Elektrine lite

← Feed

@alexanderbird@mstdn.ca

2026-09-27 02:32 UTC

Whereas the prototype-then-polish workflow optimizes for completing the first 80% as quickly as possible, I prefer an iterative, incremental process that makes the last 5% as straightforward as possible. I ask "How can we start so that we'll have the best chance of finishing our goal?" I have not found any bot-based (code generation) code process that is optimized for finishing the last 5% in a satisfactory way.

Replies (1)

  • @alexanderbird@mstdn.ca 2026-09-27 02:35

    Small-batch TDD is a classic example of a process that slows you down when you're beginning, but increases your chance of finishing. The code is always functional, and we make steady, incremental progress towards the end -- refactoring as we go to keep the codebase maleable. On the other extreme, the approach of using a code-bot to quickly crank out a Very Fancy Demo that has all the bells and whistles -- that process minimizes your chance of finishing. We've started SO MANY little things, and none of them quite work. That last 5% is full of uncertainty. The more complicated the initial prototype, the lower the chance of finishing. Is any stakeholder excited about dropping features that have already been demoed? Yet we've generated half-baked prototypes and committed ourselves to getting them working... without a cogent analysis of their feasibility or value.

    Open ##5416285