if you can't define a threat model, or know why you need one, you will write vulnerable code.