A scoring question

[Antrax]

(Feel free to move if this isn't the right forum)

So, last night at the club we were arbitrarily divided into two "sections" - each with its own set of table numbers, 9-10 tables each, eight rounds of 3 boards each. No arrow switching at any point. At the end of the evening, scoring was done for each section, then combined to create two winners, one for each axis.

Two questions:

a) What is the point of dividing into sections? I couldn't figure out how it's different from just having one section of 18 tables or however many would've been needed.

b) Does it make any sense to score matchpoints within each section and then mix the results together? It seems odd to me that on a given board, a score of 50 might be worth 80% to some pairs and 40% to other pairs seating along the same axis.

[paua]

(Feel free to move if this isn't the right forum)

So, last night at the club we were arbitrarily divided into two "sections" - each with its own set of table numbers, 9-10 tables each, eight rounds of 3 boards each. No arrow switching at any point. At the end of the evening, scoring was done for each section, then combined to create two winners, one for each axis.

Two questions:

a) What is the point of dividing into sections? I couldn't figure out how it's different from just having one section of 18 tables or however many would've been needed.

b) Does it make any sense to score matchpoints within each section and then mix the results together? It seems odd to me that on a given board, a score of 50 might be worth 80% to some pairs and 40% to other pairs seating along the same axis.

 

How can you play 27 boards with 18 tables ? If you play one-board-per-round you have to then do another half a round, not very fair. If you play two-boards-per-round then you won't complete the 36-board movement.

 

Yes, you can just matchpoint within each section, but you might as well have the same boards in both sections and get more comparisons.

[gnasher]

a) It reduces the number of different boards. With two nine-table Mitchells, you have two identical sets of nine boards in play. If, instead, you had a single 18-table Mitchell playing 9 rounds, you would have18 different sets of boards. Fewer boards means more comparisons, as long as you score it across both sections.

 

b) No, it's wrong to matchpoint the sections independently. Doing this takes away the benefit of having two sections.

[Antrax]

paua, 24 boards in eight 3-board rounds.

 

Thanks gnasher, those were my guesses too. Though I do wonder, if you have two sets of identical boards, can't you just number boards 10-18 as 1-9 again and compare? Nobody is going to play a board twice anyway.

[blackshoe]

(Feel free to move if this isn't the right forum)

So, last night at the club we were arbitrarily divided into two "sections" - each with its own set of table numbers, 9-10 tables each, eight rounds of 3 boards each. No arrow switching at any point. At the end of the evening, scoring was done for each section, then combined to create two winners, one for each axis.

Two questions:

a) What is the point of dividing into sections? I couldn't figure out how it's different from just having one section of 18 tables or however many would've been needed.

b) Does it make any sense to score matchpoints within each section and then mix the results together? It seems odd to me that on a given board, a score of 50 might be worth 80% to some pairs and 40% to other pairs seating along the same axis.

a) The point is to get a reasonable number of boards in the time available. An N-table Mitchell movement requires N rounds to be complete. If you have 18 tables, you need 18 rounds. At two boards (and roughly 15 minutes) per round, that's 4.5 hours for the game, plus any "admin" time. If you split into two sections, you can play nine three board rounds (27 boards) at 22.5 minutes per round, and be done in about 3.5 hours.

b} "Overall" scores combine both sections into one field, so the number of matchpoints anyone gets for any particular score on a given board will be the same. The downside (I think, I haven't looked at it in some time) is that the scoring program (I'm referring here to ACBLScore, other programs may not do this) treats NS and EW in a Mitchell as part of the same field when computing overall scores, which makes no sense unless you arrow switch, thus getting a "one winner" movement.

[gnasher]

a) The point is to get a reasonable number of boards in the time available. An N-table Mitchell movement requires N rounds to be complete. If you have 18 tables, you need 18 rounds.

You don't have to play a complete Mitchell.

 

An 18-table Mitchell ending after 9 rounds is better than two 9-table Mitchells scored independently. In either movement you play nine rounds against nine different opponents, with eight comparisons per board. The incomplete 18-table movement is better because each pair's score for the session is affected by the actions of the entire field, rather than just half of it.

 

The benefit of splitting into two sections exists only if you score across the sections. You still play nine rounds against nine different opponents, but you have 24 comparisons per board, and your score on each board is affected by the actions of every other player.

Edited December 2, 2012 by gnasher

[inquiry]

paua, 24 boards in eight 3-board rounds.

 

Thanks gnasher, those were my guesses too. Though I do wonder, if you have two sets of identical boards, can't you just number boards 10-18 as 1-9 again and compare? Nobody is going to play a board twice anyway.

 

To get dealer and vulnerablity right, you will have to use boards 1-9, and 17-25, not 10-18.

