[sfi]

shyams, on 2016-July-17, 05:36, said:

Mods, pretty please?

You can make it stop by yourself by simply not clicking on it. Just because xkcd has a cartoon about it doesn't mean it's a categorical imperative.

[masse24]

Spisu, on 2016-July-15, 11:25, said:

I am familiar with . . . Ignorance.

FYP

[Spisu]

masse24, on 2016-July-17, 06:27, said:

FYP

The actual quote you deceitfully alter is: "I am familiar with Bayes and the restricted choice folks' claim it is based on the Bayes postulate known also by some as the "Equidistribution of Ignorance."

Factual information to explain any related point would be welcome, however.

[Spisu]

Zelandakh, on 2016-July-17, 06:03, said:

What I know is that you quoted me but failed to answer my question. So here it is again: are you saying that RC gives the correct result in all cases?

I have said repeatedly, I am interested in the underlying premise, not any and every case, because (as surely some informed here must know), there are cases attributed to RC where a choice between equals CAN make a difference. (Some opening leads for example or possibly other cases where an actual difference can exist between options and/or their consequences.) There is a massive chasm between an actual difference and a pretended and fallacious difference...You see, if you knew the K was in an E's hand and E later won a double finesse for K-Q with that K, then there is a mathematical basis that the Q might have been played if there. But there would be then an actual difference between the K and Q by your prior knowledge.

But I did show you an example where the premise of RC breaks down.

[Stephen Tu]

Spisu, on 2016-July-17, 05:39, said:

Please come back if you have something to add. BTW, did you see the challenge on the ACBL Encyclopedia Example 1 under Restricted Choice? Explaining why the numbers don't seem to add up would be a big "something".

I don't have the same edition of the encyclopedia but assume it is probably the same. Are you asking about the table with the QJ9 opposite xxx combination how to play for 1 trick? The table omits both AKT onside (because any play leading toward the honors works), and AKT offside (because nothing works), and these two scenarios cancel out and there are no wrong moves on those combos as stated in the text above the table. The table is just showing that playing the Q or J on the second round is better than finessing the 9 which I hope everyone believes, and this is because with AK offside, 4th hand can randomize on the first round, while with KT, 4th hand can only win the K. The numbers in the table seem to add up to me. They show 6 layouts, each holding 2 out of the 3 relevant cards (A, K, 10), each is dealt 100 times. Adds up to me. KT offside and AT offside are combined twice as likely as AK (without the T), so by playing high honor 2nd round you win 200 times on those combos and lose to AK 100 times, counting both times when east won first trick with K and when east won first trick with ace.

Basically
AT K 100 cases, east wins K first, anything you do works 2nd round
KT A 100 cases, east wins A first, anything works
A KT 100, east wins K first, must play high 2nd round
K AT 100, east wins A first, must play high 2nd round
T AK 50 east wins A first + 50 east wins K first.

So on the last 3 cases where what you choose matters, you either count both the times where east wins K and east wins ace first round, and get 200 vs. 100 for high honor vs. finesse 9 2nd round. Or you count only the times for say specifically east wins the A first, and you count 100 vs. 50. Either reasoning is correct.

If it is not that table, please specify more completely, perhaps cut/paste the table. And specify which numbers in the table you think are incorrect and explain why.

And answer my previous post about the AJT9x vs. xxxx combination.

[Stephen Tu]

Spisu, on 2016-July-17, 09:13, said:

But I did show you an example where the premise of RC breaks down.

You did? Could you repeat it more clearly?

[Zelandakh]

Spisu, on 2016-July-17, 09:13, said:

But I did show you an example where the premise of RC breaks down.

Did you? Again, my bad English - could you please repeat it. Be specific with the precise card combination, the (false) odds according to RC and the correct odds.

[1eyedjack]

Spisu, on 2016-July-17, 05:39, said:

Please come back if you have something to add. BTW, did you see the challenge on the ACBL Encyclopedia Example 1 under Restricted Choice? Explaining why the numbers don't seem to add up would be a big "something".

