Schneier's law
expressioncryptographysecurity
'Anyone can invent a security system so clever that he can't think of how to break it.'
Bruce Schneier stated the observation in 1998 and Cory Doctorow later attached Schneier's name to it: the designer's inability to break a cipher is evidence about the designer, not the cipher. It is the working argument for public review, open competitions, and hiring people whose job is to break your own designs. Corollary in practice: never ship cryptography you invented yourself.
Anyone can invent a security system that he himself cannot break. Bruce Schneier's formulation is deliberately pointed, because the failure it names is not incompetence but a structural limit: you cannot evaluate your own design, since the assumptions that made it look sound are the same assumptions you would need to question.
The practical consequence is that confidence is not evidence. A cipher you cannot break tells you about your own attack skills and nothing about the cipher, which is why the field runs on adversarial review by people with different backgrounds, different tools and an incentive to find something. The publication-and-attack cycle exists precisely because self-assessment does not work.
It generalizes well beyond cryptography. A threat model written by the architect misses the paths the architect never considered, an incident response plan tested by its author finds no gaps, and a red team exists because internal certainty is worthless as assurance. The rule of thumb that follows is uncomfortable and correct: if a design has not been seriously attacked by someone who wanted it to fail, its security status is unknown rather than good.