A purported definition of "number" has to correspond to that which we are trying to define, so any system which does not succeed in making 1+1=2 has to be judged unsuccessful.
Knotted said:
At the minute I'm trying to remember who it was who defined numbers as sets, something like 0={}, 1={{}}, 2={{},{{}}},... was it Russell and Whitehead? Anyone?
Frege and Russell both proposed a set-theoretic definition of natural numbers in which each natural number
n is defined as the set whose members each have
n elements. But I think you may have in mind Cantor's definition. Cantor argued that we can define an empty set, a collection which contains nothing (i.e. from nothing we obtain something -- the empty set). The empty set contains no members, and we can write this as ...
There's only one such emptyset, so we can also say that the set which contains all the empty sets has just one member
Now, for a two element set, we need only consider emptyset and {emptyset} together
Code:
{emptyset, {emptyset}} = 2
and so on. (
source -- see 2.2)
Knotted said:
Its striking to me that these points of view do not refer to the real physical world at all. I think there is a good reason for this. If you do define numbers with reference to the real world then you end up with allsorts of different definitions using different circumstances and different interpretations. Its just not particularly fruitful.
Yes, I agree. And it seems that by not referring to the real world, these theories force us towards the conclusion that numbers arise out of the nature of thought itself (or from
the nature of all possible worlds) rather than out of any particular contingent relationship with the world.
According to Wikipedia
here, John von Neumann's similar definition of an ordinal number does work (meaning it survives Russell's Paradox), and the finite cardinal numbers can be defined from von Neumann's ordinals by means of the Axiom of Choice. "The axiom of Infinity then assures that the set N of all natural numbers exists. It is easy to show that the above definition satisfies the Peano axioms. It also (in contrast to some alternative definitions) has the property that each natural number n is a set with exactly n elements: {0,1,2,...,n-1}"
Seems like that's the best current definition.