On Saturday, 7 December 2019 12:00:49 UTC-7, Khong Dong wrote:
> On Saturday, 7 December 2019 04:23:46 UTC-7, Rupert wrote:
> > On Saturday, December 7, 2019 at 10:30:18 AM UTC+1, Khong Dong wrote:
> > > On Friday, 6 December 2019 01:48:02 UTC-7, Khong Dong wrote:
> > > > On Friday, 6 December 2019 01:26:54 UTC-7, Rupert wrote:
> > > > > On Friday, December 6, 2019 at 6:20:05 AM UTC+1, Khong Dong wrote:
> > > > > > According to this Wiki link:
> > > > > >
> > > > > >
https://en.wikipedia.org/wiki/Abc_conjecture
> > > > > >
> > > > > > the familiar statement of the ABC conjecture would involve functions of reals
> > > > > > (which of course aren't first order expressible). Let's call this Wiki link
> > > > > > version the canonical version of the ABC conjecture.
> > > > > >
> > > > > > Let's consider below the, say, NN (Natural-Number) version of the conjecture
> > > > > > that does NOT refer to non-natural numbers (i.e. real numbers):
> > > > > >
> > > > > > (NN) There are finitely many ABC numeral-triples ([a],[b],[c])'s where:
> > > > > >
> > > > > >
> > > > > > - In general, [n] is a numeral denoting a natural number n.
> > > > > >
> > > > > > - [c] denotes an even number c which is a counter example of Goldbach
> > > > > > conjecture.
> > > > > >
> > > > > > - rad([a][b][c]) < [c]
> > > > >
> > > > > As I pointed out before, the Wikipedia article states that this is known to be false.
> > > >
> > > > Wrong. Because you snipped the last part of the NN version definition!
> > > > Perhaps you want to read that version _in full_ .
> > >
> > > So now you (Rupert) would understand what the NN version states, right?
> > > Any rate, are meta statements MT-A, MT-B true to you? If not, what would be
> > > the specific errors you'd see?
> >
> > In your "NN" version, you appear to attach importance to the fact that you are referring to numerals rather than natural numbers. I am not able to see how this would make any difference to the truth-value of your claims. Your "NN" version appears to me to be equivalent to the statement that Goldbach's conjecture is true.
> >
> > I see no reason to believe MT-A or MT-B.
> >
> > From your purported proof of MT-A, for example:
> >
> > "By syntactical form, an even [c] denoting a counter example of Goldbach
> > conjecture doesn't logically exclude the distinct possible existence of an ABC numeral-triple ([a],[b],[c]) where both [a],[b] aren't S0 and one of [a],[b] must necessarily not denote a prime, and rad([a][b][c]) < [c]."
> >
> > Well, how do you know that?
>
> Consider the even number c = 128 = 2^7, expressed in the below 4 equation-cases
> involving some ABC triples:
>
> 1. 3 + 5^3 = 2^7 [ rad(abc)<c ]
> 2. 3^2 + (7*17) = 2^7 [ rad(abc)>c ]
> 3. 19 + 109 = 2^7 [ rad(abc)>c ]
> 4. 31 + 97 = 2^7 [ rad(abc)>c ]
>
> First, 128 is not a counter example of Goldbach conjecture because of cases
> 3 and 4, in each of which *it's syntactically possible* to express the entire
> equation with numerals and _without_ the language multiplication symbol '*'
> (which of course must be required to define prime numbers).
>
> Otoh, suppose 128 were a counter example of Goldbach conjecture then both cases
> 3, 4 can be ignored and only cases 1 and 2 would be considered as part of a
> finitude both versions of the ABC conjecture are about.
>
> But both of cases 1. and 2. are syntactically of the same language-symbol form
> - where the language multiplication symbol '*' must by First Order _Logic_
> formalism necessarily be present on the left side of the logical symbol '='
> for the equation to be syntactically valid (i.e. well formed).
>
> Consequently, if there is a counter example of Goldbach conjecture c (and at
> least it's not known that we can logically exclude that distinct possibility)
> being relevant to the ABC conjecture (of either version), then there would have
> to be an ABC numeral-triple, hence ABC numeral-equation, like 1. and 2. (with
> or without hypothetically assuming 128 were a counter example of Goldbach
> conjecture): an (ABC-triple) equation of that restricted form, where '*' is
> absolutely absent.
>
> In summary, yes: we do you know _that_ - syntactically _that_ must be the case.
>
> The important note here is that the common form of 3 and 4 here is ignored by
> the alluded finitude in both versions of the ABC conjecture _anyway_ : since we
> would have the undesired inequality rad(abc)>c.
>
> It's an important note since it signifies we can examine the relationship
> between Goldbach, cGC, ABC conjectures as a purely syntactical symbol game or
> manipulation. And this examination would portend some disturbing facts about
> the incoherent semantics-juxtaposition of the two FOL-with-equality logical
> symbols - 'E' (Existence quantifier) and '=' (Equality), about it being invalid
> to logically, semantically equate _general_ "equivalence" with "equality".
>
> It would portend the collapse of the mathematical truth-reasoning envisioned
> (philosophized) by Plato, built by Tarski, and invalidly codified by Gödel.
>
> IUTT (Inter-Universal Teichmüller Theory) would be just an unfortunate and
> accidental smokescreen masking of this collapse. How far have SME's on both
> sides of the alleged ABC proof been able to agree even on such simple
> (undergraduate level) issue of whether or not two isomorphic objects/structures
> can categorically be treated the "same"?
>
> Not that far at all apparently!
To Rupert et al.:
I've looked and looked again many times at MT-A and MT-B, and NN_S being a
subset of Canonical_S is the final nail in a coffin of any hope that the
current alleged proof of the ABC conjecture is valid, independent of whether
or not IUUT is properly understood.
The only way I'd be wrong is if the CGC-trichotomy proof turns out to be wrong
or invalid. But one shouldn't bet on that unless one could find a blatant
counter example of the law of thought named here as HP (an impossibility in and
of itself), or unless one could prove the phrase "when Choice is necessary" is
a logically meaningless notion (another impossibility).