When making new software, the most important task is to discover what to make. The process is: idea, build, try. (Repeat.)

Each of these three steps could be done better. Better ideas, more accurate trials. But I’d like to see more startups working on the second step: build. This is the slow part. This is the lossy part where time is spent on something that doesn’t faithfully represent the idea.

There are already excellent tools for creating static things: Figma, Photoshop. The artifact is faithful to the final experience and is relatively quick to produce.

The problem is that software has behavior which is slow to prototype faithfully.

At one end of the continuum are paper prototypes, AR glasses made out of coat hangers and perspex, click-through static mocks. The designer must simulate the behavior in their head to choose which aspects are important and should be represented in the prototype. Quick to make, but faithless.

At the other end is code. Hours or days to try one idea. Slow to make, but faithful.

There are good behavioral prototyping tools that are in the middle of this continuum, but each has shortcomings. Axure lets you lay out UIs and wire up event handlers fast, but you have to fake algorithmic or data-driven behavior. Origami lets you build almost any behavior, and it gives you pre-built pieces of behavior that can be built very quickly. But for intricate processes it becomes fiddly and slow. GameMaker lets you build some parts of the software without code, then leans into code for the rest. This works quite well, but more could be done without code.

Good behavioral prototyping tools exist, but great ones are needed. Tools that are both fast and faithful. So we can finally stop playing simulator in our heads.

Part of Mary Rose Cook’s research to build software quickly