Absence of Evidence is Evidence of Absence

[MorpheusSandman]

As far as evidence of absence/absence of evidence: a cable news channel last night featured a side-bar issue concerning the mystifying disappearance of Malaysian Air 370... Supplanting the disturbing mystery with a theory, no matter how unreasonable or irrational, somehow alleviates fear by allowing a small measure of control over the uncontrollable.

What seems to me to be a flaw in the argument is that we have no reason to suppose that anything exists without some kind of evidence for it.

I wanted to quote these two posts together as I think they're related to each other. AuntShecky hits on the reason WHY there is often such outlandish speculation for the existence of things/causes even in the absence of evidence. The problem with the human mind is that it will sooner invent a possible answer than it will immediately believe it and begin trying to find evidence for it, no matter how specious. As she says, this is directly related to our fear/anxiety at the unknown, the primal need to supplant that mystery with SOME answer, no matter how ludicrous. That instinct makes it easy to suppose the existence of any number of things with no evidence, and then suppose the absence of evidence has no bearing on the likelihood of their existence (when it does). Obviously, what Hawkman says is true, but AuntShecky's post illustrates precisely why we don't live in a world where people believe that "we have no reason to suppose that anything exists without some kind of evidence," and, indeed, people, in general, have a very poor notion of what substitutes real evidence as opposed to false evidence, weak evidence as opposed to strong evidence, or how any kind of evidence affects different possible explanations or existences. The point being that people are not rational, and not being rational leads to nonsense like "absence of evidence is not evidence of absence," even when such a saying is mathematically provably wrong.

[MorpheusSandman]

There you go - now you understand. We're not talking about the future, or at least you weren't.

I find it strange that you can type that - which completely negates the statement, yet you continue to try to defend it.

Ho hum.

I still have no idea what you're on about or how it relates to anything I've said. That things that exist CAN provide evidence for their existence some time in the future or COULD HAVE in the past, while things that don't exist CAN NOT or COULD NOT have, feeds directly into the notion that "absence of evidence is evidence for absence." If you understood the math as I've been expressing it you should realize why this is. Math and probability doesn't just stop working because we can't know the future.

[Hawkman]

I wanted to quote these two posts together as I think they're related to each other. AuntShecky hits on the reason WHY there is often such outlandish speculation for the existence of things/causes even in the absence of evidence. The problem with the human mind is that it will sooner invent a possible answer than it will immediately believe it and begin trying to find evidence for it, no matter how specious. As she says, this is directly related to our fear/anxiety at the unknown, the primal need to supplant that mystery with SOME answer, no matter how ludicrous. That instinct makes it easy to suppose the existence of any number of things with no evidence, and then suppose the absence of evidence has no bearing on the likelihood of their existence (when it does). Obviously, what Hawkman says is true, but AuntShecky's post illustrates precisely why we don't live in a world where people believe that "we have no reason to suppose that anything exists without some kind of evidence," and, indeed, people, in general, have a very poor notion of what substitutes real evidence as opposed to false evidence, weak evidence as opposed to strong evidence, or how any kind of evidence affects different possible explanations or existences. The point being that people are not rational, and not being rational leads to nonsense like "absence of evidence is not evidence of absence," even when such a saying is mathematically provably wrong.

Of course, my postulation is rather dependent on the nature of belief. If one believes the crashing of an airliner is evidence for the existence of ET, then as you say, it's rather weak evidence. It is infinitely challengeable and has to be weighed in the scales of probability and likelihood. Likewise, that absence of evidence is not evidence of absence when applied to black holes would be ample proof of the statement, unless, of course, one subscribed to the position that our belief in the possibility of black holes, as first suggested by deduction from a mathematical hypothesis, conjured them into being and subsequently allowed for their detection by the physical observation of their effects. An interesting if solipsistic argument, perhaps, but it is, in my opinion (though I concede that I'm not an astrophysicist or quantum mechanic) unlikely. They almost certainly existed before we even conceived of the idea of them or were able to detect any evidence of their existence.

[YesNo]

Rummy said it best.

There are known knowns. These are things we know that we know. There are known unknowns. That is to say, there are things that we know we don't know. But there are also unknown unknowns. There are things we don't know we don't know.

http://youtu.be/_RpSv3HjpEw


The hilarious part is the guy at the end.

