Sizes of infinity
A Socratic walk-through of sizes of infinity — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #How can one infinite set contain more members than another?
The phrase sounds like a contradiction. Infinity is the end of counting; how could one endlessness be roomier than another? But notice what the objection assumes — that "how many" must mean a number reached by counting. If the counting never finishes, that definition has nothing to say. We need a comparison that never asks us to finish.
Reasoning it through
REASONING #Is there one? Imagine a hall of people and a hall of chairs, and you are forbidden to count. Ask everyone to sit. If nobody stands and no chair is empty, the collections are the same size — and you used no number at all.
Now let that loose on the infinite. Take 1, 2, 3, … and the evens 2, 4, 6, … Surely the evens are half as many, since we discarded every other one. But pair n with 2n: nobody stands, no chair is empty. Does that trouble you? It should trouble the intuition that a part must be smaller than the whole — a property of finite collections, not a law of quantity. The fractions fall the same way: lay them on a grid, numerator across and denominator down, walk it in diagonal zig-zags skipping repeats, and every fraction is reached at some definite step.
This begins to look like a proof that all infinities are one size. Then Cantor asks the sharp question: can the real numbers be listed this way? Suppose someone hands you a list claiming every real between 0 and 1, one per line. Build a number whose first digit differs from the first digit of line one, whose second differs from the second digit of line two, and so on down the diagonal. It cannot be line one, or line seventeen, or any line — it differs from each somewhere. And nothing in the argument used their particular list. This is not a claim that the reals are hard to list; it is a proof that no listing exists.
The analogy
THE ANALOGY #Return to the hall. Countable infinity is a hall where every guest, however many, gets a numbered seat with none left standing. Cantor's argument describes a crowd for which any seating plan you propose leaves at least one guest unseated — and you can name that guest by reading the plan itself.
In a real hall you could fetch another chair. Here you cannot, because the leftover guest is manufactured from whatever plan you offer; add a chair, rerun the diagonal, and a fresh unseated guest appears.
Clarifying the model
THE MODEL #The diagonal has one loose thread worth naming: some reals have two decimal expansions, since 0.4999… equals 0.5000… A careless diagonal could build a number that merely looks new; the standard fix is to choose replacement digits avoiding 0 and 9, so the constructed number has one expansion and is honestly absent. And "more" means something precise: the reals cannot be paired with the naturals, though the naturals pair into the reals.
A picture of it
THE PICTURE #How to readEach line is an attempted seating plan between the naturals and another infinite set. The top two carry the double bar at both ends — an exact one-to-one pairing, which is why the evens and the fractions are the same size as the naturals despite looking smaller and larger. The third breaks that symmetry: the naturals end reads "zero or one", so some reals get no partner. The attribute box names that unpartnered member, which is the diagonal argument in one line.
What became clearer
WHAT CLEARED #"Same size" for infinite sets means a perfect pairing exists, and by that standard the evens and the fractions match the whole numbers exactly. The reals do not — not for want of cleverness, but because any proposed list can be turned against itself to construct a number it omits.
Where to go next
ONWARD #- Why the set of all subsets of a set is always strictly larger, building an infinite ladder.
- Whether any size sits between the countable and the continuum — the continuum hypothesis.
- Why almost every real number is one we can never name or compute.
Key terms
TERMS #| Term | What it means |
|---|---|
| Cardinality | the size of a set, defined by which sets it can be put in one-to-one correspondence with. |
| Bijection | a pairing leaving nothing unmatched and nothing doubled up, in either direction. |
| Cantor's diagonal argument | the proof that no list of reals is complete, by constructing a number differing from the nth entry in its nth digit. |
Every term the collection defines is gathered in the glossary.