If code gets cheaper to write, what gets more expensive?
An open question I do not have a settled answer to. If producing software stops being the constraint, the constraint moves somewhere. I would like to know where.
I keep circling this one without landing anywhere, so it is here as a question rather than an argument.
For most of my career, the bottleneck in building software was building software. Design, planning and review all fitted around that central expense.
If that expense falls far enough, it stops being the bottleneck. Something else becomes the thing that limits how much good work gets done. I am not sure what.
The candidates
- Review. Producing a change is fast; being confident it is correct is not. Reading is now the slow step, and reading does not parallelise well.
- Deciding what to build. If ten plausible directions can each be prototyped in an afternoon, choosing between them becomes the whole job.
- Trust. Users decide quickly whether software is worth relying on. That judgement is slow to earn and fast to lose, and no amount of production speed touches it.
- Understanding. Someone still has to hold a working model of the system in their head. If nobody does, the system becomes unchangeable regardless of how quickly parts of it can be written.
Where I currently lean
Probably understanding, with review as the visible symptom. A team that cannot explain its own system will eventually stop being able to change it safely, and it will look like a review problem long before anyone names it correctly.
But I hold that loosely. If you have a better answer, I would genuinely like to hear it.