I saw your challenge. I do not have access to that work and it is not available free to download as far as I can tell. I will let someone else respond if they feel inclined. I would feel more sympathetic to the attempt if the challenges put to you were met. Incidentally, as I recall from the last time I had access to the Encyclopedia some decades ago it was not immune to error itself, in some of the card combinations in particular. I am not saying it is wrong in this case - I have no idea on that. If I had to guess in a vacuum I would expect it to be correct but what the hey. There is no shortage of other reference works on PRC, and I tend to trust specialised academic works more than encyclopedia generally.

As for having something to add, it is not so complex a subject that there is an infinite scope for adding to the case.

If you think that I am being unfair, I think that you should accept that where you attempt to assert the fallacy of a principle, that has been peer reviewed by experts in the field, and has been accepted without question for decades, then the bar is set quite high and you have an onerous burden of proof to overcome. If that means that you have to go the extra mile, where your opponents don't feel that pressure, than that is the reason.

[EDIT] I see that Stephen Tu has taken up the cudgels on this one. His pedigree is better than mine anyway. Personally, I respect his patience.

[nige1]

gordontd, on 2016-July-16, 11:51, said:

Does anyone remember Reef Fish, I think from rgb?

Yes. He was fun!

[Spisu]

1eyedjack, on 2016-July-17, 11:22, said:

I saw your challenge. I do not have access to that work and it is not available free to download as far as I can tell. I will let someone else respond if they feel inclined. I would feel more sympathetic to the attempt if the challenges put to you were met. Incidentally, as I recall from the last time I had access to the Encyclopedia some decades ago it was not immune to error itself, in some of the card combinations in particular. I am not saying it is wrong in this case - I have no idea on that. If I had to guess in a vacuum I would expect it to be correct but what the hey. There is no shortage of other reference works on PRC, and I tend to trust specialised academic works more than encyclopedia generally.

As for having something to add, it is not so complex a subject that there is an infinite scope for adding to the case.

If you think that I am being unfair, I think that you should accept that where you attempt to assert the fallacy of a principle, that has been peer reviewed by experts in the field, and has been accepted without question for decades, then the bar is set quite high and you have an onerous burden of proof to overcome. If that means that you have to go the extra mile, where your opponents don't feel that pressure, than that is the reason.

[EDIT] I see that Stephen Tu has taken up the cudgels on this one. His pedigree is better than mine anyway. Personally, I respect his patience.

I greatly appreciate the moderate tone. I may see things a bit differently because of my background in every science, math, logic, Martin Gardner style skepticism (and bridge). So I was shocked at the reaction from what I expected to be reflective and open minded experts. I sought data here actually to see if I could possibly be wrong. But that hasn't happened. (Quite the contrary.) So the next step could be an element that may really surprise you, maybe even give you pause.

[Spisu]

Zelandakh, on 2016-July-17, 09:53, said:

Did you? Again, my bad English - could you please repeat it. Be specific with the precise card combination, the (false) odds according to RC and the correct odds.

You're just trolling now. Give it up.

[Stephen Tu]

How is he trolling? He, and the rest of us, are looking for a specific card combination where RC goes wrong, what numbers you attribute to the various layouts, and how you think this differs vs RC. All you have provided is hand waving verbiage, no concrete numbers.

How about answering my post above with the 4 questions about the ajtxx vs xxxx combo. Or post your own combo, with actual numbers, and say how you think rc comes up with a different answer. Otherwise all your arguments amount to pointless ranting, because you aren't giving any numbers to dispute, nor are you disputing our numbers in the posts presenting specific examples.

[1eyedjack]

Spisu, on 2016-July-17, 14:08, said:

I greatly appreciate the moderate tone. I may see things a bit differently because of my background in every science, math, logic, Martin Gardner style skepticism (and bridge). So I was shocked at the reaction from what I expected to be reflective and open minded experts. I sought data here actually to see if I could possibly be wrong. But that hasn't happened. (Quite the contrary.) So the next step could be an element that may really surprise you, maybe even give you pause.

Your cause is not helped by the prevalence of fly-by-night operators of no recognised provenance who periodically pop up in forums making apparently outrageous claims in the face of settled wisdom. Against that backdrop it is predictable that you would be compartmentalised in that class as an initial reaction. To overcome that you either need to establish your credentials as an expert in the field or set out your case with absolute clarity, and then on top of that be prepared to demolish the counter arguments that you expect to arise. Preferably all 3 of the above.

