Evidence, Popular Opinion, and Prejudice

[PairTheBoard]

Sklansky has made the following observation:

[ QUOTE ]
DS -
The reason I say that the fact that 90% of people disagreeing with a theory is evidence against it, is because of all theories ever proposed that were eventually settled, those that had 90% disagreeing, turned out to be right much less often than those which had few disagreeing.


[/ QUOTE ]

He has claimed that besides me, "Everyone Else Understands" this. So I ask SMP to take a closer look at it. Exactly what is it saying and how useful is it?

If by Sklansky's observation Poplular Opinion is indeed "Evidence", what kind of Evidence is it? Is it mathematical evidence? Is it scientific evidence? Is it aesthetic evidence? Is it evidence that would be admitted in a court of law? Is it common sense evidence?

How persuasive is this Evidence? Is it conclusive evidence? Is it strong evidence? Is it weak evidence? Is it reliable evidence? Is it misleading evidence? Is it Prejudicial evidence?

First, does Sklansky's observation imply Popular Opinion is Mathematical Evidence? In the past Sklansky has made allusions to mathematical support for the implication, sometimes with statements like, "Math tells us". Certainly, the observation is analogous to ones we can make in a mathematical probability or statistics setting. When in that setting it's the kind of observation provided by Bayes Theorem or just Conditional Probability in general. But this is an example of Sophomoric application of math concepts.

People with real in depth knowledge of mathematics tell us that the real problem is in coming up with the mathematical model. When dealing with a Real situation, the assumptions that go into constructing the math model are assumptions made outside of mathematics. I think Sklansky has recently become convinced of this. His observation does not carry the Authority of Mathematics. The Evidence it implies is not Math Evidence.

Does Sklanky's observation imply Scientific Evidence? Not strictly speaking. At least not in general. In a specific instance where the Proposition itself refers to Popular Opinion about it, then empirical data about that Popular Opinion would provide Scientific Evidence about the Proposition. Certain Aesthetic Propositions might fall into this category. But generally we are not talking about such a Proposition. Science needs to apply the scientific method to a Proposition to obtain evidence for it. Science does not look at Popular Opinion about it. Thats' a good thing too, because otherwise we might still be operating under the laws of physics provided us by Aristotle.

Does Sklanky's Observation provide Evidence admissable in a Court of Law? Almost but not quite. And at this point I bring in Sklanky's additional Observation that the Evidence can be improved if we limit our Population of Opinion to that of "brilliant" people. Certainly a court of law would not find admissable the fact that Most People think the defendant is guilty. Neither would the court admit the fact that most "brilliant" people think the defendant is guilty.

However, the court will hear testimony from expert witnesses. And at this level Sklansky's Observation might be allowed as Evidence for what constitutes good science according to the consensus of the scientific community. But that only provides the court with evidence of credibility for the expert witness. That only allows the expert witness to speak. Notice also, the Court would not allow the witness to speak based only on his "brilliance". He must have in depth knowledge of the specific area on which he is to testify.

The Evidence the Court is interested in is, what does the expert-identified scientific evidence say about the Proposition at Bar? The Court will weigh the evidence in making a determination on the Proposition. The Court will decide how the evidence relates to the Proposition. It will not allow itself to be Prejudiced by the opinions others have about the Proposition. The expert witness does not give his opinion about the guilt of the defendant.

So the implication Sklansky claims for his Observation is not Mathematical, Scientific, or Legal. Is it common sense evidence? Well, it is certainly a Natural inferrence for us to make. If we know practically nothing about the Proposition other than the fact that Most People agree with it, we Naturally tend to go along with the crowd and accept it. We are Naturally even more swayed by the consensus of opinion among people who have an in depth knowledge the Proposition's subject.

In subjects beyond our ability to study we have little choice but to accept the consensus of those who are able to study it. And we would naturally give more credence to those who are recognized as Brilliant in that field of study. So this might pass as a common sense way to form our opinions in these situations. However, just at the Court does not allow just any "Brilliant" person to give expert testimony, it does not make common sense for us to give extraordinary weight to the opinon of "Brilliant" people who are speaking outside their area of expertise. Especially if they tend to be Myopically Focused on concepts applicable to their area rather than ones relevant to the Proposition.

We should also recognize that this common sense evidence is not the Best Evidence. It is often unreliable, which is why it is not allowed in a Court of Law. The fact that most people think O.J. did it sways our opinion as weak evidence. The fact that most Legal Experts think he did it sways our opinion as stronger evidence. But the Best Evidence is that which was presented in Court. Those of us who were really interested in the question wanted to see the actual evidence and weigh it ourselves. It does not make common sense to depend on weak evidence when Better evidence is available.

Furthmore, what sometimes passes as common sense often turns out to be anything but good sense. Sklansky's inference can easily degrade into plain everyday Prejudice and bias. Popular opinion can quickly turn into Bigotry, even among the most Brilliant people, depending on the times. The Best Evidence is sometimes our individual conscience.

And when it comes to questions like, "Does God Exist"? Who is an expert? What does the question even mean? How do you decide what the Best Evidence is for questions like this? Nobody, no matter how brilliant, has any kind of track record of accuracy for determining such questions because there is no way making such a determination. Each individual relates to the question his own way. It is ultimately a personal question requiring a personal resolution.

PairTheBoard

[David Sklansky]

Are you implying that something you wrote refutes my quote?

Just to make it clear, let me refer you to my official definition of "evidence". To wit:

Anything that forces you to move the line.

That includes popular opinion. But it doesn't mean that there there doesn't often exists much stronger evidence for the other side. Surely you didn't think that I thought otherwise.

[kerowo]

I think that is an iffy statement because almost by definition every paridigm change has started off with 90% of the population disagreeing with it.