[paua]

How can you play 27 boards with 18 tables ? If you play one-board-per-round you have to then do another half a round, not very fair. If you play two-boards-per-round then you won't complete the 36-board movement.

 

Yes, you can just matchpoint within each section, but you might as well have the same boards in both sections and get more comparisons.

 

Another option for 18 boards is to do a Web. I think you can do 24 or 26 or 28 boards.

[VMars]

b) No, it's wrong to matchpoint the sections independently. Doing this takes away the benefit of having two sections.

 

I don't disagree with you about theory, but many clubs around here do that so that if a board is misduplicated or fouled in some other way, only one section is affected.

[VMars]

(Feel free to move if this isn't the right forum)

 

a) What is the point of dividing into sections? I couldn't figure out how it's different from just having one section of 18 tables or however many would've been needed.

 

 

Some clubs may not have boards numbered high enough for having all people in one section. We have sets of 36 boards so when we have 18.5+ tables, we HAVE to go to two sections. And then sometimes someone doesn't show up, so we have two nine table sections.

 

And yes, there are movements that use two sets of boards in one big movement. If you can trust players to move boards correctly, this is feasible but if you don't, then you can't do this.

[blackshoe]

You don't have to play a complete Mitchell.

 

An 18-table Mitchell ending after 9 rounds is better than two 9-table Mitchells scored independently. In either movement you play nine rounds against nine different opponents, with eight comparisons per board. The incomplete 18-table movement is better because each pair's score for the session is affected by the actions of the entire field, rather than just half of it.

 

The benefit of splitting into two sections exists only if you score across the sections. You still play nine rounds against nine different opponents, but you have 24 comparisons per board, and your score on each board is affected by the actions of every other player.

You don't have to complete any movement. The problem with your curtailed 18 table Mitchell is that not everyone plays the same boards.

 

I have no problem with two sections playing the same boards and scoring "across the field", so long as if there are no error switches there are still two fields - NS and EW. I do have a problem with ACBLScore's method (as I understand it) in which EW and NS are combined into one field without any arrow switches.

 

We had 17 tables today. Web Mitchell, three (I think, might have been two) sets of boards, 2 board rounds, 13 rounds, total 26 boards. Not "all play all" with only 13 rounds, but at least we all played the same boards. Two winners, neither of which were my partner and I. :D

[gnasher]

You don't have to complete any movement. The problem with your curtailed 18 table Mitchell is that not everyone plays the same boards.

But, for the purpose of producing a winner, the two-independent-sections movement has exactly the same problem. They may be the same deals in the sense of having the same cards in each slot, but if the matchpoints for a given result vary they're different boards.

 

I agree that there is a social advantage in having the same deals played in each section, but for comparison purposes it's definitely worse.

[blackshoe]

But, for the purpose of producing a winner, the two-independent-sections movement has exactly the same problem. They may be the same deals in the sense of having the same cards in each slot, but if the matchpoints for a given result vary they're different boards.

 

I agree that there is a social advantage in having the same deals played in each section, but for comparison purposes it's definitely worse.

Seems to me the matchpoints depends on whether you have an even or odd number of tables. We were talking, as I recall, about dividing 18 tables into two nine table sections. Boards in either section will have the same top.

 

If you have, say, nine and eight tables, there are some movements that, iirc, will give you the same top in both sections. And then there's factoring. Not ideal, perhaps.

 

I would say I don't agree with your last statement, but more likely I just don't understand it. :P

[gnasher]

I meant that given the choice between

(1) Two sections, each playing a full Mitchell, with the same boards but matchpointed separately

(2) One section, playing half a Mitchell

 

The advantage of (1) is that the conversations in the bar afterwards will be more interesting. The advantage of (2) is that it's more likely to produce the right winner.

 

I'm not talking about the size of a top. With two sections, if EW in Section 1 are on average worse than EW in Section 2, NS in Section 1 have an unfair advantage over NS in Section 2. Effects like that are reduced by having a single section with an incomplete movement.

Edited December 7, 2012 by gnasher

[mycroft]

I don't see much of the difference, gnasher, between two sections (with a biased E-W) and across-the-field scoring, and one section playing the same movement, with the N-S pairs that happen to hit the biased E-W section. Unless you think that the chance of there being a "weak 9" and a "strong 9" in an 18-table field over two self-selecting 9s, of course.

 

Yes, instead of "these 9 N-S get to play the weak pairs", you get "this one gets to play *all* the weak section, these two get to play 8-and-1 (and if that 1 is the weakest of the "strong" section, that isn't so bad), these two get to play 7-and-2, and so on. So maybe it's a bit mitigated. Of course, arrow-switching both sections would help there, as it does with the "oh my, there's no strong E-W pairs in this game" issue.