On the first point, no-one knows who you are

On the second point, despite all the repetition I would say that the clarity of your case is wanting. At least as far as I am concerned.

On the third point, well this is still progressing.

So I guess it is still a case of watch this space.

[Spisu]

Stephen Tu, on 2016-July-17, 09:27, said:

I don't have the same edition of the encyclopedia but assume it is probably the same. Are you asking about the table with the QJ9 opposite xxx combination how to play for 1 trick? The table omits both AKT onside (because any play leading toward the honors works), and AKT offside (because nothing works), and these two scenarios cancel out and there are no wrong moves on those combos as stated in the text above the table. The table is just showing that playing the Q or J on the second round is better than finessing the 9 which I hope everyone believes, and this is because with AK offside, 4th hand can randomize on the first round, while with KT, 4th hand can only win the K. The numbers in the table seem to add up to me. They show 6 layouts, each holding 2 out of the 3 relevant cards (A, K, 10), each is dealt 100 times. Adds up to me. KT offside and AT offside are combined twice as likely as AK (without the T), so by playing high honor 2nd round you win 200 times on those combos and lose to AK 100 times, counting both times when east won first trick with K and when east won first trick with ace.

Basically
AT K 100 cases, east wins K first, anything you do works 2nd round
KT A 100 cases, east wins A first, anything works
A KT 100, east wins K first, must play high 2nd round
K AT 100, east wins A first, must play high 2nd round
T AK 50 east wins A first + 50 east wins K first.

So on the last 3 cases where what you choose matters, you either count both the times where east wins K and east wins ace first round, and get 200 vs. 100 for high honor vs. finesse 9 2nd round. Or you count only the times for say specifically east wins the A first, and you count 100 vs. 50. Either reasoning is correct.

If it is not that table, please specify more completely, perhaps cut/paste the table. And specify which numbers in the table you think are incorrect and explain why.

And answer my previous post about the AJT9x vs. xxxx combination.

Thank you. I appreciate your going there and checking it out, but I believe there is a factor there you didn't pick up on. I had planned to bring it up tomorrow and will do so.

The other question brings in a nine card suit and a distributional issue I've avoided to concentrate on the principle without that side issue. I always finesse twice because the strategy I follow focusing on honors has me winning about 66% of the cases. That strategy is that isolating East's hand sets one up with a priori divided honors 50% of the time compared to East's a priori 25% to hold both honors. And, others to the contrary, those odds do not change when events consistent with the strategy occur...Which is all events in roughly 75% of the deals. Note that an East's winning with an expected honor that turns out to be a king (in the absence of special information) does not change divided honors' odds unless you decide to assert "He won with THE KING!", now what do I do?!

[Spisu]

Stephen Tu, on 2016-July-17, 15:35, said:

How is he trolling? He, and the rest of us, are looking for a specific card combination where RC goes wrong, what numbers you attribute to the various layouts, and how you think this differs vs RC. All you have provided is hand waving verbiage, no concrete numbers.

How about answering my post above with the 4 questions about the ajtxx vs xxxx combo. Or post your own combo, with actual numbers, and say how you think rc comes up with a different answer. Otherwise all your arguments amount to pointless ranting, because you aren't giving any numbers to dispute, nor are you disputing our numbers in the posts presenting specific examples.

There is no case where I said RC gave false or bad odds. I will post a $100 reward to the BBO account of the person who finds and cites where I said those words.

If something is based on a fallacy, that doesn't mean it can't come up with the right answer and even incorporate other compensating fallacies or errors, particularly when the answer is already known. Maybe there is an error but I do not know offhand of a case where in its common uses RC doesn't come up the correct advice. (Now, outside of normal play, I have seen experts criticized for not applying RC to some inane spot cards where the card was not a prior interest. THAT would be an error to me, but it's not "bad odds".)

[Spisu]

I'm afraid I've had it with the lies, insults, and deceit, and worst of all, bizarre and arrogant ignorance in nearly every field from a group I expected to be intelligent and open minded. I have had to explain things as crazy as being attacked for questioning RC by people with no clue as to the actual basis for RC...And accused of saying things I did not say and believing absurdities.