The only case Rummy missed was the one about those unknowns that we know.

I wonder how this would apply to the Emperor's New Clothes.

Absence of evidence (that he's wearing clothes) is or is not evidence of absence (of his clothing).

Of course, we all know he's wearing clothes. Just ask the tailors for the evidence.

[The Atheist]

I still have no idea what you're on about or how it relates to anything I've said.

Yes. I've noticed that.

I suspect that's because it doesn't fit with your flawed idea. The maths is no problem, and I've agreed that it works in cases where we already have some knowledge, and on idiotic premises as to whether Caesar wrote the Bond stories.

It just doesn't work when you can't assign a probability. I'm enjoying watching you come up with ever more-unlikely hypotheses to prove your point though. (Tip: instead of asking whether Julius Caesar wrote Bond, ask whether he wrote an ode to Romulus three weeks before his murder.)

Just carry on - it's not a subject I'm bothered enough about to argue the toss with you on. It seems most people can see the flaws even if you can't.

[Nick Capozzoli]

This "Absence of Evidence is Evidence of Absence" thread makes me think of a joke I heard many years ago. It goes like this:

"An archeologist, who happened to be an avid amateur radio hobbyist, had spent years exploring archeological sites in Peru. He noted that after years of excavations in the Andes, he had never found any sort of metallic wire, such as the type used in wired telegraphy. He concluded that this proved that the ancient Andean societies he was studying must have had "wireless" communications.":

[MorpheusSandman]

It just doesn't work when you can't assign a probability.

Of course it does! This is just an absurdly wrong statement. Yudkowsky uses the example of the possible existence of the Fifth Column during WW2. You can't "assign a probability" to the Fifth Column's possible existence, but you can certainly say that the absence of evidence it exists is evidence of its absence. What you're saying is as silly as saying that if we don't know the numbers involved, we can't know that in adding integers besides zero the sum will always be more than either integer. Of course that's nonsense as well. Just like we can know that adding positive integers will lead to a sum more than either integer without knowing the numbers involved, we can know absence of evidence is evidence of absence because of Bayes' even if we don't know the probabilities.

The only probabilities we need to know to prove the statement are given in the phrase itself: "absence of evidence" means that the probability of B|A (where A is something's "absence," | is "Given," and B is "absence of evidence") is always 100%, while B (absence of evidence even if A exists) is less than 100%. Since B|A is 100% and B is not, the probability of A's ABSENCE WILL ALWAYS INCREASE. If A always increases, then the "absence of evidence" is always "evidence of absence." That deserves repeating: whatever the probability of A's non-existence, any absence of evidence necessarily increases the probability of A's non-existence.

The only FAIR criticism to be made is that we don't (often can't) know precisely how much it increases the probability of A's non-existence, but it must be more than 0, and if it's more than 0, the original phrase is demonstrably wrong.

It seems most people can see the flaws even if you can't.

I'm not convinced anyone here actually understands either the phrase or the math. I dare you to explain the math on your own and how it shows my argument is flawed. I've presented my argument in pure mathematical terms, a proof; for you to say it's flawed or not a proof, you need to present your own. Right now all you're doing is blowing hot air.

[MorpheusSandman]

This "Absence of Evidence is Evidence of Absence" thread makes me think of a joke I heard many years ago. It goes like this:

"An archeologist, who happened to be an avid amateur radio hobbyist, had spent years exploring archeological sites in Peru. He noted that after years of excavations in the Andes, he had never found any sort of metallic wire, such as the type used in wired telegraphy. He concluded that this proved that the ancient Andean societies he was studying must have had "wireless" communications.":

At least it's better than concluding that the absence of evidence for Andean wired telegraphy isn't evidence for the absence of Andean wired telegraphy!

[MorpheusSandman]

Of course, my postulation is rather dependent on the nature of belief. If one believes the crashing of an airliner is evidence for the existence of ET, then as you say, it's rather weak evidence.

It's not even weak evidence, it's no evidence at all. The problem there is one of fake causality, like using "phlogiston" to explain "fire." It demonstrates the problem of beliefs that don't depend upon anticipated experience, but rather uses any experience as evidence for said beliefs. Any proper belief should both anticipate and constrain experience, so that, hypothetically, some events would increase the likelihood of the belief while others would decrease the likelihood of the belief.

