[whereagles]

Quoting from 2014 EBU White Book, which I think is still officially used.

3.1.3 Cross-IMP scoring for pairs events. No VP scales are available. A suggested method is to calculate the total cross IMPs, then divide by 70% of the number of scores, and then use the standard teams-of-four scale in §3.1.1: e.g. for 20 tables and an 8-board match, calculate the total cross IMPs, divide by 14 (70% of the number of scores) and then use the standard 8-board VP scale.The above recommendation is only an approximation. Further advice can be obtained from EBU Headquarters: see §0.4 for contact details. See §4.2.5.3 for details of rounding in cross-IMP scored events.

Questions:

1. Is there any motivation for the 70% figure? Why not 60% or 80%? Or N - 1 for that matter, with N number of scores?

2. Is the paragraph still recommended for the new 20-0 WBF VP scales?

Thx

[gordontd]

I'll see if I can find out more information (actually RMB1 may be able to help), but I'm fairly sure that 70% is there because it's approximately 1/sqrt(2), the relationship between IMP pairs and teams that leads to the standard amount for adjustments being 2IMPs for the former, rather than 3IMPs as for the latter.

[paulg]

The JSS scoring program factors the cross-imps by dividing by SQRT(rc/2) where r= results, c = comparisons to simulate Teams scoring (on the advice from the EBU). As Gordon says, the 70% seems to be a quick approximation to this. I suspect the name 'Max Bavin' will be included in any long answer, and possibly 'Probst'.

I don't think the new WBF VP scales make any difference. The basis of the factoring is that it converts the cross-imps on the board to a number of imps that is comparable to a team score on a board. Either you accept this premise or not. The translation of these imps to VPs is a secondary issue.

In Scotland, some trials convert cross-imps to VPs and some just total the cross-imps, normally with a cap. The new VP scales have changed the cap limit.

[whereagles]

Thx a lot. Cheers.

[Trinidad]

As a scientist, I hate this kind of wordings: "A suggested method", "The above recommendation is only an approximation", "70 %" (and why a percentage?!?). The way the text is formulated is the equivalent of saying: "To bake a cake for 9 people, it is suggested to take a recipe for 4 people and increase the needed amounts of ingredients by 100%. This recommendation is only an approximation."

sqrt(2)/2 is not 70%. (Like 9 is not 100% more than 4.)

Stating that 70% is only an approximation suggests that whoever came up with it picked a value that "felt good" without justification.

In reality, sqrt(2)/2 is not an approximation, just like 9/4 is not an approximation for adjusting the cake recipe. The factor sqrt(2)/2 follows directly from the mathematical model that the conversion of IMPs to VPs is based on. It simply is the mathematically correct way to generalize the applicability of the IMP-VP conversion to a Cross IMP-VP conversion, just like multiplying the needed amounts by 9/4 is the mathematically correct way to generalize the cake recipe from 4 to 9 people.

I don't have any problem with the fact that most people will not understand where this factor sqrt(2)/2 comes from or how the conversion from IMPs to VPs works. Everybody has their own specializations and talents and we can't all be math geeks (nor do we want everybody to be one). But if you are not a math geek and write the White Book, you have three options:

• You rely on the math geeks to know what they are doing, and state -in absolute terms, without casting any doubt- how to do the calculation.
  
• You don't trust the math geeks and don't mention it at all.
  
• You state that the math geeks say that you need to calculate like this, leaving it to the reader to decide whether math geeks are reliable.

But stating a mathematically wrong formulation (no matter how close the result will be to the correct one) and suggesting that this procedure is the result of some form of alchemy or black magic is not an option.

Rik

[gordontd]

Trinidad, on 2015-February-02, 23:12, said:

As a scientist, I hate this kind of wordings: "A suggested method"

The EBU often gets asked about regulations for events of a type that it doesn't run and for which it therefore doesn't have regulations of its own. It could just say "no, we can't help you" or it can provide "suggested methods" that event organisers can use or not as they choose.

