approach, reinforcing my interest in trying an alternative. And I am sure that there are ways in which some work can provide value to both teams.
July 15
You are ignoring the author of this comment. Click to temporarily show the comment.
@Hank: With regard to BG checker play, I would expect that in a match against a stronger player, I would be more likely to take chances for a gammon, whereas against a weaker player, I would want to just grind out the single point.
July 15
You are ignoring the author of this comment. Click to temporarily show the comment.
I believe there is an error in the analysis of the double-elimination KO. If you play full-length matches, you have, as noted in the OP, 4 winners and 8 losers after two sessions. But after the third session, you have 2 winners and 6 losers (not 4). Then after the fourth session, you have 1 winner and 4 losers.

In order to get it down even to six sessions, you need the losers' bracket to play shorter matches. Suppose you have 16 teams:
After session 1: 8 winners, 8 losers
Session 2, losers play two short matches, ending with 4 winners, 4+2=6 losers
Session 3, losers play three-ways with one survivor, ending with 2 winners, 2+2=4 losers
Session 4, losers play two short matches, ending with 1 winner, 2 losers
Session 5 is the losers' bracket final
Session 6: winner plays the undefeated team for first.

Alternatively, you can play full-length matches until you get down to the last four losers and have them play short matches in the fifth session.

Yet another possibility is to stop the event after four sessions and declare the undefeated team as the winner with the losers' winners as tied for second. (Not quite the same as the repechage, as it allows losers in sessions 2 and 3 to also keep playing.)
July 15
You are ignoring the author of this comment. Click to temporarily show the comment.
Partner will not expect this much defense if you open 2. Partner will expect more HCP if you open 1. And partner will not expect this much playing strength if you pass. The reason I choose pass is that it offers the best hope of recovering, as it is easy to force to game if partner shows a dececnt hand with a fit for one of your suits.
July 11
You are ignoring the author of this comment. Click to temporarily show the comment.
A balanced hand with 8HCP wouldn't meet the rule of 18, which Ian states is the requirement.
July 11
You are ignoring the author of this comment. Click to temporarily show the comment.
Declarer will have already taken nine tricks by the time he notices the jack of clubs not dropping, so he can't get any fewer than that. And at that point, I don't see any plausible way for him to get more, even if we let him rethink his plan.
July 10
You are ignoring the author of this comment. Click to temporarily show the comment.
Presumably East thought that the agreement was “majors,” since that's what he said. And from his hand, that seems more likely than natural. So why not throw a monkey wrench into things by passing?
July 6
You are ignoring the author of this comment. Click to temporarily show the comment.
Even if South does think 4 is going down, he is clearly not doubling on a trump stack, so Reese's argument doesn't apply here. North didn't want to save at unfavorable, but when he heard South express the opinion that it was N-S's hand, he bid hoping for a make or at worst -200 instead of -420.
July 6
You are ignoring the author of this comment. Click to temporarily show the comment.
I agree with Craig–it also makes it harder for the opponents to get involved in the auction when it's their hand and provides a bit more bidding room when it's ours, either to search for slam or to enable 4=please transfer to your suit and 4=please bid your suit.
July 3
You are ignoring the author of this comment. Click to temporarily show the comment.
If East is 1=4=4=4, West would be 3=5=3=2, not the shape in Oren's reply.

Still, it is interesting that Kit, who always stresses the importance of having a complete auction and correct spot cards so we know what inferences can be drawn, provides a made-up auction. So I am not sure what we can say about West's likelihood of interfering.
June 29
You are ignoring the author of this comment. Click to temporarily show the comment.
If you find the wording of the Laws confusing, you can also use common sense. Assuming BWS, after {1} (1) X, partner knows you don't have Axxx x KQJ Kxxxx, AJxx Kx Axxx KJx, or Kxxx - Kxxxx Kxxx because of the withdrawn call, so there is a penalty. If you instead overcall 2, partner has no extra information from the withdrawn call, so the replacement is comparable.
June 28
You are ignoring the author of this comment. Click to temporarily show the comment.
For example, I open 2 showing a three-suiter short in diamonds with opening values. Oops, it was actually my RHO's deal. He opens 1. Here, a double could be a comparable call.
June 28
You are ignoring the author of this comment. Click to temporarily show the comment.
@Ed: It is irrelevant to whether there was MI, but once MI is established and you need to determine what a likely result would have been in the absence of MI, I believe it is relevant.
June 25
You are ignoring the author of this comment. Click to temporarily show the comment.
Tough lead–any suit costs a trick.
June 24
You are ignoring the author of this comment. Click to temporarily show the comment.
@Michael: The purchase of lottery tickets does not necessarily demonstrate a lack of knowledge of statistics–we pay money for the fun of playing bridge; others pay money for the fun of playing the lottery. Also, utility functions are nonlinear. Suppose I had the opportunity to pay $1 for a one in two hundred million chance to win a hundred million dollars. It would make some sense for me to do this, as the loss of a dollar would have literally no impact on my life and the gain of a fortune would have a significant impact. However, it is not worth the effort for such a small fraction of “significant,” plus it's the principle of the thing.
June 21
You are ignoring the author of this comment. Click to temporarily show the comment.
@Monty: “Rarely” compared to Chris Compton, perhaps. Someone who plays at the club twice a month, a few weekend tournaments per year, and maybe a few days at an NABC if it is in a good location could well be serious about the GNT but have a MP count that does not well reflect their ability.
June 21
You are ignoring the author of this comment. Click to temporarily show the comment.
+1 to the pro for opening 1 rather than 1NT.
June 20
You are ignoring the author of this comment. Click to temporarily show the comment.
I've had this idea of searching (preferably with help from the BW staff) for all the articles where they both commented and editing it into a book.
June 20
1 2 3 4 5 6 7 8 9 ... 261 262 263 264
.

Bottom Home Top