Ron Pressler’s 2016 essay explains the difficulty of correct software through computability and complexity. The halting problem and Rice’s theorem show that no universal algorithm can decide termination or nontrivial semantic properties for arbitrary programs. Even short programs can be difficult to prove correct, so code length does not represent verification cost. The argument is not that testing or formal verification is useless, but that their scope and guarantees must be stated clearly.
Back to Tech
Computability explains why correct software remains hard
Better languages and tools do not provide a universal automatic proof of program correctness.