Do you also object to the use of 2 IMPs as the standard amount for score adjustments in IMP pairs? Would you prefer that pairs are awarded +/- 2.12 IMPs (still an approximation) for unplayable boards?

[Trinidad]

gordontd, on 2015-February-03, 02:11, said:

The EBU often gets asked about regulations for events of a type that it doesn't run and for which it therefore doesn't have regulations of its own. It could just say "no, we can't help you" or it can provide "suggested methods" that event organisers can use or not as they choose.

The point is that there is nothing "suggested" about the method. It is the method. Just like there is nothing "suggested" about multiplying by 9/4 when you adjust the cake recipe for 4 people to 9 people: That is simply how you do it. No "hocus pocus", no suggestion that "we can only hope it works, there are no guarantees".

When traffic light turns red: STOP.
When converting Cross-IMPS to VPs: Divide Cross-IMPs by rc and divide by sqrt(2) and use the VP table for team of four IMPs.
When you want to turn right: Turn steering wheel clockwise.

Don't be afraid that you don't understand the intricate mechanism and electronics of the steering column, power steering and its influence on the position of the wheels. That is for the engineers to worry. Just turn the steering wheel. That is neither a suggestion nor a recommendation.

But somehow when a scientist says something, based on hard cold facts, not at all open for discussion, it is interpreted as a suggestion or merely an opinion. I apologize for ranting about this, and to non-scientists it seems like a tempest in a tea cup. But I think it is symptomatic for a wrong attitude towards science and scientific methods in society that worries me. (And part of the blame certainly goes to the scientists.)

Last week I was helping my son with math. He was learning and practicing the "remarkable products", e.g. (a+b)(a-b) = a² - b². I noticed that he never applied the remarkable products in his exercises, but instead multiplied everything out. I asked him why he did that. The answer: "The religion teacher (!) had told them that scientific and mathematical theories are continuously revised and are, hence, unreliable." So, my son reasoned: the fact that these remarkable products worked yesterday is no guarantee that they will work today or tomorrow.

I know that this has nothing to do with bridge or bridge laws, but as a scientist this attitude drives me nuts.

gordontd, on 2015-February-03, 02:11, said:

Do you also object to the use of 2 IMPs as the standard amount for score adjustments in IMP pairs? Would you prefer that pairs are awarded +/- 2.12 IMPs (still an approximation) for unplayable boards?

I certainly don't object to a "verdict" from a bridge league (or committee) that for IMP pairs the adjustment should be +/- 2. After all, it is an arbitrary number and, essentially, a political decision.

However, if a claim is made that there is a mathematical relation with the 3 IMPs for a team of four match, this changes. Then, they should use the correct relation and I would prefer that they use the number of digits that they report the results in. If the Cross-IMPs are rounded to whole numbers, then use 2. If they are rounded to two digits behind the decimal point thaen use 2.12. If they report 8 digits behind the decimal point use 2.12132034.

Rik

[paulg]

Trinidad, on 2015-February-03, 08:08, said:

Last week I was helping my son with math. He was learning and practicing the "remarkable products", e.g. (a+b)(a-b) = a² - b². I noticed that he never applied the remarkable products in his exercises, but instead multiplied everything out. I asked him why he did that. The answer: "The religion teacher (!) had told them that scientific and mathematical theories are continuously revised and are, hence, unreliable." So, my son reasoned: the fact that these remarkable products worked yesterday is no guarantee that they will work today or tomorrow.

Religion always has an answer, whereas maths ... well, let's not get carried away now.

Wait until he gets to http://xkcd.com/179/.

[blackshoe]

Trinidad, on 2015-February-03, 08:08, said:

When traffic light turns red: STOP.

You are clearly not from around here.

[barmar]

Trinidad, on 2015-February-03, 08:08, said:

The point is that there is nothing "suggested" about the method. It is the method.

It's a method, but you can use some other method so long as it produces the same result.

