Deflationary Development

The new normal is that writing software gets easier - passively - with the passage of time. Any project I might start today will be done better, faster, and cheaper by next month’s models. Economics has a term for this dynamic: deflation.

The falling cost of producing code inverts heuristics for scheduling work. Big migration looming? At constant developer productivity, a day’s delay means (at least!) another day before it’s done - so start as soon as possible. That’s no longer true. The finish line gets closer while the tools sharpen themselves, especially if the delay ends with a better understanding of the problem.

The work most worth doing now is paying off debt that’s denominated in something other than development time. That’s information processed and actioned at human speed: within an engineering org, how decisions are made and reviewed. At a product level, research and adoption cycle times.

The other kind of work to do now invests in the ability to capture the gains of those future models. Better tests and documentation (better doesn’t always mean ‘more’). Reliable, concurrent builds and deploys, so batch sizes don’t increase with change rate. And experience with the tools - building a library of techniques to apply to different shapes and sizes of problems.