[SilasX]

Actually, in a good debate, the fallacies are taken as (Bayesian) evidence, even if obviously imperfect.

Appeal to authority/experts: The fact that experts believe something is evidence in favor of it (albeit imperfect and overridable).

Appeal to tradition: The fact that things have historically been done a certain way, in a highly immodular system with complex dependencies, without catastrophic failure is evidence in favor of it (albeit imperfect and overridable).

Appeal/reduction to absurdity: Absurd ideas are usually false, even if the presumption can be overridden.

Genetic fallacy: Actually, the reason a conclusion was reached is exactly how you should judge it. (The true fallacy is thinking that you can pick an arbitrary reason a conclusion was reached -- e.g. snake dreams and Benzene -- rather than the best reason -- the later empirical confirmation of the model's predictive power.)

And so on.

Generally, the "Fallacy fallacy" is to jump from

"Hey, this piece of evidence isn't perfect (i.e. an infinite likelihood ratio, which is impossible anyway)"

to

"So I can ignore it."

[dragonwriter]

> Actually, in a good debate, the fallacies are taken as (Bayesian) evidence, even if obviously imperfect.

I don't believe there is any reasonable definition of "good" where this is true. I think this characterization must rest on a misunderstanding of what the fallacies actually are.

> Appeal to authority/experts: The fact that experts believe something is evidence in favor of it (albeit imperfect and overridable).

Yes, but where that's true its not the fallacy of appeal to authority; the fallacy of appeal to authority is making an appeal to an authority where:

1. The cited authority is not actually an authority in the appropriate domain, or

2. The cited position is not the consensus of experts in the appropriate domain, or

3. In the context of deductive, rather than inductive, argument.

(Similar problems exist with your other arguments about fallacies as evidence).

Finally, your description of the "fallacy fallacy" is incorrect; it is not the (non-)fallacy of rejecting an argument because it is fallacious, it is the formal fallacy of affirming the negation of a claim because an argument containing a fallacy is offered in support of the claim.

[SilasX]

>In the context of deductive, rather than inductive, argument.

The very context that you should never use, in other words, which was exactly my point! If you're going to require all evidence deductively imply a conclusion (i.e. have an infinite likelihood ratio), then you're excluding all real-world evidence, and making the very fallacy I described (which I didn't realize was claimed for something else).

So it seems you essentially agree with using the "appeal to authority" fallacy in exactly the way I warned against. And that you generally endorse (the genuine fallacy of) rejecting evidence simply because it doesn't guarantee a conclusion.

Whatever it is, that's not good debate, I'm afraid.

[dragonwriter]

> The very context that you should never use, in other words

No, that's not other words for what I said.

> which was exactly my point!

Yes, that you should never use deductive reasoning was one of your points (and a ludicrous one). No, that point isn't an "in other words" of anything in my response.

> If you're going to require all evidence deductively imply a conclusion

Having some contexts within which deductive reasoning is used (what you are arguing against categorically) is not equivalent to requiring all evidence to deductively imply a conclusion. This is the fallacy of the excluded middle, arguing that if you aren't exclusively using inductive reasoning, then you must be exclusively using deductive reasoning.

> (i.e. have an infinite likelihood ratio)

Deductive implication isn't equivalent to having an "infinite likelihood ratio". First, because deductive reasoning is about necessary (logical) rather than probabilistic implication, so talking about likelihood ratios for it is misplaced, and second, because, insofar as it might make sense to talk about a "likelihood ratio" for deductive logic, the ratio would be 1, which is very much finite.

> then you're excluding all real-world evidence

Well, yeah, real world evidence is prior to a deductive argument, not part of it; insofar as it contributes to a deductive argument, it does so as part of an inductive justification for the premises of a deductive argument.

> and making the very fallacy I described (which I didn't realize was claimed for something else).

The thing you described is not a fallacy.

> So it seems you essentially agree with using the "appeal to authority" fallacy in exactly the way I warned against.

You didn't warn against using the appeal to authority fallacy, you endorsed (not "warned against) using something you called the "the appeal to authority fallacy", and then defined that as something that is not what the "appeal to authority fallacy" actually refers to.

> And that you generally endorse (the genuine fallacy of) rejecting evidence simply because it doesn't guarantee a conclusion.

No, I don't, and nothing in my post stated that, or supports your claim that I endorse that. That's just something you invented.

[SilasX]

>Having some contexts within which deductive reasoning is used ...

Okay, you object to my claim that there are no contexts in which to use deductive reasoning. Point taken, I wasn't entirely clear about that. I was speaking in the context of the discussion, which was about the set of exchanges in which fallacies are brought up. This is basically the realm of policy debate and empirical claims.

It is there that I claim you shouldn't be invoking fallacies only applicable to deductive reasoning, because there are almost never situations where the crucial issue hinges on a disagreement about a deduction.

>Deductive implication isn't equivalent to having an "infinite likelihood ratio". ...

It's not clear you are using the term correctly. I mean this value:

http://en.wikipedia.org/wiki/Bayes_factor

Saying that a piece of evidence necessarily entails a conclusion is indeed equivalent to an infinite (as much as the term is meaningful) likelihood ratio, as that is what is necessary to make your confidence in the conclusion equal to 100% and thus reduce to the case of Aristotelian deductive logic. (Probabilistic inference is such logic generalized to the cover uncertainty.)

