Evidence and all that

[chezlaw]

[ QUOTE ]
[ QUOTE ]
In that sentence, are you using "predictions about the world" as a synonym for "statements of any kind," or as a synonym for "testable predictions."

If you mean "statements of any kind," then the theories are identical -- in which case we don't have two theories, but one (making your claim a tautology).

If you mean "testable predictions," then I will disagree and say that, for meta-evidential reasons, "the earth goes around the sun" is more likely to be true than "the earth goes around the sund because it is being pushed by invisible angels" even though the two "theories" make all the same testable predictions.

[/ QUOTE ]

I don't think that's what he's saying (not that I'm too clear on it by any stretch), but I think it's basically that if there are two (non-identical) statements about something having nothing to do with the empirical world, then there's no evidence that can be offered for or against either.

E.g.: T1: In heaven all houses are yellow
T2: In heaven all houses are green.

Well then, yeah, I guess he's logically correct that there isn't any evidence that is relevant, but that's because "evidence" connotes an observable fact and his "theories" expressly provide that they have nothing to do with the observable world.

So it's a logically valid, I suppose, it's just meaningless and unhelpful.

[/ QUOTE ]
Got there. It's logically valid but the reason I brought it up is that people keep claiming that we should have beliefs about which of two theories are true when there's absolutely no way ever of deciding between them.

I say that it is meaningless to talk about which of the theories is true but I get told I'm being silly or wishy-washy and should make a decision based on the evidence.

chez

[chezlaw]

[ QUOTE ]
[ QUOTE ]
That is where meta-evidence comes in. [img]/images/graemlins/wink.gif[/img]

I guess it depends on why the two theories make the same predictions. If it's because the theories are identical, then one is as good as the other.

But if they are not identical, it's just that they differ only with respect to statements that are not testable, then one can still be right and the other wrong -- although it is impossible for us to determine empirically which (if either)is right and which is wrong.

In that case, I think the meta-evidentiary principle is relevant.

[/ QUOTE ]

but no-one ever answers the meta-evidence questions. There are no cases where one of two theories undecidable by normal evidence have been shown to be true, so if you apply meta-evidence you have no normal evidence that the meta-evidence approach is valid.

Hence meta-evidence is a fancy word for something that is not based on evidence.

[/ QUOTE ]

I don't think this is correct, if I understand what you are claiming. One is using meta-evidence as support for a meta-theory about which of the two indistinguishable theories is likely to be correct. That meta-evidence does not need to arise from observations about indistinguishable theories, though!

Here's a crude example: If I drop a lot of different objects, eventually I will arrive at the theory All Objects Drop. If I then find a new object of a type I have never seen before... I will expect it to drop.

Did I understand your point? Is the above clear?

[/ QUOTE ]
Possibly [img]/images/graemlins/smile.gif[/img]. If you can give one (preferably many) examples of when one of two indestinguishable theories has been shown to be correct then you would have some evidence as to how to form beliefs about them.

I'm saying they're are no such examples (logically there never could be) and hence there is no meta-evidence analagous to things falling.

chez

[benjdm]

Believing a God exists who has not interfered / will not interfere with the physical universe since its creation is indistinguishable from atheism anyway.

[chezlaw]

[ QUOTE ]
Theory 1a: Lightning is forged by Hephaestus and sent to earth by Zeus.
Theory 1b: Lightning is an electrostatic discharge... etc. etc. (the details can be easily found on the web).

Theory 2a: Volcanic eruptions are caused by the Goddess Pele using her magic stick Pa'oa and earthquakes are caused by her stamping her feet (both of which occur when she is angry).
Theory 2b: Volcanic eruptions are driven by the ascent of magma.. etc. etc. (the details can be easily found on the web)

Theory 3a: Angels move the planets.
Theory 3b: See Newton, Einstein, etc.

Lather, rinse, repeat.

Suppose that you will grant me that all the 'b's are accepted as true and all the 'a's rejected. Then one could form the theory, based on this evidence, that supernatural agents are unnecessary to explain natural events. On the basis of that theory, we can choose between your god/no god theories...