[PairTheBoard]

[ QUOTE ]
Are you implying that something you wrote refutes my quote?


[/ QUOTE ]

My post was an attempt to clarify the notion you are promoting. You claimed everybody but me understands what you are saying. I'm actually applying your notion to that claim in the one instance where it best applies. Does everybody here understand what you're saying? Let's find out.

[ QUOTE ]
Just to make it clear, let me refer you to my official definition of "evidence". To wit:

Anything that forces you to move the line.


[/ QUOTE ]

"Move the line" is an interesting phrase. As you saw in your Thread on "Belief", not everybody comes to a conviction about the validity of a proposition in "probability" terms. "Move the line" is a reference to "probability" terms.

As I said in my post, it's a Natural Inference to make in the absence of other information. However, it easily Degrades into something far from common sense. It is unreliable. In a court of law it is inadmissable. It's not science and it's not mathematics. That is not a refutation of the fact that it's something we naturally do. But it indicates we need to keep a watchful eye on how we apply it.

Just because it might "move the line" in the absence of better evidence does not mean its Placement of the "Line" in that situation should be the starting point for determining the "Line" once Better Evidence is Available. Once Better information is available we should take an UNBIASED look at it. I think this position IS contrary to your use of the natural inference.

PairTheBoard

[LA_Price]

I would say that whether the wisdom of the crowd should be considered evidence depends entirely on what the proposition is.

If the proposition lives in the world of mediocristan randomness(i.e. normal distriubtion) then I would have to agree with Sklansky. This would be a question like "there are 70 beans in this jar". You could ask a crowd how many beans were in the jar and they would get it reasonbly right and there observations should be counted as evidence to prove or disprove your proposition.

If the proposition lives in world of extremistan("black swan", unpredictable,large impact of highly improbable) I would say no, the crowd can easily be wrong. In this type of world there is also an expert problem, in that there is really no expertise. This would be a proposition like "all swans are white". 90% of the people can think this, and a single observation or outlier will render the proposition false.

This is exactly the topic discussed in Nassim Taleb's new book "The Black Swan".

[Piers]

[ QUOTE ]
The reason I say that the fact that 90% of people disagreeing with a theory is evidence against it, is because of all theories ever proposed that were eventually settled, those that had 90% disagreeing, turned out to be right much less often than those which had few disagreeing.

[/ QUOTE ]

Well its kind of ok but a lot depends on perspective.

For a random theory about a subject you know nothing I think it is fine.

However once you understand why say 90% of people disagree with a theory, you can discount the fact that they disagree and use as evidence or otherwise the reason they disagree.

So as a general statement about a random, and yet unspecified theory I think the quoted statement is fine. However once you have specified the theory in question the statement need not be true in general.

[PairTheBoard]

So here's a question. In the absence of Better information we naturally look to Popular Opinion. But suppose we start out with the Better Information. Should we then Bring In Popular Opinion to Prejudice our analysis of the Better Information? This is what I see Sklansky doing.

PairTheBoard

[LA_Price]

In some situations better information doesn't exist and popular opinion doesn't matter.

Imagine you're a turkey in a group of turkeys. Every day the humans come out and give you food and watch you get fat. You have no reason to believe humans are anything but great and friendly. 90%+ of turkey's would think the same thing and keep eating the food the humans give them...and then Thanksgiving comes and you all lose your heads. Prior observation and opinion was rendered useless in a single day.

A proposition like "Does God Exist" lives in the world of extremistan, like the turkeys. We will live our days until the day we die believing one way or the other and then we will either be sent to some heaven or hell(who knows which religion is right) or we will just die and will never know we were right or wrong. 90% of the people believing either way is useless as evidence.

[bunny]

I think it is inferential evidence. As he has said before, more theories-that-have-been-doubted-by-90%-of-smart-people have turned out to be false than turned out to be true.

Thus, we infer that any current theory-that-is-doubted-by-90%-of-smart-people is more likely to turn out to be false than true.

I dont consider it very strong, but it does seem to be evidence of some sort.

[vhawk01]

[ QUOTE ]
In some situations better information doesn't exist and popular opinion doesn't matter.

Imagine you're a turkey in a group of turkeys. Every day the humans come out and give you food and watch you get fat. You have no reason to believe humans are anything but great and friendly. 90%+ of turkey's would think the same thing and keep eating the food the humans give them...and then Thanksgiving comes and you all lose your heads. Prior observation and opinion was rendered useless in a single day.

A proposition like "Does God Exist" lives in the world of extremistan, like the turkeys. We will live our days until the day we die believing one way or the other and then we will either be sent to some heaven or hell(who knows which religion is right) or we will just die and will never know we were right or wrong. 90% of the people believing either way is useless as evidence.

[/ QUOTE ]

Awfully results-oriented, don't you think?

[PairTheBoard]

[ QUOTE ]
[ QUOTE ]
In some situations better information doesn't exist and popular opinion doesn't matter.

Imagine you're a turkey in a group of turkeys. Every day the humans come out and give you food and watch you get fat. You have no reason to believe humans are anything but great and friendly. 90%+ of turkey's would think the same thing and keep eating the food the humans give them...and then Thanksgiving comes and you all lose your heads. Prior observation and opinion was rendered useless in a single day.

A proposition like "Does God Exist" lives in the world of extremistan, like the turkeys. We will live our days until the day we die believing one way or the other and then we will either be sent to some heaven or hell(who knows which religion is right) or we will just die and will never know we were right or wrong. 90% of the people believing either way is useless as evidence.

[/ QUOTE ]

Awfully results-oriented, don't you think?

[/ QUOTE ]

I'm suprised vhawk. I thought it was a good representation of your weak-atheist view that legitimate evidence is not available so No Belief makes the most sense.