>Well, yeah, real world evidence is prior to a deductive argument, not part of it; insofar as it contributes to a deductive argument, it does so as part of an inductive justification for the premises of a deductive argument.

As above, that "deductive argument" generally is not going to be a deduction. Your induction-based premise is only going to get you e.g. "this policy probably will save money in the long run", which means your supposedly deductive continuation is tainted with the same uncertainty.

To the extent that there is a pure syllogism in there ("we should do that that will probably save money; this will probably save money; therefore we should do that") the debate is rarely about its validity, but rather, about the degree to which those things really all probable. And in that case, the fact that experts are sometimes wrong is not a reason to reject the evidential value of their opinions. (Like I said the first time around.)

>You didn't warn against using the appeal to authority fallacy, you endorsed (not "warned against) using something you called the "the appeal to authority fallacy",

I most certainly did warn against using "appeal to authority is a fallacy" to ignore the evidential value of exper opinions; read the comment again. That is how fallacy invokers use it in practice, and why it is, well, bad reasoning.

>and then defined that as something that is not what the "appeal to authority fallacy" actually refers to.

Not if your own words are to be believed: you said the real "appeal to authority fallacy" was an OR of three situations, the last of which was exactly what I said it was: the notion that authorities are only probabilistic evidence. My point is that it's called a fallacy despite the authority only being used as probabilistic evidence in the first place!

>> And that you generally endorse (the genuine fallacy of) rejecting evidence simply because it doesn't guarantee a conclusion.

>No, I don't, and nothing in my post stated that, or supports your claim that I endorse that. That's just something you invented.

Yes you did: you claimed that the debates we're referring to crucially involve deductive rather than inductive logic. You claimed that the appeal to authority fallacy is relevant. But it's only relelvant to deductive claims thereon, which means that all of its invocations in this context are invalid arguments -- nobody is trying to claim that the authorities' endorsement guarantees the conclusion.

[dragonwriter]

> Okay, you object to my claim that there are no contexts in which to use deductive reasoning. Point taken, I wasn't entirely clear about that. I was speaking in the context of the discussion, which was about the set of exchanges in which fallacies are brought up. This is basically the realm of policy debate and empirical claims.

First, deductive reasoning has a role there, though obviously it is combined with inductive reasoning. I've directly addressed how the two interact in my prior post, so I don't feel the need to repeat it.

Second, while there are a class of fallacies that are relevant only to deductive arguments, and certain named fallacies that have different applications in deductive contexts, many named fallacies are specific to inductive arguments, or have specific application to them. Including, as it is actually defined (versus how you have mispresented it) the fallacy of appeal to authority. > As above, that "deductive argument" generally is not going to be a deduction.

Yes, it will.

> Your induction-based premise is only going to get you e.g. "this policy probably will save money in the long run",

The following is a valid syllogism where the major premise states a decision rule, the minor premise states a probabilistic characterization on an issue of fact of the type that would be an natural conclusion of inductive reasoning, and the conclusion follows of necessity from the premises and is not, itself, probabilistic: P1. If more probable than not that a person has committed a tort against another, they should be pay damages P2. It is more probable than not X commit the tort of trespass to chattels against Y, C. Therefore, X should pay damages to Y.

Obviously, there are syllogisms where one of the premises is inductively defined where the conclusions would involve a probabilistic statement, but this is not necessary.

> To the extent that there is a pure syllogism in there ("we should do that that will probably save money; this will probably save money; therefore we should do that") the debate is rarely about its validity

When the syllogism is explicitly stated and structured as a syllogism, there is rarely a debate about its validity, because people rarely explicitly lay out invalid syllogisms. OTOH, in practice, the elements that are presented are only those of an invalid syllogism and that is challenged.

> I most certainly did warn against using "appeal to authority is a fallacy" to ignore the evidential value of exper opinions

Which, is a reference to the fallacy, not a use of the fallacy, first, and second, not something I endorsed, so the fact that you warned against that does not support your claim that I endorsed the use to the fallacy that you warned against, and, finally, the thing you described actually wasn't the fallacy by that name.

OTOH, you endorsed the use of (not reference to) what you explicitly (and inaccurately) described as the fallacy of appeal to authority to support claims.

> That is how fallacy invokers use it in practice

To the extent that thing that you mistakenly describe as the fallacy of appeal to authority is something people reference as the fallacy of appeal to authority for the purpose you suggest, it is an inaccurate and, as such, inappropriate reference to the fallacy. Its also not exclusively how it is invoked in practice, though certainly for every fallacy there is some set of the actual references to it which are incorrect in this way.

> you said the real "appeal to authority fallacy" was an OR of three situations, the last of which was exactly what I said it was: the notion that authorities are only probabilistic evidence.

Actually, no, the last was that the appeal was made in a deductive context, not "the notion that authorities are only probabilistic evidence".

> My point is that it's called a fallacy despite the authority only being used as probabilistic evidence in the first place!

And my point is that (in addition to misusing language in attempting to say this) you are fabricating things when you claim that I have endorsed this.

> Yes you did: you claimed that the debates we're referring to crucially involve deductive rather than inductive logic.

No, I didn't. I said that appeals to authority in a deductive context are fallacious. I did not make the generalization you make here about "the debates we're referring to", nor do I even have any idea what specific debates you think we are referring to.