 

I do know that when I run two sections at the club, I will do a once-over the field to ensure that we don't have an overly-strong (or overly-weak) line; but when I run a 17, I'll do the same check to ensure there isn't a "murderer's row" that only half the field will have to handle all of.

 

But I haven't thought this through deeply; likely I'm missing something.

[blackshoe]

I meant that given the choice between

(1) Two sections, each playing a full Mitchell, with the same boards but matchpointed separately

(2) One section, playing half a Mitchell

 

The advantage of (1) is that the conversations in the bar afterwards will be more interesting. The advantage of (2) is that it's more likely to produce the right winner.

 

I'm not talking about the size of a top. With two sections, if EW in Section 1 are on average worse than EW in Section 2, NS in Section 1 have an unfair advantage over NS in Section 2. Effects like that are reduced by having a single section with an incomplete movement.

Nobody around here holds conversations in the bar after the game any more. :o :(

 

Ideally, I think, you would seed all four fields, however many sections you have. One director here used to do that by putting a card on every fourth table that said "only A players here" or the equivalent. Of course, these days we get about half the NS's reserved for people with physical problems. It is true, IME, that if you let players randomly pick where to sit, or if you have a lot of relatively good players with physical problems, you end up with unbalanced fields.

 

Monday, we had "overalls" in our one section Web movement (17 tables, 26 boards, 13 rounds, 3 board sets). The EW pair who did best in that field got more masterpoints than the NS pair who did best in their field. There was no arrow switch. This makes no sense to me, but the program (ACBLScore) allows it, so the directors here assume it must be okay. I suppose it is, if your purpose is to give away masterpoints. :(

 

Heh. Another interesting thing that happened, irrelevant to this thread: there was a scoring correction, and new results were sent out. On one board, two NS pairs got 1430 for a shared top. However, one table played the board in 6S by NS making. The other, according to the report, played it in 5D by EW, down one. I'm told that the problem is that while it's easy to correct scores after the game in ACBLScore, that doesn't affect what the Bridgepad system thinks happened, correcting the Bridgepad data is difficult or impossible, and the contracts come from the bridgepads. :unsure: :o

[gnasher]

Just to be clear, I'm comparing two pointlessly bad methods. The obvious way to run this event is with two sections, playing the same boards, and matchpointed across the field. Seeding would also help, but it seems a bit over the top for a club game.

[Antrax]

How can you seed playing duplicate? I thought it would only work for swiss-style.

[blackshoe]

Let's say you have ten tables. 50% of your players are stratum A. That's ten pairs. Insist that the A pairs sit either NS or EW at the odd numbered (or even numbered) tables. Half of these pairs will complain because they want a NS, but at least you've ensured that all ten pairs aren't sitting in the same (NS) field.

[Antrax]

That's not how I know the term "seeding" is typically used. Is it different for Bridge?

[blackshoe]

Well, maybe it's not strictly correct, but I can't think of a better term. I'm open to suggestions. B-)

[Vampyr]

Just to be clear, I'm comparing two pointlessly bad methods. The obvious way to run this event is with two sections, playing the same boards, and matchpointed across the field. Seeding would also help, but it seems a bit over the top for a club game.

 

Why do you think that this is definitely better than a one-section web movement? The latter seems to me to be a bit better.

 

To the poster who mentioned moving the boards correctly -- so long as you have more than one set of boards, you don't have to do any fancy moving of boards to run a web. Half of the stationary pairs play the boards from the top down and the other half from the bottom up, but the boards move between tables as usual.

[awm]

That's not how I know the term "seeding" is typically used. Is it different for Bridge?

 

The purpose of seeding is to make sure that in a tournament where not every team will play every other team, all the teams (or at least all the teams in contention) have an equally difficult "schedule." This takes a somewhat different form in a pairs movement than in a single or double-elimination type format of course.

 

For pairs, you generally want to spread the better pairs evenly through the field. This makes sure that no one gets an "easy section" (if there are multiple sections) or an "easy direction" (in a Mitchell type movement) or simply fails to reach the good pairs (in a movement with more tables than rounds). The director will also typically set things up so that if there is a "skip" in the movement, top pairs are not skipping over other top pairs.

[paua]

Well, maybe it's not strictly correct, but I can't think of a better term. I'm open to suggestions. B-)

 

Seeding = ranking the pairs top to bottom

 

The process of then sewing them evenly throughout the room is different.

Maybe this should be called scattering ? Spreading ? Dispersal ?

[gnasher]

Why do you think that this is definitely better than a one-section web movement? The latter seems to me to be a bit better.

Sorry, I didn't mean to imply that. I was only comparing different configurations of Mitchell movements. I don't know anything about the web movement, except that it's clever.