PairTheBoard

[LA_Price]

expand please. I don't get what part you find results oriented.

[vhawk01]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
In some situations better information doesn't exist and popular opinion doesn't matter.

Imagine you're a turkey in a group of turkeys. Every day the humans come out and give you food and watch you get fat. You have no reason to believe humans are anything but great and friendly. 90%+ of turkey's would think the same thing and keep eating the food the humans give them...and then Thanksgiving comes and you all lose your heads. Prior observation and opinion was rendered useless in a single day.

A proposition like "Does God Exist" lives in the world of extremistan, like the turkeys. We will live our days until the day we die believing one way or the other and then we will either be sent to some heaven or hell(who knows which religion is right) or we will just die and will never know we were right or wrong. 90% of the people believing either way is useless as evidence.

[/ QUOTE ]

Awfully results-oriented, don't you think?

[/ QUOTE ]

I'm suprised vhawk. I thought it was a good representation of your weak-atheist view that legitimate evidence is not available so No Belief makes the most sense.

PairTheBoard

[/ QUOTE ]

My belief in this analogy, I think, if I'm understanding it correctly, is the same as the stupid 90% of the turkeys, i.e. that the humans are benevolent. There is no reason to think otherwise, thus, the standard position is they are good. I think you have my position confused with those who would actually say things like "I make no claims about my own knowledge or beliefs" or "I have no belief on the status of God." I have a belief on the status of God, and that is one of disbelief. I don't think God is impossible, I just don't believe in him. In this turkey example, they don't think its impossible the humans are out to eat them, they just have no reason to think it. Thus, the best guess is that the humans are benevolent.

It is entirely possible I am misunderstanding the analogy, though.

[vhawk01]

[ QUOTE ]
expand please. I don't get what part you find results oriented.

[/ QUOTE ]

Because I think the turkeys are making the correct choice in believing the humans are benevolent, even though it turns out they are wrong.

[LA_Price]

This would be true if the turkey's lived in the world of mediocristan and the argument that the humans were benevolent on the 300 days or whatever prior wasn't vastly outweighed by the final verdict of their death on a single day. Results oriented analysis is wrong in the normal "bell curve" shaped world but when the impact of a single observation disproves all others it is not. Take the "all swans are white example". Europeans in the 1700's would have made this statement. Then they discovered Austrailia and black swans. They were wrong in their statement but their is nothing they could do, because the existence of a black swan was essentially unpredictable. Their failure was in making a statement that existed in the extremistan(unpredictable world), but then behaving like it existed in mediocristan. Taking individual people's observations on an exstremistan proposition and treating them as "evidence" as if the world is mediocristan would be this same mistake.

[vhawk01]

[ QUOTE ]
This would be true if the turkey's lived in the world of mediocristan and the argument that the humans were benevolent on the 300 days or whatever prior wasn't vastly outweighed by the final verdict of their death on a single day. Results oriented analysis is wrong in the normal "bell curve" shaped world but when the impact of a single observation disproves all others it is not. Take the "all swans are white example". Europeans in the 1700's would have made this statement. Then they discovered Austrailia and black swans. They were wrong in their statement but their is nothing they could do, because the existence of a black swan was essentially unpredictable. Their failure was in making a statement that existed in the extremistan(unpredictable world), but then behaving like it existed in mediocristan. Taking individual people's observations on an exstremistan proposition and treating them as "evidence" as if the world is mediocristan would be this same mistake.

[/ QUOTE ]

Its a bell-curve world because we are talking about the general approach, not the specific outcome. Yep, the turkeys are wrong about the humans, and the Europeans were wrong about the swans, but these are the outliers, and most of the time, this approach is correct. Thats what I meant by results-oriented.

[LA_Price]

I think you're still misunderstanding. It is not always a bell shaped world. It is only a bell curve shaped world if one observation does not outweigh all the others. Take the 1000 random people and put them in a room. Now put the world's tallest man in. The average height will not change much. Now send in bill gates and see what happens to the average wealth.

Applying some "General bell shaped curve approach" to things that do not correspond to the bell curve is a big error. It isn't ok because it "works most of time". The point is that when you make a decision or want to test proposition that is extremistan in nature you can not use the bell curve because it will cause you to ignore the outliers and their huge impact.

P.S. I suggest getting taleb's book as he explains it much better than I.

[vhawk01]

[ QUOTE ]
I think you're still misunderstanding. It is not always a bell shaped world. It is only a bell curve shaped world if one observation does not outweigh all the others. Take the 1000 random people and put them in a room. Now put the world's tallest man in. The average height will not change much. Now send in bill gates and see what happens to the average wealth.

Applying some "General bell shaped curve approach" to things that do not correspond to the bell curve is a big error. It isn't ok because it "works most of time". The point is that when you make a decision or want to test proposition that is extremistan in nature you can not use the bell curve because it will cause you to ignore the outliers and their huge impact.

P.S. I suggest getting taleb's book as he explains it much better than I.

[/ QUOTE ]

Ok. I guess I'm still not following. I agree with you on the "not always a bell curve" part, I just don't see why the turkeys are making a mistake. Sure, the final result, that they die, is a huge result, and blows the previous data out of the water. But they made their decision BEFORE that, right?

Or are you saying that we should take the opinion of the hypothetical turkey that just got killed over the opinions of the 90% that think humans are awesome, even though he is the minority and the mean opinion is still "humans are nice?" If this is what you are saying, I have definitely been misunderstanding you this entire time, and I certainly agree. Although, I'm not sure I see the parallels to atheism/agnosticism. However, if you can clear this up, I'm positive you can explain the rest of the analogy to me. It seems like I just misunderstood the whole time.

