FOM: Having truth-values

[William Tait]

Some of the discussion of CH and second-order logic has concerned the 
question of whether this or that mathematical proposition A, e.g. CH, 
has a truth-value, i.e., whether

(1)  A is true or A is false [i.e. not-(A is true)]

is true. Here A is not a sentence of a formal system being 
interpreted in a model: it is an ordinary meaningful mathematical 
statement.

What meaning should the sentence  ``A is true'' have, other than the 
meaning expressed by A? The illusion that Steel mentioned (Oct 16) is 
precisely the illusion that there is a further criterion of truth, 
e.g. given by Tarski's definition of truth, where ordinary 
mathematical language, of which A is a sentence, is thought of as 
though it were a formal language (the object language) being 
interpreted in some MODEL-IN-THE-SKY within some superlanguage (the 
metalanguage). If one rejects this picture, as really we should, then 
``A is true'' can have no sense other than that of ``A'' itself, in 
which case (1) becomes

(2)   A or not-A

which most of us accept as valid in mathematics.

So CH has a truth-value; but we are not in the position to determine 
which one. We will be in the position to do so when and if new axioms 
are accepted which imply CH or its negation.

On these grounds, I think that the assertion that CH, or its 
truth-value is `indeterminate' can be nothing more than a 
sociological prediction that such axioms will never be found. I don't 
know how risky that prediction is.

Bill Tait