If you need to divide large numbers, I suggest that you use a computer or calculator. But you can do long division by hand instead if you wish.

[mycroft]

Quote

Quote

Trinidad:
When traffic light turns red: STOP.

Blackshoe:
You are clearly not from around here.

Quote

Molson Canadian Montreal parody:

Maybe I can't turn right on a red light, but tabernac, I can go right through it.

Apologies for sensitive Quebecoise eyes :-)

[whereagles]

Trinidad, on 2015-February-03, 08:08, said:

Last week I was helping my son with math. He was learning and practicing the "remarkable products", e.g. (a+b)(a-b) = a² - b². I noticed that he never applied the remarkable products in his exercises, but instead multiplied everything out. I asked him why he did that. The answer: "The religion teacher (!) had told them that scientific and mathematical theories are continuously revised and are, hence, unreliable." So, my son reasoned: the fact that these remarkable products worked yesterday is no guarantee that they will work today or tomorrow.

Is this for real?????

[Vampyr]

whereagles, on 2015-February-03, 16:20, said:

Is this for real?????

And why is your son going to a school with a religion teacher?

[Trinidad]

Vampyr, on 2015-February-03, 18:18, said:

And why is your son going to a school with a religion teacher?

Because religion (or "life perspective" as the best literal translation would be) is a mandatory subject in the Dutch school system?

And you think that children should not learn about religion? You may not like religions or other people's views, but they are there. People of the world should know about them. Refusing to learn about other people's believe system is the fastest shortcut to extremism.

So, for me, there has never been any doubt that my children should learn about religions. I feel very fortunate that my kids have attended a Hindu Upanayanam, and that my son has spent a week in a Muslim family. Of course, they also get in touch with the Catholic values of the region where we live and I will tell them all about the Protestant perspective that I grew up with. And, in this all, there is no doubt that we tell our kids that there is zero reason to assume the existence of a god, and that message has gotten home loud and clear with both our kids.

What else do you think that two sensible kids who are exposed to a variety of believe systems would believe? That Mary was a virgin when she gave birth to Jesus?!? Or that they might reincarnate as a fire fly? Do you have so little confidence in your own atheism that you would shield your kids from the reality of the existence of religions in this world (and at the same time deny them access to all the valuable things religions give)?

So, both our kids have chosen not to believe in the existence of a god. Or, more accurately, they consider the whole question utterly irrelevant, which is exactly what it is. But they will be able to build bridges across religions and cultures, with an understanding for other views and without passing judgement. How about you?

Rik

[Trinidad]

whereagles, on 2015-February-03, 16:20, said:

Is this for real?????

Yes. (My son has a dominant religion teacher and has had a weak math teacher who, fortunately, got replaced recently by one tough, but very nice, lady. )

But I set him straight fairly easily: I just did the experiment with him. I told him that scientists - in sharp contrast to religious teachers - are capable of predicting the future. Of course, he didn't believe me.

So, I told him to punch in 98*102 on his calculator and predicted that when he hit the "=" button the display would show 9996. He was puzzled when I was correct. How could I know that, just like that? I explained to him that 98*102 = (100-2)(100+2)=100²-2² = 10000-4 = 9996. We repeated this for 17*23=391 and a couple more problems, and he saw that I could do this arithmetic without a calculator or pen and paper. When he saw the confidence with which I applied these mathematical principles in arithmetic he got the message. I told him that facts like this made it possible for engineers to build cars and airplanes that they successfully predict will work and houses and bridges that they predict will be strong enough. And that all this is verifiable.

And then I asked him: What verifiable fact did your religion teacher ever predict?

His conclusion was simple and correct: The religion teacher should stick to what he knows and not stick his nose into science.

Q.E.D.

Rik

[Vampyr]

Trinidad, on 2015-February-03, 19:03, said:

So, for me, there has never been any doubt that my children should learn about religions.

Learning about religions is fine. Teaching the children a belief system is not. Your son took the teacher's statement as true, not as something "some people believe". He was not being taught about religions.