Likewise, that absence of evidence is not evidence of absence when applied to black holes would be ample proof of the statement...

Not quite sure what you mean here as there IS evidence for black holes. EG, when gas reaches millions of degrees it glows in a way that can be detected by X-ray scopes, and looking through such scopes one can see such lights "spiraling" towards a center before it disappears. Hard to explain how that observation would be possible without the existence of black holes.

[Hawkman]

It's not even weak evidence, it's no evidence at all. The problem there is one of fake causality, like using "phlogiston" to explain "fire." It demonstrates the problem of beliefs that don't depend upon anticipated experience, but rather uses any experience as evidence for said beliefs. Any proper belief should both anticipate and constrain experience, so that, hypothetically, some events would increase the likelihood of the belief while others would decrease the likelihood of the belief.

Well I did say it was dependent on belief.

Not quite sure what you mean here as there IS evidence for black holes. EG, when gas reaches millions of degrees it glows in a way that can be detected by X-ray scopes, and looking through such scopes one can see such lights "spiraling" towards a center before it disappears. Hard to explain how that observation would be possible without the existence of black holes.

Tish Tish and pshaw! Taking a line out of context and building an argument around it is no argument and is the first step on the road to sophistry If you have no evidence then you have an absence of evidence, regardless of whether the evidence exists or not. This was the point I was making, unless of course you subscribe to the position that black holes did not exist until we had the capacity to predict their existence, etc.... It is obviously necessary to define what one means by 'absence'. I use it in the sense of 'not present' in that one does not have it. This does not mean that it does not exist somewhere else or that it cannot be found at some time in the future. Ergo, simply put, an absence of evidence is not evidence of absence. It is, as you say, 'no evidence at all.'

[MorpheusSandman]

Well I did say it was dependent on belief.

Real evidence doesn't just depend on belief, it depends on how belief anticipates and constrains experience. If a belief can claim anything that happens (after it happens) as evidence for itself, then nothing is really evidence "for it" because there can't be any evidence "against it."

If you have no evidence then you have an absence of evidence, regardless of whether the evidence exists or not. This was the point I was making, unless of course you subscribe to the position that black holes did not exist until we had the capacity to predict their existence, etc.... It is obviously necessary to define what one means by 'absence'. I use it in the sense of 'not present' in that one does not have it. This does not mean that it does not exist somewhere else or that it cannot be found at some time in the future. Ergo, simply put, an absence of evidence is not evidence of absence. It is, as you say, 'no evidence at all.'

Me saying that the crash of a plain isn't evidence for the existence of aliens is completely different than me saying "absence of evidence is evidence for absence." These are two different things: one is about what is and isn't evidence IN GENERAL and the other (absence of evidence is...) assumes we already know what evidence is and that there isn't any.

Where you're going wrong with black holes is in thinking in terms of hindsight bias: "black holes always existed even before we knew they did, so the absence of evidence for black holes before we knew they existed wasn't evidence they didn't exist." This is just plain wrong. We can't know before we find the evidence of something's existence whether or not the something exists, but we do know two things:

1. If the thing doesn't exist the probability of there being a continuing absence of evidence is 100%
2. If the thing exists the probability of there being a continuing absence of evidence is less than 100%

Because of this, a continuing absence of evidence IS evidence that something doesn't exist. Why? Because we'd EXPECT an absence of evidence for something that doesn't exist MORE than we'd expect an absence of evidence for something that exists. The only way for this NOT to be the case is if 2. was also 100%. But 2. CAN'T BE 100% EVER! To suppose that 2. could be 100% is to assume that everything we know exists leaves no evidence of its existence; which is absurd and obviously wrong.

The fact that we found evidence for black holes supports the notion that "absence of evidence is evidence for absence," because the reverse is also true: "existence of evidence is evidence for existence." With black holes, we now have "evidence for existence," not an "absence of evidence."

I think this thread is just another example of how counter-intuitive probabilities are to people, like the Monty Hall Problem that continues to confuse everyone who hasn't studied probabilities.