It's obvious there will be no use to go forward. I haven't encountered a single open mind. I can't imagine anyone here interested in truths outside his/her dogmas or who could even recognize the truth. If there was any interested person here I'm letting down, I wish you had made it known, and I apologize.

[Stephen Tu]

You haven't encountered a single open mind, because your arguments are totally unconvincing. Apparently, you concede that RC gives absolutely correct advice on how to play suits and comes up with the exact same odds as you do. And your only objection apparently is your philosophy that people should treat equal honors as unidentifiable equivalent quacks and not see that an opponent specifically won with say the king. So you think only option 1 (count all cases, either equal honor winning) of how to explain strategy is correct, and option 2 (count only 1 particular honor winning, but divide the combined cases by the number of equal honors) is fallacious. But really there is nothing wrong with option 2. People are allowed to see what card the opponent won with, and see that it is a king. There are a certain set of people who don't grasp best strategy immediately who need it explained to them why the odds are not 50:50 when the previous plays have winnowed down the possibilities into only 2 positions of originally equal frequency. They have to be explained why, and it's because the opponent has choice of plays in one case but not the other. And best play against people in general is not the same as against a hypothetical opponent who has a kibitzer standing behind him with a pistol who will shoot him in the head if he doesn't always play the highest of equals first round. So the odds most definitely arise from the opponent having choice of plays. The overall success rate of two finesses is fixed, but the success rate when a particular identifiable honor won the first trick most definitely varies with opponent tendencies.

So RC explained in an alternate manner gives all the right answers and numbers, but it's a fallacy. OK believe whatever you want dude. Clearly you aren't going to convince anyone here. I mean basically from what I can see you are claiming it's a fallacy that people can look and see which honor won the first trick, and that it's a fallacy that an opponent having a choice of what to play from equals matters. That's ridiculous, which is why you haven't found any open minds.

[1eyedjack]

If they did not have open minds they would not have spent 5 pages on it. Your definition of a closed mind is someone who does not accept your case

[1eyedjack]

RC, as it was taught to me, stated that it was irrelevant whether or not you took specific note of the particular "equal" card played, RC comes up with the same answer.

I attempt to demonstrate this in another thread here

http://tinyurl.com/hjps5mv

To which I have linked earlier.

Hats off to Stephen Tu if he has successfully distilled the nature of your case, because despite repetition it was not clear to me, nor to Zel it seems, and we stand accused of insults to your use of language because of that.

From a philosophical viewpoint I have a lot of difficulty accepting that a principle which always returns the correct mathematical odds (which I gather you accept to be the case) can ever be a "fallacy", just because it expresses a manner of arriving at the solution that differs from some other, perhaps equally valid, mathematical technique.

[1eyedjack]

I wonder whether it might be profitable to define PRC here. 5 pages into the discussion is rather late in the day for that, and we are all rather jaded, but if the differences of opinion are simply as a result of differences in an understanding of the definition of the principle, then that would go a long way to resolving it.

As I understand it, the principle simply states that

p2 = f(p1)

where

p2 is the probability that RHO (last to follow) has the missing (key) card, having earlier (necessarily) followed with one or more equal cards

p1 is the probability that RHO (last to follow) would have had the missing (key) card had he earlier followed with an insignificantly small card.

f is an algorithm that derives p2 from p1.

The algorithm f requires that you divide p1 by the factorial of the number of potentially equal cards in the initial state. In most illustrations that number is 2, but it could be more.

The main practical benefit of the principle to bridge players is that an experienced bridge player will have memorised a table of the most common values for p1, not having the time at the table to regenerate a load of nCr values, but the application of this function at the table is well within their capabilities.

Say North holds AK8x and South has Qxx. You cash the Ace and Q, and lead up to K8. All have followed so far, West with low cards and East with 2 of JT9.

Here there would be three key cards (JT9). 3! = 6.
Absent any other information, p1 = 11/21, so p2 = 11/126. The finesse is very much more heavily favourite than, say had North held KTxxx opposite A9xx and East had played one of just 2 key cards earlier.

Have I successfully set out the principle? I don't need only Spisu to respond but his view would certainly be of value.