Claim: there is no largest whole number. Suppose there were one, and call it N. Then N+1 is a whole number bigger than N, so N wasn't the largest after all. The assumption defeats itself, so it is false: there is no largest whole number. That is the whole method: assume the opposite, follow valid steps, reach something impossible, and blame the assumption.
The power comes from what the assumption gives you. "There are infinitely many primes" is hard to attack head-on, but "suppose there are only finitely many, p1,p2,…,pk" hands you a list. Multiply them all and add 1: M=p1p2⋯pk+1. Dividing M by any prime on the list leaves remainder 1, so none of them divides M. Yet every whole number above 1 has a prime factor, so M has one that is missing from the list, and the list was not complete. M itself need not be prime: 2⋅3⋅5⋅7⋅11⋅13+1=30031=59⋅509.
The ending must be truly impossible: 1=0, an integer that is both even and odd, a fraction in lowest terms whose top and bottom are both even. A statement that is only surprising, such as "then n is bigger than a million", is not a contradiction. The method also differs from the contrapositive. To prove "if P then Q" by contrapositive, you assume only "not Q" and prove "not P" directly; by contradiction, you assume P and "not Q" together and look for any impossibility.
Think of it like this. It works like checking an alibi. Suppose the suspect was at the scene; then they were in two cities at 9 p.m., which is impossible, so they weren't there. The reasoning holds only if every step is solid: a wrong clock, like a faulty algebra step, can produce a false impossibility.
Core idea. Proof by contradiction assumes the negation of your claim, follows valid logic until it produces something impossibly false, and concludes the original claim must be true.