Hello, I've found it useful in software design to distinguish between failure modes that will be obvious and those that may be confused with good behavior. This is important for safety-related calculation-type software, in which the numbers may be taken and used to do something important. I audited one organization that makes stress-analysis software, which classifies errors as follows:
Type A -- minor problems that dont affect results (ie warnings)
Type B -- major problems that cause results that are obviously wrong (eg crashes, or results in impossible values -- you discard the results)
Type C -- major problems that cause results that are not obviously wrong (eg a math error that adds 10 to the result, which user will use and causes a dangerous situation)
I think this was an organization-specific classification. I'm looking for a handy or industry-standard term for their "Type C" issues. I always find myself saying "errors that don't look like errors" or "non-obvious errors," which is a mouthful. It seems like this is an obvious enough concept where I'd be surprised if there wasn't some industry standard that addresses, defines, and assigns terminology.
Anybody know of any standard terminology for this?