‘Sexed-up’ numbers should not always be accepted as science

[snopes]

In the recent flurry of moves to ban plastic bags a frequently cited statistic is that more than 100,000 marine mammals and sea turtles die each year from entanglement in, or ingestion of, plastic bags.

The original scientific study upon which this estimate relied actually attributed these deaths to fishing tackle in the oceans, not plastic bags. Yet the terms “100,000 marine deaths” and “plastic bags” now circulate happily through our public discourse, solidified as established fact.

But when is a fact a fact? Can facts change over time? And does it matter if they do? Science is instinctively referred to as the source and authenticator of facts such as the one cited above, and rightly so.

Yet as this example shows, we need to be very careful about the veracity of the numbers we latch on to, and about what they signify. What may start out as a credible, yet qualified and provisional, scientific estimate may end up, either through distortion or mere negligence, enduring as an urban myth, apocryphal numbers – the modern equivalent of folklore.

http://www.timesonline.co.uk/tol/new...cle3508277.ece

[snopes]

Scientists and environmentalists have attacked a global campaign to ban plastic bags which they say is based on flawed science and exaggerated claims.

The widely stated accusation that the bags kill 100,000 animals and a million seabirds every year are false, experts have told The Times. They pose only a minimal threat to most marine species, including seals, whales, dolphins and seabirds.

http://www.timesonline.co.uk/tol/new...cle3508263.ece

[Grand Illusion]

Modernfolk are enamored of scientific studies because they are viewed as objective and infallible, especially when properly conducted, but there are some underlying problems when dealing with their practical applications. I realize that I'm preaching to the choir on this board, but:

1) Scientific studies are limited in their scope and only address specific, pre-determined issues. Their findings must not be transferred or analogized. A health study on South Americans who subsist on a diet of leaves cannot be used to draw conclusions about leaves in the diets of Africans. This is tempting to do because there might not be any studies on African leaf diets. "This is the closest we could get" is not good enough to stand up to scientific scrutiny.

2) The transition from objective numbers to practical application is subjective. There are a lot of judgment calls between "18% of Americans cannot name the first five presidents" and "Americans are ignorant about their history." I might say that 18% is acceptable, and you might say that it's not.

3) There is no such thing as a no spin zone. Casting a statistic into a context automatically puts a bias on it. "There are more guns per capita in the U.S. than in Canada" might be a true, objective statistic, but it sends a different message than "There are fewer guns per capita in the U.S. now than 100 years ago," also a true, objective statistic. This is an easy way to make statistics say what you want them to say while still claiming that the statistics are accurate and therefore your position is indisputable.

Let me know if I've missed any key points.

* Disclaimer: The statistics above are ficticious.

[Troberg]

Let me know if I've missed any key points.

Not really missed, but I'd like to expand #1.

Scientific tests and models based on them are only valid within the scope of that model.

There is a famous thought experiment on this: "You live in a room. It's warm and cosy, and every day someone comes to feed you and take care of you and makes sure you are healthy. You are not hurt. This is all you know, so after a while you make the assumption that the being that takes care of you means you well and is your friend. Then Christmas comes, and you're the pig (or, in the US, the turkey)."

This is kind of what they are doing with the climate change today. They are using historical data to try to predict something that goes outside the frame of that historical data, which is very dicey science. Mathematicians are stricter about this, they are not satisfied that no one has found (as an example) a recurring period of digits in pi as evidence, they require evidence that such an occurance can not happen.

It's easy to become locked up in a model and don't see where it's not valid anymore. Somewhat like this old riddle:

At the Logic club in Berlin, only those who have the correct password gets in.
If the doorman say 28, the answer is 14, if he say 8, the answer is 4 and if he say 16, the answer is 8.
Now, an enterprising fellow thinks he has solved the problem. He walks up to the doorman and the doorman says 14. Our hero answers 7 and is immediately thrown out.
What would have been the correct answer?

