designlisted

# Design Explore several options before committing to one. Based on "Design It Twice" (Ousterhout). Your first idea is rarely the best, and sketching alternatives costs far less than building the wrong one. Works for any design problem: an interface, a data model, an algorithm, a control flow, a system architecture, a process. A design is a tree of dependent decisions. Walk it one decision at a time, resolving each branch and respecting the dependencies between decisions. Give your recommended answer for each. ## Process ### 1. Explore Understand the current state before proposing anything. - Explore what already exists and how it works - Identify the constraints any solution must satisfy - Identify the dependencies a solution would rely on - Understand relevant history and existing patterns - Check for available skills relevant to what's being designed, and load them for domain-specific vocabulary, angles, and constraints Optionally ground the constraints with a rough sketch. This is not a proposal, just a way to make them concrete. Present your understanding to the user, then proceed. ### 2. Produce options Produce 2 radically different options. Push them apart deliberately: if two differ only in detail, replace one. Vary the axis that matters for this problem, such as what's optimized, what's hidden, where complexity sits, or which case is made trivial. The output for each option is: 1. The approach — what it is and how it works 2. A concrete sketch or example