And I will appreciate it if you will not tell me what I thnk.

[campboy]

At the risk of derailing the thread by answering the original question...

The justification for the 70% rule (which, like the pirate code, is really more of a guideline) is as follows. First, the EBU VP scales are (or were – not sure if this is still true) designed so that in matches between equal-strength teams of average ability, each VP result will be approximately equally likely. To maintain that property when applying to other forms of scoring, the scaling factor we should use is the quotient of the standard deviations -- that is, the standard deviation of the total crossimp of one pair from n tables of equal-strength pairs, divided by the standard deviation of the IMP score between two equal strength teams. Different boards are independent if we assume the teams/pairs are indistinguishable, so we can do everything board-by-board and get the same scaling factor.

Now, assume that on a given board these indistinguishable average pairs will produce scores independently with some distribution, with mean m and variance s². It is easy to check that for a teams match the variance of the swing in total points is 2s², and that the variance of the total of a pair's swings against all the other pairs is n(n–1)s². So, assuming the IMP scale is linear, you should divide by the total number of scores (n) times sqrt((n–1)/2n). This last figure will be a little less than sqrt(0.5), and 70% is indeed a little less than sqrt(0.5).

Now of course the IMP scale is not actually linear. But this means that the problem is no longer exactly soluble – the scaling factor will depend on the unknown distribution above, and will be different for every board. The overall average scaling factor will be something close to sqrt(0.5), will depend on n, will probably be less than sqrt(0.5) for smallish n and there is simply no way of getting enough information on the distribution of swings between average evenly-matched teams to give it any more precisely than that.

So I for one do not object to the WB giving a figure of 70%. It is an approximation to an exact solution to an approximation of the problem. Since we can't do anything about the second "approximation" in that sentence, it is silly to nit-pick about the first.

Likewise it does not bother me that average-plus is 2 IMPs rather than 2.12. The figure of 3 IMPs for teams of four is rounded to the nearest integer in the first place, and it is pointless to take something rounded to an integer, multiply it by sqrt(0.5), and then give 2 decimal places of the answer as if they mean something.

This post has been edited by campboy: 2015-February-04, 06:21

[Trinidad]

campboy, on 2015-February-04, 06:14, said:

The figure of 3 IMPs for teams of four is rounded to the nearest integer in the first place, and it is pointless to take something rounded to an integer, multiply it by sqrt(0.5), and then give 2 decimal places of the answer as if they mean something.

While I agree with the rest of your post, this part is incorrect. The figure of 3 IMPs for teams of four is not rounded to a nearest integer. It is defined by the laws (Law 86A). That means that the value of 3 IMPs has absolute accuracy and anything derived from it does not carry any inaccuracies from the value of 3 IMP. So, reporting 2.12 IMP for Cross-IMPs is not silly at all.

The fact that this 3 IMP definition was conjured up a long time ago because people guessed that it would be approximately equivalent to 60% in MP pairs (and some other value might be more accurately resembling 60% at MP pairs) is irrelevant. Once "3 IMPs" was written into the law book, the value of 3 IMPs is a given by definition (until someone decides to change it).

Rik

[whereagles]

Trinidad, on 2015-February-03, 19:35, said:

1. And then I asked him: What verifiable fact did your religion teacher ever predict?

2. My son has a dominant religion teacher and has had a weak math teacher who, fortunately, got replaced recently by one tough, but very nice, lady.

1. Touché. That is indeed the crux of the matter. Philosophers are right that both science and religion can be seen as explanations for the world, and that one is free to decide which to believe. However, there is one very fundamental difference between the two: science has predictive power. Only the most devote niilist can deny that - but even he cannot convince the rest of the world. Only himself.. lol

2. That sort of woman gets me hot. You happen to have her number...?

[whereagles]

campboy, on 2015-February-04, 06:14, said:

At the risk of derailing the thread by answering the original question...

Thx a lot. I'll have a closer look at your reply later.