[ganzfeld]

Originally Posted by Troberg

This is kind of what they are doing with the climate change today. They are using historical data to try to predict something that goes outside the frame of that historical data, which is very dicey science.

Without any specifics (who "they" are, for example), this accusation is baseless.

[Troberg]

Without any specifics (who "they" are, for example), this accusation is baseless.

The scientists who use models based on historical data and extrapolates it to draw conclusions about future climate that goes outside the bounds of the historical data. For instance, claiming that based on measurements and historical data, we will see a climate that has no previous historical counterpart is just wrong. It's guessing that the models will hold outside the tested limits. They might work, but we don't know.

[Friends of Alfred]

Originally Posted by Troberg

At the Logic club in Berlin, only those who have the correct password gets in.
If the doorman say 28, the answer is 14, if he say 8, the answer is 4 and if he say 16, the answer is 8.
Now, an enterprising fellow thinks he has solved the problem. He walks up to the doorman and the doorman says 14. Our hero answers 7 and is immediately thrown out.
What would have been the correct answer?

Pure guess here - 8?

[Richard W]

Originally Posted by Troberg

At the Logic club in Berlin, only those who have the correct password gets in.
If the doorman say 28, the answer is 14, if he say 8, the answer is 4 and if he say 16, the answer is 8.
Now, an enterprising fellow thinks he has solved the problem. He walks up to the doorman and the doorman says 14. Our hero answers 7 and is immediately thrown out.
What would have been the correct answer?

The answer could also be 8...

[Silas Sparkhammer]

Originally Posted by Troberg

The scientists who use models based on historical data and extrapolates it to draw conclusions about future climate that goes outside the bounds of the historical data. For instance, claiming that based on measurements and historical data, we will see a climate that has no previous historical counterpart is just wrong. It's guessing that the models will hold outside the tested limits. They might work, but we don't know.

While it is true that all extrapolations are incorrect, some have more validity than others.

We have very good reasons to believe that the earth's human population will increase in the next twenty years. Sure, a vast, all-encompassing war might come along and scotch that, but the population model can't include everything.

Ditto, a gigantic volcano might erupt, putting dust into the atmosphere, and leading to short-term global cooling. But absent such events, the global warming model appears to be valid.

"The battle is not always to the strong, nor the race to the swift, but that's the way to bet." And since global warming is an issue of both abstract science and practical public policy, we are betting the farm. Literally.

Silas

[Troberg]

Pure guess here - 8?

and

The answer could also be 8...

Why? The riddle is not just the answer, it's also the explanation.

[Dr. Dave]

Originally Posted by Troberg

and



Why? The riddle is not just the answer, it's also the explanation.

8 would fit the following pattern:
Doorman states number >20- password = 14
Doorman states number >10, <20- password = 8
Doorman states number <10- password = 4

So doorman states 14, password = 8

Or

The code could be closest even number to the doorman number / 2, always round up.
i.e. doorman number/4, round up to nearest integer, times 2

[Esprise Me]

Originally Posted by Troberg

At the Logic club in Berlin, only those who have the correct password gets in.
If the doorman say 28, the answer is 14, if he say 8, the answer is 4 and if he say 16, the answer is 8.
Now, an enterprising fellow thinks he has solved the problem. He walks up to the doorman and the doorman says 14. Our hero answers 7 and is immediately thrown out.
What would have been the correct answer?

Originally Posted by Dr. Dave

8 would fit the following pattern:
Dorman states number >20- password = 14
Dorman states number >10, <20- password = 8
Dorman states number <10- password = 4

So doorman states 14, password = 8

I was thinking that perhaps there was no mathematical pattern and each number was paired with another, so that if the doorman says 28, you say 14; if the doorman says 14, you say 28. Or, it could be that any number over 15, you give half, any number under 15, you give double. Or, perhaps, if the number given cannot be divided by two and yield an even-numbered result, then it is doubled instead. Anyway, if the answer is not 28 I'm stumped.

[unbroken]

I can't see the title of this thread without thinking "What, like 69?".

