[inquiry]

Elianna, on May 12 2005, 04:13 PM, said:

Quote

  I would let the entry with the highest minimum win, I think it's better to be consistant, 4 60s should be better than 3 80s and a 0.

I'm not entirely clear on what you mean by highest minimum.

Person A gets 20, 40, 100, 100, 100 for a score of 360.

Minimum on person A is "20" (minimum = lowest score)

Person B gets 40, 80, 80, 80, 80 for a score of 360.

Minimum on person B is 40 ("40" is lowest score)

Which has the "highest" minimum? person B because 40 is higher than 20.

[luis]

Elianna, on May 12 2005, 08:13 PM, said:

Quote

  I would let the entry with the highest minimum win, I think it's better to be consistant, 4 60s should be better than 3 80s and a 0.

I'm not entirely clear on what you mean by highest minimum. For example, with five questions:

Person A gets 20, 40, 100, 100, 100 for a score of 360.

Person B gets 40, 80, 80, 80, 80 for a score of 360.

Do you want Person A to win, because he has the highest number of scores that are the minimum of the 10 (positive) scores, (it could be so, following your example)?

Or, by looking at the words "highest minimum", do you want person B to win, because the minimum of his scores is greater than the minimum of person A's scores (that makes more sense to me)?

Or is there another interpretation that I am missing?

B wins because his minimum is 40 and A's minimum is 20. As I said I think it's good to value a consitant entry.
Againts what Justin says I'm strongly agaisnt time based tiebreakers since with that will be unfair for some particular timezones.

[awm]

My experience with these polls has been that someone usually scores pretty close to "perfect." So it doesn't actually matter that much which of these methods you use; often the result is the same.

I'd recommend to go by total score. This is what other polls seem to do. If the top people are tied for total score, then break ties by their minimum score (so 2 90s would beat 80 and 100). If they're still tied, you could take multiple people for the panel, or you could break ties by LAST week's score.

[luke warm]

no strong feelings on it, but ben's original (earliest entry) idea seems fine...

[Echognome]

awm's idea of using LAST week's score is going to provide the best measure of consistency. It also encourages people to participate each week. Bonus!

[shoeless]

1st in works for me - KISS

[pclayton]

Echognome, on May 13 2005, 03:31 AM, said:

awm's idea of using LAST week's score is going to provide the best measure of consistency. It also encourages people to participate each week. Bonus!

Agree. There shouldn't be any bonus for turning scores in earliest, just as there isn't in finishing boards quickly in real life.

[mike777]

pclayton, on May 13 2005, 10:20 AM, said:

Echognome, on May 13 2005, 03:31 AM, said:

awm's idea of using LAST week's score is going to provide the best measure of consistency.  It also encourages people to participate each week. Bonus!

Agree. There shouldn't be any bonus for turning scores in earliest, just as there isn't in finishing boards quickly in real life.

AGREE

[Echognome]

One point on the scoring. Although the scoring is vastly dictated by the expert opinion (and it should be), it seems that the scoring drops off very rapidly if it's not one of the top 2 choices. Most other polls give a smoother gradation. So whereas our scores are often 100... 90... 10... 0, you might consider 100... 80... 70... 50... 30.. etc. It doesn't make much difference as it's just for fun, but psychologically it's nicer.

[luke warm]

a great big round of applause for ben (and the panel, of course)... ben's explanations have been lucid and comprehensive... i know we all appreciate his work on the forums in general and this thread in particular

[whereagles]

Actually, Ben's choice of awarding percentages based on the nr. of votes each option had isn't perhaps the best. But it does have the priceless virtue of being 100% transparent.

[Blofeld]

An ideal solution might be to get the panel members to each award a percentage to the bids, and then take the averages (also consider that so far, the fact that there are 101 different scores available hasn't seemed to have any application, as everything awarded has been a multiple of 10% ; if taking averages this would clearly not be the case).

But this has the not insignificant disadvantage of demanding a lot more from the panellists.

[luke warm]

actually he *hasn't* always awarded points based on the votes... remember last poll?... i think that if you're gonna have someone judge the choices, etc, that person should have the leeway to choose what seems best, after weighing the logic used for each... that represents a change of mind for me, but that's the way i see it now

[Chamaco]

Is it possible to create INTO the " BPO - Bridge Poll Online" Section, several "subfolders", having one Subfolder for each Problem set ?

This way, all the problems, the proposed answers and the following discussions would all be nested into a foldr pertaining to a single problem.

It would be easier to navigate through the problem sets and hands. :-)

[Rebound]

IMO the scoring system is fine.

I am curious as to why there has not been more support for allowing both participants to be on the panel in the event of a tie. I was under the impression that the composition of the panel is more or less dynamic in any event. It makes sense to break ties if they are a frequent ocurrence, but are they? Just wondering really.