Speakpigeon said:
I think all statements are either true or false.
Alright, so let us stipulate from now on, and for the sake of the argument, that that is correct (it makes my case easier, though I could make my case as well it if you had said otherwise). Now consider the definition of validity in classical mathematical logic (CML for short). A proof is valid in CML if and only if it takes a form that makes it impossible for the premises to be true but the conclusion to be false.
It follows that CML-valid proofs with true premises have true conclusions, always. So, CML leads from truths to truths.
Now, Aristotelian-valid proofs with true premises also have true conclusions. But there are proofs that are CML-valid but not Aristotelian-valid. In fact, for any proposed system of logic PL in which all valid proofs with true premises have true conclusions (whether the Aristotelian-system, or intuitionism, or any other system), any proof that is PL-valid is also CML-valid. Why? Because, by definition, a proof is valid in CML if and only if it takes a form that makes it impossible for the premises to be true but the conclusion to be false.
The conclusion?
CML-validity is the strongest form of truth-preserving validity. Any mathematical truth that can be discovered by any other truth-preserving proposed system of logic (whether intuitionistic, Aristotelian, or any other), can also be discovered (and with no greater difficulty) from the same premises with CML-logic. On the other hand, for any proposed truth-preserving system of logic PL, there are arguments with true premises that are CML-valid but not PL-valid. As a result, there are mathematical truths that can be discovered by means of CML, but not PL. Moreover, in the case of systems like Aristotelian logic, some proofs might be extremely difficult to obtain even if valid, whereas CML provides more tools for finding truths.
Note that all of this is
independent of whether CML matches human logic, or not.
Now, it is also true that if the premises are false, CML will likely lead to more errors than weaker systems, like the Aristotelian system or Intuitionistic Logic. However, in mathematics, that is not likely problematic, for two reasons:
1. The basic, self-evident statements taken as axioms in different fields are clearly true (and if some of them weren't, then human mathematical intuition would seem hopelessly lost, so that a weaker system of logic would be of no help).
2. Sometimes, proofs begin with previous results, which already are in no way self-evident. But those more complicated results have been proven from earlier results, etc., until we get to something really basic. And the proofs have been checked by other mathematicians. Errors are usually soon discovered, and when they rarely persist for a while, they're still discovered.
So, in short, CML is the best way of finding mathematical truths. Now, there are mathematical truths that we will never find. But we will find a lot more with CML than with any other proposed system PL. Granting now for the sake of the argument that CML is the wrong logic in the sense that it does not match human logic, and further granting that your arguments on the matter are persuasive, then I would say we should still do mathematics using the best truth-finding method we have - namely, CML
Again, in this case, the wrong logic is
superior to the right logic, as a tool for finding mathematical truths.
Speakpigeon said:
No. Some implications (in effect an infinity of them) are valid according to Aristotelian logic and not valid according to mathematical logic.
That is false (of course, we are talking about deductions). There are no implications that are Aristotelian-valid but not CML-valid. Remember, if a proof is CML-invalid, then by definition, it takes a form such that it is possible for the premises to be true but the conclusion false. That would imply that Aristotelian logic
fails to be truth-preserving, which would be a devastating blow for it, regardless of whether it matches human logic.
Fortunately for Aristotelian logic, every proof that is Aristotelian-valid is also CML-valid.
Speakpigeon said:
And you also have the opposite situation.
Indeed, you do have that. There are infinitely many proofs that are CML-valid, but not Aristotelian-valid. But it's not merely that there are infinitely many proofs like that. More importantly, examples among those infinitely many are all over the place. As I mentioned, a very simple example was provided by Bomb#20
here.
Speakpigeon said:
So, no, not all things that follow in Aristotelian logic also follow in mathematical logic. As I said from the start, mathematical logic is seriously wrong.
Granting for the sake of the argument that mathematical logic is seriously wrong as you believe (which I do not believe at all, but that's not the point here), the fact remains that all proofs that are valid in Aristotelian logic are also valid in CML. Moreover, if there were Arisotelian-valid proofs that are not CML-valid, then Aristotelian logic would not be truth-preserving.
Speakpigeon said:
Again, that's not true. Mathematical logic produces two types of error.
It produces neither. But I'm willing to assume in this thread and for the sake of the argument that it is the wrong logic, in the sense that it fails to match human logic. Under that assumption, it produces the first but not the second type of error. But it's an error easily fixed
if we assume there is a persuasive argument to the conclusion that CML fails to match human logic.
Speakpigeon said:
First, it declares valid inferences that are not valid.
Yes, by assumption in this thread (not in reality). Adding the further assumption that there is a good argument for the conclusion that some inferences are CML-valid but not human-logic-valid, the solution is to refrain from claiming that all CML-valid inferences are also human-logic-valid, or valid in the colloquial sense of the words.
However, CML remains superior to human logic as a logic for mathematics, as it can be used to obtain all of the mathematical truths that can be obtained with human logic, and more truths as well (assuming human logic is Aristotelian logic or some other truth-preserving system; on the other hand, if human logic is not truth-preserving, the superiority of CML is even greater).
Speakpigeon said:
Second, it is myopic in that it cannot prove valid some inferences that are valid. And again, when I say some, there is in fact an infinity of these inferences.
And again, that is false. Fortunately so, because any inference that is not CML-valid is of a form that is not truth-preserving.
If there were a single inference that is valid but not CML-valid, then validity (unlike CML-validity) would fail to be truth-preserving.