Security is a property of human outcomes, not technical systems. The security community understands how to think about the security of code and is learning how to think about the security of large systems, but has barely begun to start to think about how to improve security outcomes for humans. Security for humans affects the entire software development and deployment lifecycle, but it’s most strongly present in the design and requirements phase where it has been largely ignored by the security community. The need for outcome-oriented security design is particularly obvious in the design of systems intended for use by high-risk, specifically-targeted users. In this talk, we’ll describe the security design problem, the challenges of supporting high-risk users, and some steps toward practical processes and solutions for the missing discipline in the security community.