[enigmisto]

I recently read the book "Good, Better, Best" by Jan Eric Larsson, in which an AI researcher programs bots to understand a variety of bidding systems, and pits them against one another.

Generally speaking, the Precision systems outperformed Standard and 2/1. Interestingly, the top bidding system (excluding the author's own unpublished system) was the Blue Team Club system, a strong-club canape system that predates Precision.

I know there's a strong overlap behind the kind of people who enjoy artificial systems and the kind of people that are "data-driven", so given this result, I would expect to see a resurgence of interest in Blue Club. After reading Good, Better, Best, I was certainly very eager to learn more about this system that had bested the others.

I found a few English resources for Blue Club, which I have really enjoyed reading:
* The original English translation for the system: https://smile.amazon...uct/0571092659/
* A description of the system written by Omar Sharif: https://smile.amazon...uct/B000GX0ECA/
* A modern, somewhat simpler adaptation: https://smile.amazon...a/dp/1771401893
* A free English translation of another modern version of the system: https://bridgewithda...2000_gmeier.pdf

All the links given in the last document for the Blue Club community appear to be broken. Is anyone here aware of a current online community of Blue Club players?

[hrothgar]

https://smile.amazon...uct/0571092659/ is quite good

[johnu]

The original Blue Team Club book

The Italian Blue Team Bridge Book

This book has a lot of detail about various auctions, including canape auctions.

There was a short book (or section of a larger book?) on Neapolitan Club which was the predecessor to Blue Team Club. The main differences were 1♦ response to 1♣ showed 0 controls, 1♥ showed 1 control, etc. Opening 2♦ was just a weak 2 bid, and 1NT didn't have the strong any balanced distribution NT option. I remember checking out the book from the public library way back when I was in college.

Can't help with links to any Blue Team support groups. There are usually a few people in your local metro area that played some version of Blue Team at some point in their bridge careers, and there may be a pair or even several that still play a version of Blue Team.

[enigmisto]

 johnu, on 2022-May-12, 20:04, said:

There was a short book (or section of a larger book?) on Neapolitan Club which was the predecessor to Blue Team Club. The main differences were 1♦ response to 1♣ showed 0 controls, 1♥ showed 1 control, etc. Opening 2♦ was just a weak 2 bid, and 1NT didn't have the strong any balanced distribution NT option. I remember checking out the book from the public library way back when I was in college.

Interestingly, "Good, Better, Best" looked at Neapolitan Club as well; it fared quite poorly.

[blindsey]

Interesting. I thought the weakness of strong club systems to interference bidding was well-established. Did Larsson say anything about that?

[mycroft]

• The weakness of strong club *auctions* to interference is well-established. But quote from forever - "if I break even on 1♣ auctions, I'm happy. I'll take my wins from all the hands I don't open 1♣". 1M (Precision), in particular, is very EV+, especially in competitive auctions, over a 11-21 system.
  
• Computers are bad at "judgement". Even when an AI researcher trains a neural network. The more rules they have - which they will never get wrong, forget, or worry that their partner will forget, and which doesn't impact "memory load" and "energy" the way these kinds of system can in humans - the less judgement situations they have, and the less judgement they need to apply.

Both of those is why I played strong club systems in the past, and would do so again with a partner I could work on it with. I am (and was much more 20 years ago) someone who had (relatively) poor judgement and (relatively) little memory load for serious system and arbitrary sets of rules, so the tradeoff in 2. is a net win. I also know that my brain is wired weird, and this does not necessarily apply to 80-90% of bridge players.

I don't know what Larsson has to say about that, but I'm sure it was part of the result.

[hrothgar]

Please note:

I'd take Larson's results with a hefty grain of salt... (And don't expect a bunch of people to renew their interest in Blue Club based on this publication)

[straube]

Read Blue Team Club a long time ago. Forgot the details. My impression is that they employed canape frequently so as to differentiate strengths and shapes of both opener's and responder's hands, but they didn't have relays to complete these patterns. They also used control-showing responses to 1C and I think that has almost universally fallen out of favor; nowadays shape first and then controls. The Blue Team Club 1C opening was stronger than most strong club systems of today.

I'd recommend Dan's book on Meckwell Lite because it's a pretty good system, played by many and doesn't require too much memory work.

RMPrecision (Meckwell) has a lot of fans, but the notes are mostly kept secret.

TOSR (Transfer-Oriented Strong Relay) is one of the better systems using relays after the 1C opening and it has some interesting alternatives for 1C-1D continuations.

You could invest some time in IMPrecision by our own frequent poster's Adam and Sieong. It's quite complicated but my vote for the best system. It uses the 1D response to 1C for very weak or very strong hands so that the majority of in-between hands can start showing their shape right away.

[enigmisto]

 blindsey, on 2022-May-13, 05:02, said:

Interesting. I thought the weakness of strong club systems to interference bidding was well-established. Did Larsson say anything about that?

Yes, he had a section devoted to the question of whether strong club systems were vulnerable to interference. He programmed a number of different interference strategies, and found that none of them were particularly effective at diminishing the value of a strong club system. The more destructive the interference (e.g., bid 1S over 1C no matter what), the less it helped. The most effective interference strategies were ones that were plausibly constructive for the interferers in finding a useful contract to compete in, but even against these interference strategies, the strong club systems still outperformed the others. He concluded that his results effectively debunk "the myth" that strong club systems are defeated by interference.