[Phil153]

You mention courts. The irony is, the jury system is set up on exactly this premise - the more people that agree with a position on a question of fact, the more likely it is to be correct.

However on difficult technical questions (including evolution), popular opinion is irrelevant, as the basis for those opinions is not based on an understanding of the evidence or lack of, but merely the cherry picking of bits and pieces they understand or that fit with their feelings.

[LA_Price]

The mistake the turkeys are making is in believing that in an extremistan world, their observations or their commonly held belief based on past observations will help them predict the future. They can never assume that the humans will not chop their heads off tomorrow, no matter how many previous days they observe. The 90%+ of turkeys that operate on the assumption that the humans have "always been benevolent and always will be" can not predict what is going to happen. The humans know what happens on thanksgiving , but they don't and so to them they're just going to get their heads chopped off on a random day.

[vhawk01]

[ QUOTE ]
The mistake the turkeys are making is in believing that in an extremistan world, their observations or their commonly held belief based on past observations will help them predict the future. They can never assume that the humans will not chop their heads off tomorrow, no matter how many previous days they observe. The 90%+ of turkeys that operate on the assumption that the humans have "always been benevolent and always will be" can not predict what is going to happen. The humans know what happens on thanksgiving , but they don't and so to them they're just going to get their heads chopped off on a random day.

[/ QUOTE ]
Right, but aren't you sort of just ignoring the situations where you are FORCED to make some prediction? The turkeys are going to be wrong a lot of the time, in an extremistan world....but not as often as someone just randomly picking some guess, right? Their previous experiences have far LESS value than in a bell-curve world, but not 0 value.

[PairTheBoard]

[ QUOTE ]
You mention courts. The irony is, the jury system is set up on exactly this premise - the more people that agree with a position on a question of fact, the more likely it is to be correct.


[/ QUOTE ]

Yea, I thought about that and how it applies here. It is Ironic. I think it's the same principle the Court applies when vetting the expert witness. As I pointed out, the Court does consider the consensus among the specialists in the field as to what the best science is that the expert witness is applying. The Jury can be seen similiarly as specialists on the evidence presented in Court. However the point relevant to Sklansky's Inference is that the Jury forms its opinion based on Unbiased Evidence speaking directly to the Merits of the case. They do Not start their deliberations from a view Prejudiced by Public Opinion.

They do Not say to themselves, "most people think he's guilty so I will Start the Line at 60% probabable for guilty and then consider the Trial Evidence from there." In fact, the Court goes to great length trying to find Jury members who know nothing about the case or its public opinion from the beginning for exactly this reason. And the Jury is instructed not to read or listen to media about the case. Flagrant violations can mean juror removal or even mistrial. The Court does not want the Jury Prejudiced by Sklansky's Inference.

So when we at SMP are considering a Proposition we are like the Jury. We want to look at evidence that speaks directly to the merits of the Proposition. We will consider expert testimony if we can determine who is an expert and what constitutes a consensus among specialists in the field for the science involved. But when David steps in and tries to prejudice our deliberations with his Sklansky Inference I think an Objection should be made.

PairTheBoard

[Ron Paul]

http://www.youtube.com/watch?v=ImzFdHd70vU

[David Sklansky]

"They do Not say to themselves, "most people think he's guilty so I will Start the Line at 60%"

It makes no difference which piece of evidence comes first and thus where you "start the line". Baye's Theorem combines the evidence properly regardless of the order of the evidence. I'm trying to remember if were one of those who disagreed with me that if a defendent was about to be reluctantly acquitted by the jury, they should convict if there was last minute evidence that showed that both the perpertator and the defendendent wore 14B shoes but probably not if it was 11D. Those who do disagree are frankly too dumb to discuss anything further with.

[PairTheBoard]

[ QUOTE ]
"They do Not say to themselves, "most people think he's guilty so I will Start the Line at 60%"

It makes no difference which piece of evidence comes first and thus where you "start the line". Baye's Theorem combines the evidence properly regardless of the order of the evidence. I'm trying to remember if were one of those who disagreed with me that if a defendent was about to be reluctantly acquitted by the jury, they should convict if there was last minute evidence that showed that both the perpertator and the defendendent wore 14B shoes but probably not if it was 11D. Those who do disagree are frankly too dumb to discuss anything further with.

[/ QUOTE ]

Evidently you don't realize that resorting to ad hominems like "dumb" here and "silly" in your other thread is tantamount to admitting defeat in the debate. You evidently also didn't learn much in your Thread about how people come to believe things. It is not like you thought. People don't generally calculate a precise "probability".

So you had difficulties constructing a hypothetical where a precise "probability" is calculated that is exactly on the border of a decision. If you could get people to accept the premise of that precise borderline probability then they should follow your conclusion that the most infitesimal additional evidence should sway them. So you were frustrated when people would not accept your premise. Out of that frustration you evidently concluded they were "dumb".

Evidently you continue to hold to your peculiar views on how convictions for beliefs are formed. As I recommended on that thread, I think you should try programming a robot's belief forming software. I think you would be startled at some of the absurd results you would get.

You are evidently frustrated by my observation that in a Court of Law, everything possible is done to insure that the Sklansky Inference is Not applied. In the Manson Case when Nixon's opinion of his guilt was displayed by Manson to the jury, it almost caused a mistrial. The trial went on but the jury was instructed to Disregard that "evidence". The judge did not mention Bayes.

PairTheBoard

[David Sklansky]

"You are evidently frustrated by my observation that in a Court of Law, everything possible is done to insure that the Sklansky Inference is Not applied. In the Manson Case when Nixon's opinion of his guilt was displayed by Manson to the jury, it almost caused a mistrial. The trial went on but the jury was instructed to Disregard that "evidence". The judge did not mention Bayes."

