A note I sent to a Parasoft sales rep in March 1998:

    I know that it's been almost two months since you sent this to me, so you've probably forgotten my query regarding Parasoft's Insure tool. But I did read the description of Insure++ that you sent to me.

    First, I appreciate very much that you sent it.

    But, I have to be honest, and tell you that as a scientist, an educator, and as a concerned member of the software testing industrial community, I am appalled at the misuse of the mutation terms.

    Here is my summary of the description of the process that Insure++ follows:

    1. Use the original program P to create "equivalent" mutants, Mi, 1 <= i <= n
    2. Run mutants on test cases tj, 1 <= j <= m
    3. If Output(Mi, tj) != Output(P, tj) then a fault is found in P

    Here is my analysis of the statements and claims that are made:

My final assessment is one that I am sure your company will not like, but I will say it anyway. My assessment is that Insure is questionable technology that is based on paper foundations.

Inventing new technology is okay. Inventing new ideas is great. Selling ideas that have not been validated is very risky. Using terms that have been established and well defined to mean something completely different is misleading. The paper says "Insure++ takes a different approach to mutation testing" ... in fact, Insure++ has almost nothing to do with mutation. Saying it is does a disservice to the years of hard scientific and engineering work that have gone into inventing, validating, and refining mutation. I think of someone who may someday use Insure, think "wow, mutation doesn't work!", and I feel very disappointed ...

Jeff Offutt