4 ideas
Because every language has a stable statistical fingerprint — in English 'e' is most common, followed by 't', 'a', etc. — any cipher that consistently swaps one letter for another preserves these frequencies. By counting how often each symbol appears in the ciphertext and matching it to known language frequencies, a codebreaker can reconstruct the substitution without ever knowing the key.
Security is never permanent: each new cipher is eventually broken, which forces the invention of a stronger one, which is in turn broken. Progress in cryptography is driven by this oscillation, where a method's secrecy is only a temporary lead over the analysts trying to defeat it.
Mary believed her enciphered letters plotting Elizabeth I's assassination were unreadable, so she wrote with full incriminating frankness. Walsingham's cryptanalysts broke the cipher and even forged an addition asking for the conspirators' names, leading to her execution — demonstrating that overconfidence in a weak cipher is more dangerous than using no cipher at all.
Traditionally both parties needed to share a secret key in advance, creating a chicken-and-egg distribution problem. Public-key systems use a pair of mathematically linked keys — one published openly for encryption, one kept private for decryption — so anyone can send a secure message without any prior secret exchange, relying on one-way functions that are easy to compute but practically impossible to reverse.