The idea of a jury of your peers is absurd whan it comes to pure whodonut cases. Much better if the jury was comprised of professional evidence evaluators. Meanwhile, because judges know that the average jury member is often close to a borderline moron, literally, they are forced to eliminate certain pieces of evidence because it will sway the jury a lot more than a professional evidence evaluator would be swayed. That's especially unacceptable if the swaying is against the defendant. But what does the fact that people form their opinions in less than cogent ways have to do with the fact that each new piece of evidence SHOULD change your opinion?

[PairTheBoard]

[ QUOTE ]
literally, they are forced to eliminate certain pieces of evidence because it will sway the jury a lot more than a professional evidence evaluator would be swayed.

[/ QUOTE ]

Is that why they don't allow Sklansky's Inference as evidence? Or is it because they consider Public Opinion to be irrelevant to the merits of the case? Sometimes there is no Jury of peers. The Judge or group of Judges decide the case. Are they morons too, or do they qualify as expert evidence evaluators? Do you think in their deliberations they give proper weight to the Sklansky-Inference? Or do you think they hold to the legal principle that Public Opinion is irrelevant to the merits of the case and give it 0 weight?

[ QUOTE ]
But what does the fact that people form their opinions in less than cogent ways have to do with the fact that each new piece of evidence SHOULD change your opinion?

[/ QUOTE ]

It depends on what you mean by "change" and what you mean by "opinion". At what stage are they in forming their "opinion"? What do you call their state of mind before they have formed an "opinion"? In court they are directed Not to form an opinion until all the evidence is in. Let's call that State a State of Deliberation. Does every piece of evidence affect their Deliberation? Of course. Now if you want to try to force some conclusions on that scenario I suggest you start a thread and discuss it.

But the topic of this Thread is the Relevance of the Sklansky Public Opinion Inference when considering a Proposition. In a court of law, when evidence speaking directly to the merits of the Proposition is available, the legal principle is that the Sklansky-Inference is Not Relevant. ie. It should be given zero weight.

Notwithstanding your weak argument about moronic juries.

PairTheBoard

[Taraz]

vhawk, did you hear an interview with Nassim Nicholas Taleb? Extremistan and whatever?

[Trier]

As far as I can see, the original quote from Sklansky is about 99 per cent wrong (no offence DS – love your poker books).

Put into an historical context, there are many theories where popular opinion has been very hardy. Theories like ‘if I don’t eat I’ll die’ and ‘if I step off this 100 metre cliff, things are going to get messy’ are ones where the dissenting minority are uninsurable. Occasionally idiots do try the highly unlikely. The Shakers thought sex was dirty (agree, ain’t it wonderful) and unnecessary, and promptly bred themselves out of existence. So it goes.

But if it comes to a ‘new’ theory, which is subsequently widely accepted, it is always the case that the majority initially disagrees with it. Note this goes for all current scientific theories and the vast majority of social theories, since these are of such recent origin. For a long time scientists gazed into fires and watched the emission of phlogiston. Then someone gazed into a fire and saw the absorption of oxygen. We don’t know 100 per cent that either theory is correct. We do have now, however, far better reasons for believing in oxidisation than the emission of phlogiston. Flat Earthers, we hope, are similarly running out of room.

The big names of science, from Copernicus through Descartes, Kepler, Darwin, Einstein, Watson & Crick, and all the others, are big names PRECISELY BECAUSE they at one stage believed things that the vast majority not only didn’t believe, but thought were self-evidently nonsense. “Of course,” the doubting majority said, “the sun goes around the earth. There it is right now, rising out of one horizon, going across the sky, and setting in the other horizon. It’s absurd to think otherwise.” The doubting majority even enforced this belief with the weight of law for many centuries. So, on the sheer bulk of theories, Sklansky is completely wrong.

The thread has been interesting. The turkey example has gone completely out of control. So far as I know, the genesis of this example was a philosopher of science by the name of Alan Chalmers, long a lecturer at Sydney University and now at the University of Newcastle UK, and a former student of Karl Popper. Alan’s ‘naïve inductionist turkey’ (one) seems to have grown into a super species about to declare a republic and start testing nuclear weapons. Alan’s book, What is this thing called Science? deserves a wide readership.

And most of our swans, do you mind, are black, although I’m beginning to think that’s a worry. Is a civil war afoot in the wide brown land?

That said, Sklansky and some others seem to believe that because nothing whatever can ever be known with absolute 100 per cent certainty, therefore it’s a probability problem, and they miss the point. Scientific theories, and many (but still not very many) social theories, depend on WEIGHT OF EVIDENCE, and the absence of conflicting evidence. A mathematical probability result is merely one piece of evidence.

It pays not to confuse poker with the rest of life. Poker is a closed system among players (as mathematics is a closed system – dead language really – among objects called numbers). I’d like my pair of deuces to beat the three bullets, but it ain’t going to happen. But in life it’s wholly up to me whether I declare the deuces winners, pull out a gun and dispose of the player with the bullets (this strategy, I believe, has been the subject of considerable experimentation).

PairTheBoard can’t escape scrutiny either. His touching belief in the American legal system takes some digesting. Like all common law adversarial systems, it is inherently flawed in that the purpose of the court is to decide on a contest between the advocates, and not to arrive at the truth. But the American system with its excessive theatricality is way behind similar systems in other common law countries such as the UK, Australia, Canada, New Zealand and the majority of other former British colonies. The magisterial legal systems of Continental Europe, and their former colonies, are far superior to the lot.

And PairTheBoard’s apparent belief in a god or gods going ‘Foop’, naturally with a capital ‘F’, to create the universe also beggars, well, belief.

