comparing systems

[straube]

I've received a suggestion that we pause after 25 deals for a few days to give folks a break. Happy to do so if wanted. How does everyone feel about this?

[relknes]

I've received a suggestion that we pause after 25 deals for a few days to give folks a break. Happy to do so if wanted. How does everyone feel about this?

That's fine. I have some system tweaking to do anyways. This has been a very helpful exercise!

[rbforster]

I would also suggest that, to the extent there is competitive information where you provide spoilers for different responding bids, that you leave those up for people who might want to contribute later. You can add a second post and/or an additional spoiler that shows the whole hand and result summary afterwards.

[straube]

Zelandakh did some analysis of the 25 deals thus far. He compared imp results of each system to a natural system and came up with some raw figures. I don't precisely understand what those figures represent (imps I guess) but he remarked that they are pretty subjective, depending on the actual hand layouts for the defenders and such.

 

I've always wanted to make sure this didn't turn into a contest or it would increase the likelihood of resulting action, but I think most folks want to see some sort of score assigned to their system. So do we want me (or Zelandakh) to post the results?

 

My feeling about them is that 1) Zelandakh was very likely being fair and 2) they are not very meaningful. Personally, as we went through the deals I'd have reactions like "This system really knows while this other system is guessing" or "This system shuts out the competition or this system preempts itself" and I think that we each had that sort of thought process and that's more meaningful than a raw number. Still...

 

But would we like to see it? Or would folks like me to send individual scores to each person? Or post results in spoilers? Let me know.

[Free]

I don't care about the scores. All I'm interested in is to see how other strong C systems work (for example with semipositives and double negatives) and have some friendly discussions about difficult auctions. I'm already happy with the contracts I reached so far, and I hope that our vanilla system van keep this up (otherwise we'll have to invest some time to improve this part).

[Kungsgeten]

I'm interested in reading the results. On my part, however, I think it is more interesting to see the differences in other types of hands in the systems (not a strong opener). The most interesting auctions so far has (according to me) been those where some systems opens strong while some do not. In a big club context, the main differences between the systems seems to be the other modules in the system, not the 1C one. However I guess this is harder to compare since the differences probably would be pretty large. I'm thinking nebulous diamonds vs natural diamonds, 4-card vs 5-card majors, transfer openings, (feather) light vs sound, canapé, multi-way openings etc.

[hrothgar]

I am somewhat surprised at the low number of contested auctions over the strong clubs

 

I agree with Kungsgeten that the major differences will involve preemptive strucutres and constructive openings.

[rbforster]

I think scores with spoilers would be good, interesting despite the obvious caveats.

 

I am somewhat surprised at the low number of contested auctions over the strong clubs

I believe straube's deal selection algo removes any with 2nd hand interference, which is more common than the 4th hand actions we saw.

[straube]

I think scores with spoilers would be good, interesting despite the obvious caveats.

 

 

I believe straube's deal selection algo removes any with 2nd hand interference, which is more common than the 4th hand actions we saw.

 

Yes. That's what I did and there was some discussion of this. My main objective was to see how the arrangement of initial responses (e.g. the meaning of 1D) affected outcomes. I've thought occasionally to run a separate study of coping mechanisms for 1S and other interference...but this would be an entirely different task.

[Zelandakh]

To be honest, I had not intended this "results list" to be made public. I did it as an exercise for myself and thought straube might be (privately) interested in it too. If we had publicised results then I feel that the desire to reach the "top spot" will be even higher than it currently is. Most systems can reach a particular end-contract if you work hard enough at it; even more so for systems with reverse relays when you know from the off which hand should be doing the asking.

 

What I am quite certain of is that the results we have reflect much more closely the pesonalities of the posters than the relative merits of the systems. In this respect I have been disappointed. To my mind the exercise has not produced the results that straube set out, at least not in terms of the final contracts. On the other hand, you can read between the lines to get a reasonable idea of which systems are bidding confidently in some cases. But not all - I am quite sure that some of our most confident posters have options and bracnhes along the way that are not mentioned, some of which would not result in the top spot being reached. Whereas some other posters are very concerned with saying where there is doubt, sometimes even to the point of not producing a final auction (which means I could not give them a score for the hand).

 

As I wrote before, I think these threads are an interesting window on the personalities of the contributors. I have not seen much value from a system theory point of view. Now that straube has switched from SCREAM to IMPrecision, it would be rather interesting to see him post auctions for that in his style (in 6 months perhaps) to see if he is bidding these hands with more confidence in the new system. My guess is that the difference would be minimal because it is straube's style to be honest about his doubts and bid each hand as it comes rather than with an end-contract in mind.

