But in that scenario the subject of interest (who we’re trying to determine if they’re stupid or malicious) is the commander regardless of whether they were stupid or they were malicious. (Actually, you could apply the razor to the commander, the soldiers, or the system of both in combination; it works in any of those scenarios, as long as it applies to the same entity the whole time). In your original scenario, you aren’t comparing the hypothesis of a malicious employee to a stupid employee, you’re comparing a malicious employee to stupid instructions. Hanlon’s razor does not imply the employee is not malicious because you aren’t using it to imply the employee is stupid, you’re claiming the instructions are stupid, which is a perfectly good motivation for malicious behavior from the employee. A correct usage of Hanlon’s Razor here would be to say that you should assume the employees are stupid rather than malicious. I disagree with that interpretation because Hanlon’s Razor is often wrong, but it’s at least a valid usage of it.