“It is wrong always, everywhere, and for anyone, to believe anything upon insufficient evidence”, WK Clifford, The Ethics of Belief, 1879.

[PairTheBoard]

It looks like you misunderstand both Sklansky and myself.

[ QUOTE ]
A mathematical probability result is merely one piece of evidence.


[/ QUOTE ]

I don't think Skalnsky claims mathematical authority anymore for his Popular Opinion Inference. Only that it is a piece of evidence. He doesn't claim it is a large piece of evidence when considered next to the Weight of Evidence that speaks directly to the merits of the Proposition. He only claims that it is a relevant piece of evidence, however small, that should be included in the deliberations on the Entire Weight of Evidence available for the Proposition. From the quote above it looks like you agree with him.

My contention is that once Better Evidence is available it should be weighed independently of Sklansky's Inference. The Proposition should be analyzed free of prejudice due to Popular Opinion. I only point to the Legal System to show precendent for my position in a setting where the question of weighing evidence has been seriously studied.

As far as my believing in a Foop. You don't know what I believe. I've never disclosed my most personal beliefs on this Forum. I do my best to act as an advocate for positions I see being unfairly attacked.

PairTheBoard

[David Sklansky]

"I don't think Skalnsky claims mathematical authority anymore for his Popular Opinion Inference. Only that it is a piece of evidence. He doesn't claim it is a large piece of evidence when considered next to the Weight of Evidence that speaks directly to the merits of the Proposition. He only claims that it is a relevant piece of evidence, however small, that should be included in the deliberations on the Entire Weight of Evidence available for the Proposition. From the quote above it looks like you agree with him.

My contention is that once Better Evidence is available it should be weighed independently of Sklansky's Inference. The Proposition should be analyzed free of prejudice due to Popular Opinion. I only point to the Legal System to show precendent for my position in a setting where the question of weighing evidence has been seriously studied."

You are aware of course that prosecutors often can not bring up evidence of past crimes. And you hopefully realize that statistically speaking those past crimes change the probability of someone's guilt. Meaning that if two defendents had the same exact evidence against them it would be more likely that the one who committed past crimes is guilty this time. Even judges must surely realize this. They don't allow the evidence because statistical evidence about crimes in general has a feel of unfairness about it. That's why you won't hear it said in court that "its usually the husband". but as a poker player you should realize that evidence like this IS fair unless it is improperly evaluated. Especially if you are betting rather than judging.

[chezlaw]

[ QUOTE ]
You are aware of course that prosecutors often can not bring up evidence of past crimes. And you hopefully realize that statistically speaking those past crimes change the probability of someone's guilt. Meaning that if two defendents had the same exact evidence against them it would be more likely that the one who committed past crimes is guilty this time. Even judges must surely realize this. They don't allow the evidence because statistical evidence about crimes in general has a feel of unfairness about it.

[/ QUOTE ]
but someone with a criminal record is more likely to be incorrectly in the dock in the first place. Maybe it would be unfair to hold their past against them again - not in a perfect world but in the real one.

chez

[jason1990]

[ QUOTE ]
Sklansky has made the following observation:

[ QUOTE ]
DS -
The reason I say that the fact that 90% of people disagreeing with a theory is evidence against it, is because of all theories ever proposed that were eventually settled, those that had 90% disagreeing, turned out to be right much less often than those which had few disagreeing.


[/ QUOTE ]

He has claimed that besides me, "Everyone Else Understands" this. So I ask SMP to take a closer look at it. Exactly what is it saying and how useful is it?

[/ QUOTE ]
The argument roughly has the following form:

1. Object X has property P.
2. Among all objects with property P, relatively few have property Q.
3. Therefore, the probability is small that object X has property Q.

In this case, X is the theory in question, P is the property that 90% of people disagree with it, and Q is the property that the theory is correct. You seem to be focused on the "popular opinion" aspect of this particular argument. That is, you seem most concerned with Sklansy's use of this particular property P. However, if you accept the form of this argument, then it does not matter what property P is used. In other words, you must take issue with the form of the argument, and not with the particular choices of X, P, and Q. So what, if anything, is wrong with this argument?

I find it easier to consider the following equivalent argument, reworded to avoid the double negative:

a. Object X is a member of group G.
b. Most objects in group G have trait T.
c. Therefore, the probability is high that X has trait T.

There is actually an implied premise in this argument. To make it explicit, let us rewrite the argument as

a1. Object X is a member of group G.
a2. We know nothing else about object X.
b. Most objects in group G have trait T.
c. Therefore, the probability is high that X has trait T.

Since the conclusion is probabilistic, a probability model is hiding in the background. The sample space is G, the probability measure is the uniform measure, and the object X is regarded as a random element from this probability space. Under this model, the conclusion is valid. So what, if anything, is wrong with this model?

The model assumes that X is random, when in fact it is some fixed object. There is nothing wrong with this, in my opinion. If you want to make a probabilistic statement about X, then you must assume at least this. The only other aspect of the model to take issue with is the use of the uniform distribution. Why should uniform be the right probability measure to use? This is justified by premise a2 (together with a1). The justification is the "indifference principle": If we know nothing about X other than the fact that it is a member of G, then we should regard X as equally likely to be any member of G.

Is this evidence? If you accept the indifference principle as a valid principle of inference, then yes it is. I suspect most people regard the indifference principle as valid, so this is probably a convincing argument to a majority of people. In that sense, one might call it commonsense evidence.

