What Is a Mathematical Proof and Why Does It Matter?

How do you know something is true in math, forever, with zero exceptions?

Not "probably true." Not "true in every case anyone has ever checked." True, full stop, for every case that will ever exist, including ones nobody has thought of yet.

That's what a proof gives you. And it's a stranger, stricter thing than most people assume. It isn't a pile of convincing examples. It isn't an expert saying "trust me." A proof is a chain, built one link at a time, where each link is forced by the ones before it. Once you accept the first link, you can't coherently deny the last one.

Let's see why that bar is so much higher than it sounds.

Isn't checking a bunch of examples enough?

Try this claim: the expression $n^2 + n + 41$ always produces a prime number, for every whole number $n$.

Plug in $n=0$: you get 41, prime. $n=1$: 43, prime. $n=2$: 47, prime. Keep going, and it keeps working. Not for a few cases, for dozens.

Drag that slider up. The claim survives 40 straight checks in a row. If you stopped at $n=39$, you'd swear you'd found a law of nature. Then $n=40$ arrives, and $f(40) = 1681 = 41^2$, which is about as composite as a number gets.

Forty successes, one failure, and the claim was false the whole time. This is the trap: no finite pile of confirming examples ever proves a claim about infinitely many cases. There's always one more $n$ you haven't tried. A pattern holding a thousand times is still just a pattern. It isn't a proof.

So what does a real proof look like?

A proof doesn't check cases one at a time. It finds a reason the claim has to be true, a reason that covers every case simultaneously, without ever running out.

Here's a classic one. Add up the first few odd numbers: $1$, then $1+3=4$, then $1+3+5=9$, then $1+3+5+7=16$. Notice anything? Each sum is a perfect square: $1^2, 2^2, 3^2, 4^2$.

We could check this for $n=100$ and it would still hold. But why, for every $n$ at once? Watch what happens when you build the sum as a picture instead of a list of numbers.

Each new odd number isn't just some number that happens to fit, it's a border, an "L" shape wrapped around the previous square, adding exactly one row and one column. The $k$-th border adds $2k-1$ new tiles, because that's exactly how many tiles it takes to wrap a square one layer bigger. Stack $n$ of those borders and you've built an $n \times n$ square, by construction, not by coincidence. That's a proof: not "we checked and it worked," but "here's the structural reason it can't fail."

We write this more formally as:

Same fact. The picture just tells you why it's forced.

What does "forced" actually mean, step by step?

Real proofs are usually built from words and logic rather than tiles, but the idea is identical: each statement has to follow from the one before it, with no room for a maybe.

Here's a famous example: proving that $\sqrt{2}$ is irrational, meaning it can never be written as a fraction of whole numbers. The strategy is sneaky. Assume the opposite of what we want, and show it collapses.

Follow the chain all the way through. Assume $\sqrt2 = p/q$ in lowest terms. Square it, and $p^2$ turns out even, which forces $p$ itself to be even. Substitute that back in, and now $q^2$ is even too, which forces $q$ to be even as well. But if $p$ and $q$ are both even, they share a factor of 2, which breaks "lowest terms," the thing we assumed on step one.

That's the trick of proof by contradiction: assume a claim is false, follow the logic wherever it leads, and if it leads somewhere impossible, the assumption must have been wrong. Every step here was forced. Not persuasive, forced. That's the difference between a proof and an argument that merely sounds right.

Can proof reach conclusions no amount of checking ever could?

This is where proof earns its keep. Take one of the oldest questions in math: are there infinitely many prime numbers, or does the list eventually stop?

You could check: 2, 3, 5, 7, 11, 13... they keep appearing, for as far as anyone has ever computed. But that's the exact trap from the very first section. Checking further and further never proves "forever." We need the same trick as the last section: assume the list is finite, and see what breaks.

Say you had the complete list of primes, all $k$ of them. Multiply every one together and add 1. Call that $N$. Now here's the move: $N$ can't be divisible by any prime on your list, because dividing by any of them always leaves a remainder of exactly 1. So $N$'s smallest prime factor, whatever it is, has to be a prime that was missing from your "complete" list all along.

Notice something at $k=6$: $N$ itself isn't even prime this time (it's $30{,}031 = 59 \times 509$). That's fine, the proof never needed $N$ to be prime, only for its factors to escape the list. No matter how large you make your "complete" list, this construction always finds a prime outside it. So no finite list is ever complete. There are infinitely many primes, not because we checked far enough, but because the list breaking is forced, for every possible list size at once.

Why does this actually matter outside a math class?

Because "probably true" and "true" behave very differently once something depends on them. A bridge engineer doesn't want "this design held up in every test we ran so far", they want the load calculation that guarantees it, the same way $f(n)=n^2+n+41$ fooled everyone for 40 cases before quietly failing.

Software has the same trap. A function can pass a thousand test cases and still contain a bug nobody's input has triggered yet, that's testing, evidence collected case by case. Formal verification, the discipline of actually proving a program correct, is the difference between "we tried a lot of inputs" and "this cannot fail, for a reason we can point to." The same gap separates a cryptographic system that's merely "never been broken yet" from one with a security proof behind it.

Proof is what lets a claim outlive the person who checked it, and outrun every case anyone thought to try.

The short version

Checking examples, however many, only ever gives you evidence, and evidence can hold for forty straight cases and still be wrong on the forty-first. A proof is different: a chain of statements where each link is forced by the last, landing on a structural reason the claim has to be true for every case at once, not just the ones you tried. That's why the sum-of-odd-numbers picture, the $\sqrt2$ contradiction, and Euclid's endless supply of primes all count as proofs, and forty prime numbers in a row does not.

A proof doesn't just make you confident. It removes the need for confidence at all.