[Hawkman]

It seems to me your model of the universe is one of perpetual existence, static,unchanging. See your points 1 & 2. Take the case of the encylopedigazumper. The encylopedigazumper does not exist. Why? Because I haven't built it yet. Ergo there is no evidence for its existence. At least until this moment where the concept of an encylopedigazumper has been postulated, even though I have declared that it does not yet exist. Therefore the potential for it has increased from zero to more than zero. If I actually build it, then it will exist and therefore your point 1 will be unsustainable. It is therefore a false premise. An equation built on a false premise is therefore itself a false premise.

[Hawkman]

"An archeologist, who happened to be an avid amateur radio hobbyist, had spent years exploring archeological sites in Peru. He noted that after years of excavations in the Andes, he had never found any sort of metallic wire, such as the type used in wired telegraphy. He concluded that this proved that the ancient Andean societies he was studying must have had "wireless" communications.":

Well it's quite true, they did have wireless communications. Communication was by runner with a piece of knotted string or a cleft stick

[MorpheusSandman]

It seems to me your model of the universe is one of perpetual existence, static,unchanging. See your points 1 & 2.

No.

Take the case of the encylopedigazumper. The encylopedigazumper does not exist. Why? Because I haven't built it yet. Ergo there is no evidence for its existence. At least until this moment where the concept of an encylopedigazumper has been postulated, even though I have declared that it does not yet exist. Therefore the potential for it has increased from zero to more than zero. If I actually build it, then it will exist and therefore your point 1 will be unsustainable. It is therefore a false premise. An equation built on a false premise is therefore itself a false premise.

This is just silly: something that doesn't exist but that we have the power to create, like your encylopedigazumper, is quite different than things that may or may not already exist, like Black Holes or God or a book written by Caesar or undiscovered species of frogs in a rain forest, that we can't create; the latter is what's meant by the "absence." I mean, it's apparent in the premise since the phrase deals with the things for which we don't know whether or not they exist; you know your encylopedigazumper doesn't exist so 1. isn't even applicable in that situation. 1. only applies for things in which we DON'T know if they exist.

That said, we can modify the phrase to fit your encylopedigazumper in this way: let's say we don't know if it's possible to create a encylopedigazumper. If it's NOT possible, then we'd to always have no evidence of its existence (ie, it not being created). Yet, if it's possible to create a encylopedigazumper then there is some probability more than 0% for there being evidence of its existence in the form of its invention. The longer you try to invent a encylopedigazumper without success, the more likely it is that it can't be invented, as you'd expect continued failure if it was impossible to create the thing. So Bayes can be applied as much to your encylopedigazumper, to things that may or may not be possible to create, as to things that may or may not already exist.

Here's a better example. You have a coin that may or may not have a tails side. You are not allowed to look at both sides. The coin is always placed on your thumb to flip heads-side up. After 2 flips of heads, is this absence of evidence for a tails-side evidence of the tail sides' absence? What about after 5 straight heads? 10? 100? The answer is fairly obvious, but it feeds directly into the phrase: if the coin has a tails side, we expect to see no evidence for it, the coin coming up heads, only SOME of the time; if the coin DOESN'T have a tails side, if it is "absent," then we expect to see not-tails ALL of the time. The continuing absence of evidence is evidence for the tail-side's absence.

Now, The Atheist would object to the above by saying "but with the coin you KNOW the probabilities precisely, so of course the 'absence of evidence is evidence for absence.'" but The Atheist is wrong because you can phrase the issue this way: you have 5 dice in your hand with an unknown number of sides. These dice may or may not have the number 6 on them. They are not loaded. Even without knowing how many sides there are on the dice, if you continue rolling them, the absence of 6 becomes more likely with each roll where no 6 appears. Even if we say we don't know HOW MUCH more likely the absence of 6 becomes with each roll, the probability of its non-existence must increase by more than 0. The only ways in which it wouldn't increase by more than 0 was if:

1. The number of sides on the dice were infinite (impossible)
2. The dice had a 6 which couldn't be rolled (impossible if the dice aren't loaded, which they aren't).

[Emil Miller]

The only case Rummy missed was the one about those unknowns that we know.

You haven't been paying attention to what Rummy was saying. He distinctly mentions known unknowns just before he refers to unknown unknowns. Now if he had said there are known unknowns that we know are known or unknown knowns that we don't know are unknown that would have answered all queries relating to the problem; except for those unknown unknowns that we don't know are known.