Altering without change

Abstract A method is presented for showing that a revised program is similar to its original. The method allows formal expression of intuitive concepts about why the revision has not changed algorithmic features. It breaks the original problem of comparing programs down to subproblems requiring that program fragments be described and allows the concept of program similarity to be adjusted to the particular situation.