[gwnn]

we should open 30 polls. I don't think it is a big issue is it? It sounds like an exciting and new project that has never been done before!

[jdonn]

gwnn, on Jan 19 2010, 01:53 PM, said:

we should open 30 polls. I don't think it is a big issue is it? It sounds like an exciting and new project that has never been done before!

I think you are the man for the job. In fact I think this warrants a new temporary forum.

[gwnn]

I will open all the threads later tonight. (unless it was sarcasm - I am as able to detect it as Sheldon Cooper)

[jdonn]

Are you saying if my comment wasn't sarcastic you'll really do it? Gwnn you are my vote for honorary forum member of the year.

[Cascade]

helene_t, on Jan 20 2010, 02:48 AM, said:

rogerclee, on Jan 15 2010, 07:34 PM, said:

whereagles, on Jan 15 2010, 11:23 AM, said:

jdonn, on Jan 15 2010, 06:18 PM, said:

Sample size is too small.

Why?

n = 30 is the point where the student t distribution starts to get very close to a standardized normal, i.e. the point where small sample size effects start to dissipate.

30 is also the atomic number of Zinc.

He-he very appropriate analogy. Also, if a sample size of 30 is always enough, why bother letting everyone vote in general elections. Just pick 30 citizens at random.

I agree with Josh's criticism of Mikeh's method but I also can't find of a better way of easily addressing the issue. If GIB or some similar software could be manipulated to chose either pass or dbl with this hand it would be better (and could also give a larger sample size). Using BridgeBrowser is also an option but then we have the problem of defining which hands are sufficiently similar to include, as the BB database probably won't include this exact hand.

But nice that Han did the same analysis.

GIB can be so manipulated.

I have done simulations like that in the past.

Jeff Miller has done similar simulations which have been published in Bridge World regarding explore for major fit or bash 3NT decisions.

In some of the simulations that I have done there has been some bias from GIBs actions in the later auction after some particular actions. In particular I think in some situations I have noticed that GIB is particularly bad at doubling. I am not sure how generic this is.

[gwnn]

Yes sure. Why not? I am bored.

[Fluffy]

I suggest you PM to inquiry if you are doing so, so he creates a temporary forum .

Using BPO poll contest wich is dead is another option

[gwnn]

yea I'll just use that. I hope they will move it if it's especially bad. please don't merge them though.

[Fluffy]

moving 30 threads is not as boring as creating them, but comparable nevertheless

[jdonn]

Gwnn you are amazing. But you should be naming the threads "can of worms" imo. I will try to get to them later tonight, as any opportunity to bump up my post count by 30 is worthwhile.

Eugene your algorithm should deduct for singleton K/Q/J. Look at hand 17.

[gwnn]

boo hoo, it's spades! we're favorable! I think I'd open it to be honest but maybe this is just the euphoria from finishing the threads

[eyhung]

jdonn, on Jan 19 2010, 04:45 PM, said:

Gwnn you are amazing. But you should be naming the threads "can of worms" imo. I will try to get to them later tonight, as any opportunity to bump up my post count by 30 is worthwhile.

Eugene your algorithm should deduct for singleton K/Q/J. Look at hand 17.

So do you want me to model it based on expert-Josh judgement or random bridge player? I can add a deduction of 1 point for a stiff king/queen/jack into my model. But I think there are plenty of people who would open #17, it's certainly not out of line especially since the stiff queen is in hearts (seriously annoying the opponents with your 1S open).

[jdonn]

It's not some random special formula I made up to deduct for singleton honors. I'm pretty sure almost every bridge player does it!

[gwnn]

I'd prefer the Bowles method to the jdonn non-random formula.