It is not, however, scientific evidence. Probability models involving the uniform distribution are always built on symmetry. A coin is symmetric, so both sides are equally likely. A die is symmetric, so all faces are equally likely. These symmetries are physical symmetries, so these models have an empirical foundation and can be legitimately regarded as scientific models. The indifference principle, however, relies on symmetry in our lack of knowledge. This is not a physical symmetry. A model built on the indifference principle does not have an empirical foundation. Its foundation is purely philosophical. It is in this sense that the argument does not produce scientific evidence. At best, it produces "philosophical evidence."

Is it prejudicial evidence? We typically think of prejudice as forming an opinion about someone/something based solely on membership in a particular group. In this sense, the argument is about as close to prejudice as you can get. The argument is dressed up a little, though, to avoid looking like true prejudice. We tend to think of true prejudice as given by the following argument:

A1. Person X is a member of group G.
A2. I will ignore all other information about X.
B. I believe the people in group G have trait T.
C. Therefore, X has trait T.

Premise a2 improves A2 by asserting that it is either not possible to learn more about X, or that we are open to revising our conclusion when new facts come to light. Premise b improves B by asserting that it is known as fact that most members of G have trait T. And the conclusion c improves C by introducing probability so as not to unilaterally conclude something definitive about X. With these improvements, the argument does not immediately strike our "anti-prejudice nerve." And, to be fair, the improved argument is actually logically valid, provided we accept the indifference principle. But it is still an argument that draws conclusions based solely on membership in a group, and in this sense it is prejudicial evidence.

Should we bet on the conclusion of the argument? Should we bet even money that X has trait T? An EV argument could justify this if the bet were one in a sequence of independent bets, where each bet was placed on an object chosen uniformly from G. But the bet is not part of such a sequence. The bet cannot be repeated because X is a fixed object. In order to apply the EV argument, we need to imagine that this bet is part of a hypothetical sequence. Are we justified in doing this? At best, one can offer only philosophical justification for this by appealing to the indifference principle. So the same issues remain, whether we are betting or judging.

[luckyme]

[ QUOTE ]

but someone with a criminal record is more likely to be incorrectly in the dock in the first place. Maybe it would be unfair to hold their past against them again - not in a perfect world but in the real one.

chez

[/ QUOTE ]

Not only for this reason, this court example is a terrible fit for the point DS is trying to make. The court system is an agreement to resolve issues within a confined set of rules. Much larger concerns are what determine a lot of what evidence gets in, so the fact that some evidence is excluded says nothing about the value of it as evidence. ( you could have a clear video of the crime excluded because of how it was obtained).

It's been a good idea to explore but very clouded by the comparison to court rules, which aren't about truth but about correct, agreed upon procedure. A fancy version of settling by duel, coinflips or dunking in ponds.

luckyme

[PairTheBoard]

[ QUOTE ]
[ QUOTE ]
Sklansky has made the following observation:

[ QUOTE ]
DS -
The reason I say that the fact that 90% of people disagreeing with a theory is evidence against it, is because of all theories ever proposed that were eventually settled, those that had 90% disagreeing, turned out to be right much less often than those which had few disagreeing.


[/ QUOTE ]

He has claimed that besides me, "Everyone Else Understands" this. So I ask SMP to take a closer look at it. Exactly what is it saying and how useful is it?

[/ QUOTE ]
The argument roughly has the following form:

1. Object X has property P.
2. Among all objects with property P, relatively few have property Q.
3. Therefore, the probability is small that object X has property Q.

In this case, X is the theory in question, P is the property that 90% of people disagree with it, and Q is the property that the theory is correct. You seem to be focused on the "popular opinion" aspect of this particular argument. That is, you seem most concerned with Sklansy's use of this particular property P. However, if you accept the form of this argument, then it does not matter what property P is used. In other words, you must take issue with the form of the argument, and not with the particular choices of X, P, and Q. So what, if anything, is wrong with this argument?

I find it easier to consider the following equivalent argument, reworded to avoid the double negative:

a. Object X is a member of group G.
b. Most objects in group G have trait T.
c. Therefore, the probability is high that X has trait T.

There is actually an implied premise in this argument. To make it explicit, let us rewrite the argument as

a1. Object X is a member of group G.
a2. We know nothing else about object X.
b. Most objects in group G have trait T.
c. Therefore, the probability is high that X has trait T.

Since the conclusion is probabilistic, a probability model is hiding in the background. The sample space is G, the probability measure is the uniform measure, and the object X is regarded as a random element from this probability space. Under this model, the conclusion is valid. So what, if anything, is wrong with this model?

The model assumes that X is random, when in fact it is some fixed object. There is nothing wrong with this, in my opinion. If you want to make a probabilistic statement about X, then you must assume at least this. The only other aspect of the model to take issue with is the use of the uniform distribution. Why should uniform be the right probability measure to use? This is justified by premise a2 (together with a1). The justification is the "indifference principle": If we know nothing about X other than the fact that it is a member of G, then we should regard X as equally likely to be any member of G.

Is this evidence? If you accept the indifference principle as a valid principle of inference, then yes it is. I suspect most people regard the indifference principle as valid, so this is probably a convincing argument to a majority of people. In that sense, one might call it commonsense evidence.

It is not, however, scientific evidence. Probability models involving the uniform distribution are always built on symmetry. A coin is symmetric, so both sides are equally likely. A die is symmetric, so all faces are equally likely. These symmetries are physical symmetries, so these models have an empirical foundation and can be legitimately regarded as scientific models. The indifference principle, however, relies on symmetry in our lack of knowledge. This is not a physical symmetry. A model built on the indifference principle does not have an empirical foundation. Its foundation is purely philosophical. It is in this sense that the argument does not produce scientific evidence. At best, it produces "philosophical evidence."