 

One final example which I think I can make without upsetting anyone. Justin's Meckwell Lite is currently sitting in 9th place, and this is very likely to go down as other posters complete the rest of the 25 hands. Hands up who thinks this system, or perhaps Justin himself, is the 9th best taking part. It is not that Justin has bid "badly" - he has gained more than IMP a hand by my reckoning - but rather that everyone else has managed to find the top spots more regularly, in some cases every time, to give results in excess of +2 IMPs per board. Seriously, you guys need to get in a JEC match - you will slaughter them! Then move on to a World Championship...

[straube]

I misunderstood and thought Zelandakh wanted me to preview his results before possibly sharing them with the group. Sorry Zelandakh. Well, I won't post them then. If someone is very curious they can email me for them, but I would caution that (for various reasons) they are mostly meaningless.

 

I've personally benefitted from this exercise and it has swayed my thinking a bit in several respects. I'm certainly less enthusiastic about SCREAM's 1H semipositive; it seemed mostly to preempt our auction and never produced a gain in subsequent competition.

 

I think a tendency for all of us to reach the par spot was to be expected, but I also noticed very many instances where folks submitted rather pessimistic auctions. I think I encouraged folks to give themselves the benefit of the doubt as much as I questioned outcomes that I thought were unlikely to be reached. Anyway, I agree that there is bias in the outcomes, but I think we did rather well considering. It would be lazy to look at some number at the end and announce a winner. I think, especially for such a small number of deals, that one has to look at how each system handles individual deals. Twenty-five is not nothing.

 

Thanks for all who participated. I'm still mulling whether or how to continue at some point. Suggestions welcome.

[hrothgar]

I consider myself lucky that I have Jack to sanity check some of my decisions

[Zelandakh]

I consider myself lucky that I have Jack to sanity check some of my decisions

The biggest differences between you and Jack came on Hand 3 (where Jack missed a game); Hand 8 (where Jack did something silly) and Hand 16 (where Jack made a SP response and you a GF one). These small changes made a massive difference in the IMP results. Incidentally, I do not have final results for you on hands 10 and 25, although I doubt either hand will result in a major swing.

[JLOGIC]

I enjoyed participating, it helped me find a couple of areas where I don't know exactly what I play or I need to improve on what I play (mainly, 4S 5H auctions after 1C 1H 1N should obviously be able to show their shortness below 3N imo, wtf!). Reading others auctions helped me see some plus sides and downsides for those systems. Even if I'm not a relayer maybe thats the way the world will go down the road and I don't want to be an old dinosaur not understanding it or adapting.

 

FWIW I would say I'm very lucky to even be 9th currently, there was one hand where I had to just play 6 because I didn't know if partner had 3rd round club control or not and all the relayers did know, and would have bid 7 opp the third round club control (which I would not). It would not surprise me if over a long enough sample I would be close to last in this because relays should kill me in slam bidding and esp grand slam bidding even if I am a good slam bidder in my system.

 

As zelandakh said I don't think this means my system is bad, the whole 1C 1H thing often allows minimal information leakage when not in the slam zone which is good, the system is incredibly easy to learn/memorize which is a plus for me, and it allows wiggle room for judgement and tactics (is this good or not? I don't know but I like it, maybe it is an unsound reason to like something though). Basically, if you're happy with your basic system and it fits your personality and memory load preferences, then go with it. This excersize can help you find some holes that you need to fix though which is very valuable.

 

Thanks for all the effort straube, and all the participants. It was enjoyable.

[Free]

It's not finished Justin, there are at least 75 more deals waiting for us :)

[straube]

It's not finished Justin, there are at least 75 more deals waiting for us :)

 

Nice to hear a vote for that. I was thinking of breaking for awhile and in the meantime looking at 1C (1S) and other interferences. Is there any interest in that? What I was thinking is that we could offer our own coping mechanisms and then have folks who haven't suggested such (i.e. aren't biased) bid or help bid those hands for us.

[hrothgar]

Nice to hear a vote for that. I was thinking of breaking for awhile and in the meantime looking at 1C (1S) and other interferences. Is there any interest in that? What I was thinking is that we could offer our own coping mechanisms and then have folks who haven't suggested such (i.e. aren't biased) bid or help bid those hands for us.

 

I'd be most interested in dealing unconstrained hands and seeing what happens.