[/ QUOTE ]
I like this post, maybe we can grab hold of this meta-evidence idea.

On one hand you have theories that can be distingished by evidence. On the other you have theories that cannot be distingished by evidence.

Do you have any evidence that you can extrapolate from one to the other?

chez

[ QUOTE ]
Possibly . If you can give one (preferably many) examples of when one of two indestinguishable theories has been shown to be correct then you would have some evidence as to how to form beliefs about them.

I'm saying they're are no such examples (logically there never could be) and hence there is no meta-evidence analagous to things falling.


[/ QUOTE ]

I can't show you any such examples without circularity, but I don't have to, and that was the point of my "things falling" example. One of the points of theory is to be able to generalize! In my analogy, the actual objects dropped match up with theories that we actually can independently distinguish. The theory this induces, All Things Drop, matches up with the No Supernatural Explanations theory. The new type of object that we have never observed before matches up with "indistinguishable" theories...

[chezlaw]

[ QUOTE ]
[ QUOTE ]
Possibly . If you can give one (preferably many) examples of when one of two indestinguishable theories has been shown to be correct then you would have some evidence as to how to form beliefs about them.

I'm saying they're are no such examples (logically there never could be) and hence there is no meta-evidence analagous to things falling.


[/ QUOTE ]

I can't show you any such examples without circularity, but I don't have to, and that was the point of my "things falling" example. One of the points of theory is to be able to generalize! In my analogy, the actual objects dropped match up with theories that we actually can independently distinguish. The theory this induces, All Things Drop, matches up with the No Supernatural Explanations theory. The new type of object that we have never observed before matches up with "indistinguishable" theories...

[/ QUOTE ]
Lets have a go at making this precise.

T1, T2 are distingusishable on the basis of evidence and we have found that the true one requires nothing supernatural

same goes for T3, T4 ...

generalisation

For all x,y: If Tx, Ty are distingusishable on the basis of evidence then the true one requires nothing supernatural.

I think that's analogous to your objects dropping but its not the result you need.

Somehow you have to make a leap to claims about which of two indistinguishable theories are true.

chez

[ QUOTE ]
On one hand you have theories that can be distingished by evidence. On the other you have theories that cannot be distingished by evidence.

Do you have any evidence that you can extrapolate from one to the other?

[/ QUOTE ]

The No Supernatural Explanations theory applies on the face of it to the god/no god pair of theories. The point of theory making in general is to extrapolate, even to situations that are counterfactual and thus can never be directly tested. You could ask on what basis is any theory generalized from data... what evidence do you have that the new data will obey the same rules as the old? At some point an inductive bias is needed. I will grant you that.

[chezlaw]

[ QUOTE ]
[ QUOTE ]
On one hand you have theories that can be distingished by evidence. On the other you have theories that cannot be distingished by evidence.

Do you have any evidence that you can extrapolate from one to the other?

[/ QUOTE ]

The No Supernatural Explanations theory applies on the face of it to the god/no god pair of theories. The point of theory making in general is to extrapolate, even to situations that are counterfactual and thus can never be directly tested. You could ask on what basis is any theory generalized from data... what evidence do you have that the new data will obey the same rules as the old? At some point an inductive bias is needed. I will grant you that.

[/ QUOTE ]
I think our other discussion in this thread is probably clearer (my post above). can we let this sub-thread die and persue that one? No need to reply to this post if this is ok with you [img]/images/graemlins/smile.gif[/img]

chez

[Piers]

You really like multiple neagtives don't you.

I think I remeber a post of yours with a hextruple negative in it [img]/images/graemlins/ooo.gif[/img]

[chezlaw]

[ QUOTE ]
You really like multiple neagtives don't you.

I think I remeber a post of yours with a hextruple negative in it [img]/images/graemlins/ooo.gif[/img]

[/ QUOTE ]

This is me trying to be clear [img]/images/graemlins/smile.gif[/img]

Anyone you said 'people tend to use ....'

I responded with 'either people dont use .... or .... is not on the basis of evidence'

I don't think there's a double negative there or I think its not the case that there wasn't no double negative. [img]/images/graemlins/tongue.gif[/img]

chez