[enigmisto]

 hrothgar, on 2022-May-13, 10:22, said:

Please note:

I'd take Larson's results with a hefty grain of salt... (And don't expect a bunch of people to renew their interest in Blue Club based on this publication)

Why? I was certainly surprised by some of the results, some of which directly contradicted what I have seen in other data-based analyses derived from actual human tournaments. But the methodology impressed me, because it keeps precisely constant all elements except the variables being tested, and does it across several thousand boards. I'm inclined to give his results quite a bit of weight.

I wish he'd open-source his program. I can imagine many further experiments I'd like to do. Furthermore, imagine how useful it would be for bidding system designers to be able to quickly test different variations and ideas!

[sfi]

 blindsey, on 2022-May-13, 05:02, said:

Interesting. I thought the weakness of strong club systems to interference bidding was well-established. Did Larsson say anything about that?

My experience is that they are vulnerable specifically to 3-level preemption. Interference at the 1- and 2-levels can be managed just fine with good agreements.

[blackshoe]

 sfi, on 2022-May-13, 18:41, said:

My experience is that they are vulnerable specifically to 3-level preemption. Interference at the 1- and 2-levels can be managed just fine with good agreements.

Not at all surprising, is it?

[sfi]

 blackshoe, on 2022-May-14, 21:38, said:

Not at all surprising, is it?

No, but it is a trope which deserves clarification - both if one is playing against such a system and if a partnership is considering whether to start playing it themselves.

[mycroft]

Which is, of course, why my suggestions to interfere are designed to get to 2♠ or higher, when there is a fit, before opener's rebid.

Yes, it's easier to deal with 1♣-1♠-(whatever)-2♠ than 1♣-2♠-(whatever)-pass, but it's still a problem.

As opposed to say "they don't get a cuebid, but partner needs two suits to pump-raise" systems.

[hrothgar]

 enigmisto, on 2022-May-13, 18:33, said:

Why? I was certainly surprised by some of the results, some of which directly contradicted what I have seen in other data-based analyses derived from actual human tournaments. But the methodology impressed me, because it keeps precisely constant all elements except the variables being tested, and does it across several thousand boards. I'm inclined to give his results quite a bit of weight.

I'm worried about implementation issues.

I know how many mistakes I am capable of making when I run a sim.

For that reason, whenever I post the results of a sim I also post my code so people can see what kind of assumptions I am making and check for mistakes.

Larson built a large complicated system that is producing counter intuitive results.

Who knows. Maybe he's right and his own bidding system and Blue Club are the be all and end all of bidding system design.
Or, alternatively, perhaps he had a blind spot or two when he designed his Sims or the competitive methods that the opponents are using or what have you.

I have no way of knowing which is true.

[etha]

If you took 4 Gorillas and taught them bridge I expect opening 3NT every hand would end up being the best system. I don't think bot AI is ready for this challenge. Make one playing any system you like that can beat say the Dutch team over 100+ boards and you can start testing systems.

[avonw]

The greatest weakness of Blue Club/Neapolitan is responder's canape. Consider this auction:

1S 2C
2S 3D

Is that a reverse into diamonds - a hand like xx xxx AKJxx AKx ?

If so, how do you bid xx xxx AKx AKJxx?

There is a reason no one plays Blue Club any more... it's full of holes. And the fact is, neither Forquet nor Garozzo nor Siniscalco played Blue Club/Neapolitan. Go through every WC book and look for the systemic violations. It was all smoke and mirrors... especially their takeout doubles. What they said in print bore no relation to what they did at the table.

[johnu]

 avonw, on 2022-June-16, 19:31, said:

The greatest weakness of Blue Club/Neapolitan is responder's canape.

Responder's canape would be a weakness if anybody played it (notwithstanding the original Blue Team). I had barely learned bridge when I started playing Blue Team Club and getting rid of responder's canape was one of the first things my partner and I did. Minimum non-shape takeout doubles was another thing that didn't last more than a couple of disasters sessions.

And nobody these days plays the "original" Blue Team Club since the system books were written 50+ years ago and many of the ideas have not stood the test of time. This was from an era of sound opening bids, conservative defensive bidding, there wasn't even Key Card Blackwood in the system.

[nullve]

 avonw, on 2022-June-16, 19:31, said:

neither Forquet nor Garozzo nor Siniscalco played Blue Club/Neapolitan.

Because?

[PrecisionL]

 johnu, on 2022-June-17, 05:24, said:

Responder's canape would be a weakness if anybody played it (notwithstanding the original Blue Team). I had barely learned bridge when I started playing Blue Team Club and getting rid of responder's canape was one of the first things my partner and I did. Minimum non-shape takeout doubles was another thing that didn't last more than a couple of disasters sessions.

And nobody these days plays the "original" Blue Team Club since the system books were written 50+ years ago and many of the ideas have not stood the test of time. This was from an era of sound opening bids, conservative defensive bidding, there wasn't even Key Card Blackwood in the system.

4NT responses were 1430 and with 2 aces the replies were per Roman Blackwood: Rank/Other/Color

I tried canape by responder and found it a poor design.