[hrothgar]

One thing this thread has done is get me a bit more interested in relay auctions again.

 

I just updated my copy of Jack and am looking through the system it has implemented.

I'm considering etending my notes to cover the MOSCITO variant that Jack uses.

 

Who knows, I might be able to get the developers to ship an e-book version of the notes with Jack.

(I would want to get them to implement a more basic response structure to provide an easier on-ramp for new relay bidders)

[Mbodell]

it allows wiggle room for judgement and tactics (is this good or not? I don't know but I like it, maybe it is an unsound reason to like something though)

 

I think it is a very sound reason to like something. After all your judgement is likely to be significantly better than most peoples. If we played the same system and it had strict rules with no or nearly no judgement then you'd end up in the same place as everyone else, including all those with much worse judgement than you.

 

I did really enjoy the 25 and hope we can keep doing them. I'm interested both in what my system bid is, and also how my partner and I can bid them (because your system is no good if you don't know it or can't use it). I also thought it was useful to see how everyone else bid them too.

[awm]

I suggest that you give the two hands in separate spoiler text (same thread is fine) so we can actually bid them with partners. Might reduce the bias... and we can always still look at both hands if that is easier.

[nige1]

I suggest that you give the two hands in separate spoiler text (same thread is fine) so we can actually bid them with partners. Might reduce the bias... and we can always still look at both hands if that is easier.

In the past, when we were developing a system,

1. We chose a set of hands from a magazine competition e.g. Bridge World's Challenge the Champs
2. I would write down my bids on all the hands where West had the first bid and email the result.
3. Partner would simultaneously do the same on all hands where East had the first bid.
4. We would keep exchanging emails until all auctions were complete..
5. We would keep notes of uncertainties/alternatives and discuss them afterwards.
6. We would agree on a "correct" auction for each pair of hands (Often this wouldn't be an auction to the par contract)

You don't need to treat bidding Straubes hands as a bidding competition. It's a helpful exercise, anyway. For example, you can plug holes in your methods, as and when you discover them. Please would Zelandakh share his results as he goes along?

[akhare]

I'm considering etending my notes to cover the MOSCITO variant that Jack uses.

 

Please, please :D...

[relknes]

I am currious how "Phantom Club" would perform compared to other strong/multi club systems.

For those who don't know, Phantom is an approach that treats all auctions as competitive, with 1♣ being a takeout double of clubs. Opening bids are light, usually with stronger lead directing properties than standard openings, and some hands are passed in first or second seat that others might open if the long suit is weak. This system will vary considerably from partnership to partnership depending on their style of overcalls.

I expect that it will not do so well in the slam auctions, since it is geared toward competitive bidding.

[straube]

I am currious how "Phantom Club" would perform compared to other strong/multi club systems.

For those who don't know, Phantom is an approach that treats all auctions as competitive, with 1♣ being a takeout double of clubs. Opening bids are light, usually with stronger lead directing properties than standard openings, and some hands are passed in first or second seat that others might open if the long suit is weak. This system will vary considerably from partnership to partnership depending on their style of overcalls.

I expect that it will not do so well in the slam auctions, since it is geared toward competitive bidding.

 

The basic idea of these deals was to compare strong club systems. Some systems that had an element of strength to their 1C (i.e. Polish club) entered auctions and some systems that used a strong club along with other openings (2N as 20-21 for example) also participated. Forcing pass? Well...

 

But I'd rather not expand this as far as Phantom Club or other systems that don't have an element of a strong club opening because then we're comparing apples and oranges.

 

The other thing, for late entrants, is that there is no review process. As I posted the deals, we kind of bid the hands together. As we're taking a break now, the group element is not present...and primarily is the reason I haven't been updating the outcomes so as to include late entrants.

[relknes]

The basic idea of these deals was to compare strong club systems. Some systems that had an element of strength to their 1C (i.e. Polish club) entered auctions and some systems that used a strong club along with other openings (2N as 20-21 for example) also participated. Forcing pass? Well...

 

But I'd rather not expand this as far as Phantom Club or other systems that don't have an element of a strong club opening because then we're comparing apples and oranges.

 

The other thing, for late entrants, is that there is no review process. As I posted the deals, we kind of bid the hands together. As we're taking a break now, the group element is not present...and primarily is the reason I haven't been updating the outcomes so as to include late entrants.

Ok, sorry I posted the first two already. I will stop now. I had thought Phantom Club would qualify since you start with a 1♣ "takeout double" with any hand too strong for a smple overcall.