I wasn't going to reply to this thread, but since you keep changing the definition in the other one, let us do this.
First, I asked you in the other thread whether you believed all mathematical statements were either true or false. You replied:
https://talkfreethought.org/showthr...ct-mathematics&p=683824&viewfull=1#post683824
Speakpigeon said:
I think all statements are either true or false.
I will grant this for the sake of the argument.
Second, by 'CML-valid' or 'valid according to classical mathematical logic' or similar expressions, I mean that a deduction (or argument, inference, or whatever one calls it) is valid
according to the definition provided by you in the OP
Namely, a deduction is CML-valid if and only if it takes a form that makes it impossible for the premises to be true and the conclusion nevertheless to be false.
In particular, if A is a CML-valid mathematical argument with true premises P1,...,Pn and conclusion C, then C is true. Why? Because the premises are true, and the argument
takes a form that makes it impossible for the premises to be true and the conclusion nevertheless to be false. Therefore, C is not false. Given the previous hypothesis (i.e., every statement is either true or false), C is true.
This gives us an important feature of CML-validity: it is truth-preserving. If one starts with truths, by CML-valid arguments one gets only more truths.
Third, let us consider another proposed system of validity that is truth-preserving, say V. Since V is truth-preserving, it never is the case that an argument is V-valid but it takes a form such that it is possible for the premises to be true but the conclusion false. Hence, V-validity implies CML-validity. This gives us another important point: CML-validity is the strongest form of truth-preserving derivation of conclusions from premises.
Fourth, while mathematicians sometimes make logical errors in applying CML, that is usually corrected before a paper is published, because the authors and other mathematicians check it repeatedly. Moreover, if some errors make it into a paper, readers - who are usually also mathematicians - will almost certainly sooner or later (very probably sooner, if the paper has readers) spot it. So, while the system is not perfert, it is generally very reliable in getting CML-validity right.
Fifth, in order to prove things in mathematics, we need to start with something, right? Well, our starting points are true. If we were wrong about
that, mathematics would be pretty much hopeless regardless of what method of deduction is used. We could discuss whether those starting points are true because they're self-evident, or because we just set up a hypothetical abstract scenario and stipulate that such-and-such things hold, so they do hold in the scenario we are considering, or it depends on the case (e.g., the natural numbers for the former; a Banach space for the latter), or something else, but at least we have true starting points (or first statements, or axioms, or whatever one calls them.)
So, by CML, we find new mathematical truths, and we can find any truth that could be found from the same starting points by another truth-preserving method, but also more truths than any weaker method.
So, that's a good reason to adopt CML-validity as the way of deriving statements from others: we find true statements from true statements, and it's the strongest method for doing that.
Additionally, I would add that CML is intuitively right, for most mathematicians. This is not so because they were told so. Where I live, most mathematicians never take a course in mathematical logic (I think it would be a good thing if they did, but anyway), but those who do (or who decide to study it on their own), when they first encounter it, usually find it very intuitive, and in particular, they find the definition very intuitive.
Now, do non-mathematicians find it intuitive?
It depends on the person. But for that matter, there is a difference between trained and untrained intuitions, and the former are often better. For example, there are plenty of cases where the intuitions of people with no previous training, in nearly all cases, go wrong (purely for example, the
Monty Hall Problem, where the folk probabilistic intuitions nearly always fail). On the other hand, after learning mathematics (more specifically in the example, probability theory), humans are considerably less likely to go wrong.
Regardless,
if CML-validity is not the same as the folk conception of validity, then for the previously given reasons, either CML-validity is stronger, or the folk conception fails to be truth-preserving. Either way, CML-validity is a superior tool for finding mathematical truths. On the other hand, if CML-validity is just the folk conception of validity, then no problem, either.