Is it prejudicial evidence? We typically think of prejudice as forming an opinion about someone/something based solely on membership in a particular group. In this sense, the argument is about as close to prejudice as you can get. The argument is dressed up a little, though, to avoid looking like true prejudice. We tend to think of true prejudice as given by the following argument:

A1. Person X is a member of group G.
A2. I will ignore all other information about X.
B. I believe the people in group G have trait T.
C. Therefore, X has trait T.

Premise a2 improves A2 by asserting that it is either not possible to learn more about X, or that we are open to revising our conclusion when new facts come to light. Premise b improves B by asserting that it is known as fact that most members of G have trait T. And the conclusion c improves C by introducing probability so as not to unilaterally conclude something definitive about X. With these improvements, the argument does not immediately strike our "anti-prejudice nerve." And, to be fair, the improved argument is actually logically valid, provided we accept the indifference principle. But it is still an argument that draws conclusions based solely on membership in a group, and in this sense it is prejudicial evidence.

Should we bet on the conclusion of the argument? Should we bet even money that X has trait T? An EV argument could justify this if the bet were one in a sequence of independent bets, where each bet was placed on an object chosen uniformly from G. But the bet is not part of such a sequence. The bet cannot be repeated because X is a fixed object. In order to apply the EV argument, we need to imagine that this bet is part of a hypothetical sequence. Are we justified in doing this? At best, one can offer only philosophical justification for this by appealing to the indifference principle. So the same issues remain, whether we are betting or judging.

[/ QUOTE ]

In that case I would prefer to have my Judgement of the Proposition based solely on factors speaking directly to its Merits and be free from contamination with philisophical debates over a general issue like the indifference principle.

I've seen the "indifference principle" in action with the Two Envelope Problem.

PairTheBoard

[Trier]

Let's get back to where we started.

[ QUOTE ]
DS -
The reason I say that the fact that 90% of people disagreeing with a theory is evidence against it, is because of all theories ever proposed that were eventually settled, those that had 90% disagreeing, turned out to be right much less often than those which had few disagreeing.


[/ QUOTE ]

Presumably we’re dealing only with this bald quote, so let’s untangle it (let’s face it, it’s pretty tortuous). It devolves to the following propositions:
1. 90% of people disagreeing with a theory is evidence against that theory.
2. This is true BECAUSE of the historical evidence.
3. Historically, of all theories ever proposed and ‘settled’, those with 90% disagreeing turned out to be wrong more often than they turned out to be right.
4. (Implied corollary) Historically, of all theories proposed and ‘settled’, those with only 10% agreeing turned out to be wrong more often than they turned out to be right.

I’ve already pointed out above that propositions 3. and 4. are simply wrong, a misreading of history. So we’re left with propositions 1. and 2. But 2. gets eliminated because of the invalidity of 3. and 4..

Here it gets a bit confusing because we don’t have any context. Given David’s interests, we have to make the assumption that the contextual argument is probabilistic/statistical/mathematical. If that is the case, we have to evaluate the vox populi as a source of evidence for any proposition.

PTB and, for all I know, David, blinds us with what amounts to jargon. But the value of the vox populi as evidence cannot be evaluated probabilistically or syllogistically. The only way that voice can be evaluated is in terms of what it says about this or that proposition in these or those circumstances. In other words, the question always has be “What is the real world proposition?”

Now, the two ways in which the real vox populi is generally expressed are (a) at the ballot box and (b) in the modern deluge of populist polls. Neither of these is very encouraging. If you think you’ve had it rough with your George Bush for eight years, we’ve had ours for ten. Indeed, for the last eight years your George Bush has been running our country, and he wasn’t even a candidate at our election. This is because we have a Prime Minister stuck so far up The Shrub’s rectum he could give him a downstream tonsillectomy.

Here, both the 2001 and 2004 elections were single-issue elections. In 2001 the PM told the people that hordes of ‘boat people’ (read desperate asylum seekers) were about to descend on Oz as some sort of gravitational imperative. “Look at the map. They’re up there, we’re down here. Of course they’re coming.” And the vox populi voted him in.

In 2004 the PM went on interest rates, and that everyone was going to lose their homes (to whom we weren’t told) if they voted for the Opposition. And he got in again.

Of course, in both cases, with hindsight, he was seamlessly wrong on both issues. No boat people and interest rates went up anyway, and no one lost their home.

Bottom line, if anything, if the vox populi agrees with you, it’s time to go and have another look at your facts. That is, if anything, the agreement of the majority of the population historically looks like evidence AGAINST a proposition.

As for PTB’s sensitivities, I simply took the odds after you criticised vhawk for a ‘weak-atheist’ position. Let’s face it, with Septics, they’re pretty good you believe in some sort of deity. Personally, I don’t hide and don’t make any bones about where I’m coming from. I’m a strong, indeed evangelical, atheist.

But I'm with PTB on his last post.

[MaxWeiss]

No, the general population has been generally schooled to have SOME semblance of logic and understanding and thus, when 90% agree, it is because they have heard through the grape vine that people who "know" (scientists and such) agree.

This is of course not always true, but as a Bayes deduction it certainly does increase the likelihood of something being true, prejudice or not.

[PairTheBoard]

[ QUOTE ]
No, the general population has been generally schooled to have SOME semblance of logic and understanding and thus, when 90% agree, it is because they have heard through the grape vine that people who "know" (scientists and such) agree.

This is of course not always true, but as a Bayes deduction it certainly does increase the likelihood of something being true, prejudice or not.

[/ QUOTE ]

Exactly what did I say that you're saying "No" to? Do you think the evidence should be admissable in a court of law? Do you think Jurors should bias their weighing of the evidence presented in court because "everybody thinks he's guilty"? Do you think you should apply this "Baysian deduction" to a new scientific theory when analyzing its possible validity? Exactly what did I say that you disagree with? You replied to a lengthy OP.

PairTheBoard