Hacker News Reader

Comment by dataflow | Hacker News Reader

Could you expand on why it can't work?

So if N is the natural numbers {1, 2, 3, ...} and E is the even numbers {2, 4, 6, ...}, by the definition you propose, we must have size(E) < size(N), right?

But now let's divide each element of E by 2 to produce a set D. Now that set D is {1, 2, 3, ...} aka N the natural numbers. But we just applied a function to each element of E, so how can D have a different size than E? Each element of E maps to exactly one element of D: 2 -> 1, 4 -> 2, 6 -> 3, etc. So does size(D) = size(E)? Or does size(D) = size(N)?

Thanks. How I would've imagined this could work (if it could be made to work) is to somehow account for the notion that your "infinity" just got halved (yes I get that infinity is not part of the set, but let me wave my hands here while we're breaking our axioms), and everything "after" it is no longer in the set. Which would imply that "{1, 2, 3, ...}" is no longer a sufficient description for the set of natural numbers; you'd probably need extra information (maybe a "scale factor" for the infinity or something).

I imagine you're right that this leads to a contradiction somewhere, but I (obviously) haven't thought it through. I just would love to see someone try to break enough axioms to make it work and see what comes out of it, or show that it contradicts either itself or something we see in the real world if we do that.

You can have the partial order A ≤ B iff A ⊆ B. But if you want size({1}) = size({2}), size({2}) = size({3}), etc., then you’ll find that size({1,2,3,…}) = size({2,3,4,…}) and there’s really nothing you can do about it.

But yes, the ordinals might be more to your liking. If you equip your sets with more structure you can say more things about them.

> But now let's divide each element of E by 2 to produce a set D.

You are assuming that this doesn't change the size and certainly that's how the normal notion of size works. But the question is whether we can create any order relationship on the sets with the desired properties.

The properties he mentioned defines a partial order and partial orders can be extended to total orders (given axiom of choice). So it is in fact possible.

A partial order of the elements of some set can be extended to a total order.

But a partial order of all sets? How is this function "size" even defined? Functions don't have a domain of all sets in conventional set theory.

Reply on news.ycombinator.com