[Dr. Dave]

Originally Posted by Esprise Me

I was thinking that perhaps there was no mathematical pattern and each number was paired with another, so that if the doorman says 28, you say 14; if the doorman says 14, you say 28.

That's a good thought, but Troberg said if the doorman says 16, the answer is 8 and if he says 8, it is 4. So 8 is not paired with 16.

Or, it could be that any number over 15, you give half, any number under 15, you give double.

But 8 yields 4, not 16.

Or, perhaps, if the number given cannot be divided by two and yield an even-numbered result, then it is doubled instead.

Excellent- that would fit the pattern too and the answer would be as you say, 28.

Anyway, if the answer is not 28 I'm stumped.

How about 8, for the reason I gave?

[Richard W]

Originally Posted by Troberg

Why? The riddle is not just the answer, it's also the explanation.

It's the number of letters in the German word for each number (they're in Berlin, so the doorman is saying the numbers in German).

[Dreams of Thinking Machines]

Originally Posted by Troberg

It's easy to become locked up in a model and don't see where it's not valid anymore. Somewhat like this old riddle:
At the Logic club in Berlin, only those who have the correct password gets in.
If the doorman say 28, the answer is 14, if he say 8, the answer is 4 and if he say 16, the answer is 8.
Now, an enterprising fellow thinks he has solved the problem. He walks up to the doorman and the doorman says 14. Our hero answers 7 and is immediately thrown out.
What would have been the correct answer?

Hmm, this is going to be a complete shot in the dark, but maybe the Logic club is giving us a deviously misleading trick from propositional logic? Suppose they're not asking us to find a mathematical pattern, but rather, the logical form of the riddle:
P1: If the doorman says 28, then the answer is 14.
P2: The doorman says 14.
Conclusion:Therefore, 7 is the answer. [That's clearly not valid]

The answer being 14 is a consequent of P1's antecedent. When the doorman says '14' he could simple be reinterating that '14' is the consequent of 28. We cannot validly infer 7 from 14 because there is no conditional of the form, "If the doorman says 14, then the answer is 7" to allow this. We cannot validly infer anything (except for things like tautologies) from the antecedent of a conditional.

We might assume that the answer is simply taking what the doorman says and dividing it by two, or we might assume that the number of German letters being the same as the intended answer. However, like most ordinary language, the intended meaning is ambiguous and so we can interpret it in a number of ways.

Also, one interpretation of the riddle, which I highlight in bold, suggests that the man was already in the club before he was thrown out.

[TomToad]

Originally Posted by Dreams of Thinking Machines

Also, one interpretation of the riddle, which I highlight in bold, suggests that the man was already in the club before he was thrown out.

I noticed that when reading the question. I just assumed it was a mistake when he typed it and meant that the guy was not allowed in. But if it was intentional, then it is possible that the guy got in, but was immediately thrown out for an unrelated reason. So that would mean the answer is, in fact, 7.

[Troberg]

It's the number of letters in the German word for each number (they're in Berlin, so the doorman is saying the numbers in German).

Correct.

You see how easy it is to get stuck in a model, in this case mathematics.

[damian]

Originally Posted by Silas Sparkhammer

While it is true that all extrapolations are incorrect, some have more validity than others.

Unless you have tested all extrapolations, you cannot say that they are all wrong.

Originally Posted by Richard W

It's the number of letters in the German word for each number (they're in Berlin, so the doorman is saying the numbers in German).

There is nothing in the text that states they are speaking German. In fact, there is nothing in the text that states the "Logic Club in Berlin" is even in Germany. You made that inference yourself. (Goes with the above point ).

[Barns & No Bull]

Originally Posted by Troberg

What would have been the correct answer?

Any answer that is an even number will get you inside the club. The guy answered with an odd number and was barred from the club.

We can't know if this rule would still apply if the doorman gives an odd number because no examples of that are given.

Unless you know more than this problem offers you, any valid pattern is as good as any other valid pattern. From what I can see, even numbered answers get you in and an odd numbered answer gets you tossed.